Developing Soft Bio-Inspired Cooperation Methods and Mechanisms for Flapping Wings Aerial Robots (FWAR)

Proyecto Fin de Carrera Ingeniería de Telecomunicación Formato de Publicación de la Escuela Técnica Superior de Ingeniería Autor: F. Javier Payán Somet Tutor: Juan José Murillo Fuentes Dep. Teoría de la Señal y Comunicaciones Escuela Técnica Superior de Ingeniería Universidad de Sevilla Sevilla, 2013 Tesis Doctoral Ingeniería Automática, Electrónica y de Telecomunicación Developing Soft Bio-Inspired Cooperation Methods and Mechanisms for Flapping Wings Aerial Robots (FWAR) Autor: Vicente Pérez Sánchez Directores: Begoña C. Arrue Ullés Anibal Ollero Baturone Ingeniería de Sistemas y Automática Escuela Técnica Superior de Ingeniería Universidad de Sevilla Sevilla, 2023 Tesis Doctoral Ingeniería Automática, Electrónica y de Telecomunicación Developing Soft Bio-Inspired Cooperation Methods and Mechanisms for Flapping Wings Aerial Robots (FWAR) Autor: Vicente Pérez Sánchez Directores: Begoña C. Arrue Ullés Profesor Titular Anibal Ollero Baturone Catedrático Ingeniería de Sistemas y Automática Escuela Técnica Superior de Ingeniería Universidad de Sevilla 2023 Tesis Doctoral: Developing Soft Bio-Inspired Cooperation Methods and Mechanisms for Flapping Wings Aerial Robots (FWAR) Autor: Vicente Pérez Sánchez Directores: Begoña C. Arrue Ullés Anibal Ollero Baturone El tribunal nombrado para juzgar la Tesis arriba indicada, compuesto por los siguientes doctores: Presidente: Vocales: Secretario: acuerdan otorgarle la calificación de: El Secretario del Tribunal Fecha: A mi familia A la familia que te da la vida A mis compañeros A mis profesores A mis amigos A Ana Acknowledgements Developing a Ph.D. thesis is challenging and implies years of work and struggle. First, thank the Grupo de Robótica Visión y Control (GRVC) for funding my research. I can not imagine this process without my advisors, Anibal Ollero and Begoña Arrue, providing their feedback and helping me to improve my research results. I want to thank my lab colleagues. They made it possible to afford this period with good moments and contributed to my research with their opinions. I want to give a special mention to my old collaborators: Ale,Malgani, Ricardo, Manu, Emanuela, Batista, Marco, and Pablo. They put up with my bad moments and encouraged me to continue, contributing considerably to my work. The period of the thesis is lengthy, and also new colleagues support the last part of this dissertation: thanks to Alejandro, Alberto, Carlos, Jorge, Diana, Hong Fan, Pol, Kike and Juanma. I want to thank also to my colleagues at Laboratory of Intelligent Systems (LIS) at the École Polytechnique Fédérale de Lausanne (EPFL). The time I spent was amazing, learning from different cultures and researchers. I want to thank the researchers who provided a solid baseline to obtain the result presented in this research. I would also like to thank all the teachers who contributed to increasing my knowledge. The doctoral thesis is just a period of life. However, your education gives you the resilience to face it. I want to thank my family for providing me with the education that makes it possible and supporting me over the years. Thanks to my father for teaching me the meaning of work. Thanks to my mother for stressing the importance of the study. Thanks to my sister for being an example to me. I want to make a special mention to my grandparents, who passed on to me the meaning of being a good person. I would also like to thank the family that gives you life. Thanks to my friends. I want to apologize for the time my research stole from me with them. Finally, I want to thank Ana for supporting me during this long period. It has not been easy at times, but we have made it. We never know what the future has in store for us, but this doctoral thesis is for you. Vicente Pérez Sánchez Sevilla, 2023 III XAcronyms LiPo Lithium Polymer. 87 LIS Laboratory of Intelligent Systems. 12, III LMPA Low Melting Point Alloy. 7 MAV Micro Aerial Vehicles. 131, 132 MFC Macro Fiber Composite™. 8, 33–35, 92–104, 106, 108, 142, 159, 167, VII MoCap Motion Capture. 139, 145 MOSFET Metal–Oxide–Semiconductor Field-Effect Transistor. 86–88, 105, 134 NASA National Aeronautics and Space Administration. 8, 33 PD Proportional Derivative. 134 PDMS Polydimethylsiloxane. 32 PLA Polylactic Acid. 59, 67–71, 159 RC Radio Control. 104, 105, 143, 145 RPAS Radio Piloted Aircraft Systems. 1 SMA Shape Memory Alloys. 7, 8, 29, 32, 33, 40–43, 46, 47, 50, 57, 73–75, 78–84, 86–88, 119, 122, 127, 129, 135, 158, 163, VII SMP Shape Memory Polymers. 7 TPU Thermoplastic Polyurethane. 59, 62, 67–71, 73, 76, 127, 159 UAV Unmanned Aerial Vehicles. 1–5, 8, 11, 27, VII WWI World War I. 1 WWII World War II. 1 Short Contents Resumen V Abstract VII Acronyms IX Short Contents XI 1 Introduction 1 1.1 Motivation and Objectives 1 1.2 Thesis Framework 12 1.3 Thesis Outline 13 1.4 Contributions 14 2 Bio-Inspired Soft Flapping Wing Aerial Robots(FWAR) 19 2.1 Bioinspiration 19 2.2 Ornithopthers Flapping Wings Aerial Robots(FWAR) 26 2.3 Soft Robotics Technologies 31 3 Bio-Inspired Manipulators for Ornithopters 37 3.1 Shape Memory Alloy(SMA) Artificial Muscles 41 3.2 Analysing the Forces Involved in the Perching 51 3.3 Claws Morphology 55 3.4 Leg Morphology 72 3.5 Actuation Method and Integration 82 4 Bio-Inspired Morphing Surfaces 91 4.1 Morphing Tail 92 4.2 Morphing Wing 108 5 Experimental Validation 121 5.1 Bio-Inspired Claws Experiments 121 5.2 Soft Aerodynamic Surfaces Experiments 139 6 Conclusions and Future Works 157 6.1 Conclusions 157 XI XII Short Contents 6.2 Future Works 162 List of Figures 163 List of Tables 167 Bibliography 169 Contents Resumen V Abstract VII Acronyms IX Short Contents XI 1 Introduction 1 1.1 Motivation and Objectives 1 1.2 Thesis Framework 12 1.3 Thesis Outline 13 1.4 Contributions 14 2 Bio-Inspired Soft Flapping Wing Aerial Robots(FWAR) 19 2.1 Bioinspiration 19 2.1.1 Birds Manipulation Capabilities 21 2.1.2 Birds Perching Capabilities 23 2.1.3 Birds Human Interaction Capabilities 25 2.2 Ornithopthers Flapping Wings Aerial Robots(FWAR) 26 2.3 Soft Robotics Technologies 31 3 Bio-Inspired Manipulators for Ornithopters 37 3.1 Shape Memory Alloy(SMA) Artificial Muscles 41 3.1.1 SMAs Properties 41 3.1.2 Modeling SMA Muscles 44 Muscles dynamic model 44 Muscles electric and energy model 46 3.1.3 Developing SMA Muscles 50 3.2 Analysing the Forces Involved in the Perching 51 3.3 Claws Morphology 55 3.3.1 Claws Conceptualization 55 3.3.2 Claws Design 58 3.3.3 Claws Design Evolution 67 3.4 Leg Morphology 72 3.5 Actuation Method and Integration 82 3.5.1 SMAs Actuation Configuration 82 XIII XIV Contents 3.5.2 Tendon Driven Actuation 83 3.5.3 Actuation Control 86 4 Bio-Inspired Morphing Surfaces 91 4.1 Morphing Tail 92 4.1.1 MFC 93 Modeling MFC 95 4.1.2 Tail development 101 Design 102 Manufacturing 103 Control Electronics 104 Feastibilty Test 105 Tail Comparisons 107 4.2 Morphing Wing 108 4.2.1 Design 110 4.2.2 Manufacturing Process 115 5 Experimental Validation 121 5.1 Bio-Inspired Claws Experiments 121 5.1.1 Grasping Experimentation 122 5.1.2 Perching Experimentation 132 5.2 Soft Aerodynamic Surfaces Experiments 139 5.2.1 Bio-Inspired Tail Experimental Validation 139 Kinematic analysis 139 Outdoors experiments 142 5.2.2 Bio-Inspired Wings Experimental Validation 145 Wing test bench experiments 146 Free flight experiments 150 6 Conclusions and Future Works 157 6.1 Conclusions 157 6.2 Future Works 162 List of Figures 163 List of Tables 167 Bibliography 169 1 Introduction Se una persona è persistente, anche se difficile da capire, sarà intelligente; e anche se è debole, diventerà più forte. Leonardo Da Vinci, (1452-1519) 1.1 Motivation and Objectives The presence of drones in our lives has increased exponentially over the last few years. At first, drones were conceived as military weapons. Today, drones are present in our daily lives. This presence generates challenging objectives in the interaction with this aerial platform. However, they are crucial in transport, surveillance, maintenance, and industrial inspection tasks. Moreover, drones provide a new vision of the world, avoiding risky situations and reducing transport time. Military applications were the starting point for drones. The first record of drone use was in 1849 [ 1 ] when the Austrian forces launched 200 incendiary balloons at Venice. Due to the lack of control of these platforms and the wind gusts of that day, most did not reach their objectives. Nevertheless, after this first attempt, most of the forces in the world realized the potential use of these platforms, and systems evolved quickly. The first milestone of drones’ evolution was the development of Radio Piloted Aircraft Systems (RPAS). Leonardo Torres Quevedo, in 1903[ 2 , 3 ], used a radio system to control the trajectory of an airship for the first time (see Figure 1.1a. This concept sets the base of drones, and it is currently maintained. However, before this development, the absence of control over their trajectory made impossible their use effective. During 1914-1915, the effort made during the World War I (WWI) concludes developing the base of the first Unmanned Aerial Vehicles (UAV). Figure 1.1b represents the first concept of drones developed during the WWI. The platforms were improved and gained prominence during WWI and World War II (WWII). As a result, they were crucial in military usages like surveillance tasks and localization. All the knowledge learned in the military use of UAVs was directly reflected in civil applications of these systems. In 2006, the Federation of Aviation Administration (FAA) registered the first commercial drone permit. The motivations to begin the civil applications of drones were their potential applications. Initially, these motivations were their use in surveillance, inspection, and helping in natural disasters. 1 2Chapter 1. Introduction (a) Telikino first radio control system Leonardo Torres Quevedo 1901[4]. (b) Hewitt-Sperry Automatic Airplane in 1918 [5]. Figure 1.1 Early drones. At that moment, the classification of UAVs contemplates three types of drones: helicopters, multi-rotors, and fixed-wings. • Helicopters:(see Figure1.2a). They have a high payload and can hover. Also, their flight is stable and efficient. • Multi-rotors:(see Figure1.2b). As helicopters, they can hover. However, these platforms have higher maneuverability than helicopters. In addition, these aerial systems can lose one of their propellers continuing with the flight. That makes it possible to operate with these platforms in smaller areas. Nevertheless, they are less energy efficient than fixed wings. • Fixed wings:(see Figure1.2c) These platforms can glide. This capability makes it possible to cover higher areas wasting less energy. However, their high flight speed makes using this platform in confined spaces difficult. (a) Hellicopter [6]. (b) Multicopter [7]. (c) Fixed Wing [8]. Figure 1.2 Drones clasification. Since 2006, UAVs civilian applications have experienced exponential growth. The first one was the use UAVs for video recording. This application is one of the most straightforward. It involves being capable of collecting images in the aerial platform while controlling the trajectory of the platform. The first method to know the course of the drones was the Global Positioning System (GPS). At that moment, civilian drones only use GPS to control their position. But, later, the miniaturization of the systems makes it possible to use multiple sensors to improve the recognition of the environment. However, these data have to be treated to be useful. Data fusion is one of these processes. 1.1 Motivation and Objectives 3 Data Fusion combines the information given by different sensors to understand the environment around the aerial platform. The camera is one of the primary sensors to recognize the environment around the UAV. The process merges this data with the data of the flight sensors (altimeters, gyroscopes, Inertial Measurement Unit (IMU), and GPS). It makes it possible to improve recognition of the environment around the platform. Surveys like [ 9 ] highlight the importance of data fusion processes. Notably, the authors focus on using cameras and the influence of computer vision in Flight Control, Navigation, and Guidance. The following lines summarize examples of the leading vision techniques used in data fusion processes. Regarding navigation, the visual serving method [ 10 ] gives the position and orientation of the aerial platform as a function of the target. [ 11 , 12 , 13 ] shows examples of visual servoing thecniques. SLAM [ 14 ] is one of the main methods in navigation using visual odometry and mapping [ 15 , 16 , 17 ]. Obstacle avoidance [ 18 ] combines flight control with navigation and object recognition [ 19 , 20 ]. Target tracking [ 21 ] is another navigation technique by the recognition of the objective[ 22 , 23 ]. These techniques were developed before their use on drones. However, their use of drones makes these methods gain prominence. These techniques are crucial for the use of drones in unknown situations. They make it possible to react to uncertainties in the environment. Figure 1.3 Data fusion process. Data fusion enabled merging the information to recognize the environment around the aerial platform and improve the control skills of the UAVs. Therefore, it makes achievable the interaction with the environment. Firstly, this topic tries only to analyze the contact. The aim was to investigate the contact and the perturbation generated by the environment in the entire platform. Then, however, a new era in aerial robotics was born:Aerial Manipulation (AM). At the beginning of this era, the manipulation was tested using helicopters that carry some customized end effectors [ 24 , 25 ](Figure 1.4a). The capabilities of helicopters to 4Chapter 1. Introduction carry out higher loads and maneuverability made them the first candidate to develop these tasks. Then, researchers used traditional industrial arms to overcome the limited capabilities of the first customized arms in [ 26 , 27 ]. However, the heavy weight of these arms was limiting the control capabilities of the platforms. Moreover, the helicopters’ flight performance was limiting these works. Their big rotors made it challenging to work in confined areas. Also, the significant influence of the environment in the flight of these was causing failures. Then, the objective was to develop manipulation tasks on smaller multi-rotors. This platform has redundant propellers that make the flight safer. Also, these propellers are smaller, reducing the influence of the environment on the flight. However, their payload is smaller. Then, researchers focus on cooperation to carry out higher loads. The aim was to distribute the weight among the platforms involved. The work presented in [ 28 ] shows one of the first examples of helicopter cooperation to carry out higher loads outdoors under windy conditions. The authors in [ 29 ] show an example of these applications using multirotors indoors. (a) Yale aerial manipulator from [25]. (b) Helicopter using a 7 Degrees of Freedom arm from [27]. (c) Cooperative multi UAV transportation from [29]. Figure 1.4 Early manipulators. The research world then extended collaborative solutions to carry out a higher load. However, manipulation is not only to carry out loads. It is also to grasp the objects and perform manipulation with them. In that way, the necessity of developing adapted arms and customized end-effectors for aerial robotics continued. Size and weight were the main problems of the early manipulators. Their size caused disturbances throughout the flight, and the weight made using them on lighter platforms impossible. These characteristics prevent the work on smaller spaces. Thus, the study of smaller customized manipulators starts. [ 30 ] developed a manipulator for a mini UAV. Works such as [ 31 ] analyzed the influence of the manipulator in the flight. These works made it possible to reach new horizons in aerial manipulation. Subsequently, works such as [ 32 ] combined the drone’s control, knowing the perturbances generated by a robotic arm with seven Degrees of Freedom (DoF) robotic arm. In [ 33 ], the authors studied the use of compliant robotics arms to reduce the disturbances generated by the weight and possible impacts. Figure 1.5 shows some examples of manipulators considering the environment’s influence on the platform. 1.1 Motivation and Objectives 5 (a) 7 DoF manipulator from [32]. (b) Compliant manipulator from [33]. Figure 1.5 Manipulators’ publications to reduce their influence on the flight. The basis of AM was set at that time. Works like [ 34 , 35 , 36 ] made a considerable effort to register these evolutions. However, aerial robotics continues its evolution. Therefore, new requirements appear interaction between robots and humans. Traditionally, robotics was isolated from humans; an excellent example of this isolation is using robotic arms in a factory. They used to be separated into cells to avoid damage to humans (see Figure 1.6a). However, it is impossible to isolate an aerial robot. So then, a trend started developing safer aerial robots that made interactions with humans or animals possible, avoiding risk (see Figure 1.6b). Figure 1.6 shows an example of traditional robotics systems against the future of aerial manipulation. (a) Actual robotics arms in a factory [37]. (b) Future robotics soft, bioinspiration and interaction[38]. Figure 1.6 Traditional manipulation and future aerial manipulation. Then, researchers presented two ways of reducing the risk during the UAVs operation. The first solution comprises the aerial platform. Multi-rotors were, at that time, the most used platforms for cooperation. They have sharp propellers that break the air to generate Thrust and maintain the flight. However, these sharp propellers can easily damage the cooperators. Therefore, researchers looked for some solution to make assumable the risk in cooperation with the platform. This solution comprises the isolation of that risk. The idea was to maintain the flight performance and isolate the propellers. In that way, several researchers tried to develop 12 Chapter 1. Introduction (a) Bioinspiration. (b) Cooperation. (c) Soft-Flap. Figure 1.12 Thesis objectives. 1.2 Thesis Framework This research work has been funded under the framework of General compliant aerial Robotic manipulation system Integrating Fixed and Flapping wings to INcrease range and safety (GRIFFIN) [ 70 ] and AERIAL COgnitive integrated multi-task Robotic system with Extended operation range and safety (AERIAL-CORE) [ 71 ] projects. GRIFFIN is an European Research Council (ERC) Advanced Grant hosted by the University of Seville with grant agreement ID 788247. This project tries to overcome the capabilities of the ornithopters, giving them manipulation capabilities and improving flight abilities. AERIAL-CORE is a Horizon 2020 (H2020) project coordinated by the University of Seville with grant agreement ID 871479. This project involved a big consortium of different European universities. AERIAL-CORE project aims to improve the conditions for inspecting and maintaining powerlines. The project proposes using aerial vehicles to mitigate the risks of height works in powerlines. The author carried out an stay at Laboratory of Intelligent Systems (LIS) École Polytechnique Fédérale de Lausanne (EPFL) to study novelty soft electro-adhesion actuators to contribute to the manipulation. These actuators aim to control the adhesion to the surface actively, improving the grip generated by the manipulator. This study is crucial to improve the balance during the perching. However, their application in aggressive environments is still under development. 1.3 Thesis Outline 13 1.3 Thesis Outline The following list presents the thesis structure summarizing each chapter’s content. • Chapter 2: This chapter establishes the foundation of the developments in this thesis. First, it focuses on the avian capabilities aimed to mimic in the aerial robot. Then continues with analyzing the flying conditions of the aerial platforms used. Finally, the chapter describes the soft robotics technologies used, focusing mainly on actuation technologies and the works found in the state of the art. • Chapter 3: The chapter focuses on developing systems to provide manipulation to ornithopters. First, it starts describing, analyzing, and characterizing the soft actuation technologies used in the manipulator. It then discusses the design process and evolution of the manipulator through various versions. Continuously, the chapter presents the integration of the actuation systems to reach the desired specifications. Finally, at the end of the chapter, the integration of the manipulator into the aerial platform is shown. • Chapter 4: The flight capabilities of ornithopters limit the rest of the systems used. In that way, the chapter focuses on developing soft bioinspired aerodynamic surfaces to improve the actual flight capabilities of ornithopters. The chapter divides the developments into the wings and tail. Concerning the tail, an innovative solution is presented using morphing soft actuators. This section will describe the design process, actuators used, and integration in the platform. On its side, the wings present a novelty geometry avian-inspired, capable of generating actuation on the flight and improving the gliding capabilities of traditional ornithopters. The section describes the geometrical analysis, designing process, manufacturing, and integration of the wings in the aerial platform. • Chapter 5: This chapter presents the experimental validation of the systems developed. The experimental validation will assess the flight behavior, manipulation capabilities, and safety added to the platform by the soft bio-inspired devices developed in this thesis. •Chapter 6: Presents the author’s conclusions and future research line of the work. Figure 1.13 presents the interrelationship between the different chapters. All the chapters contribute to developing the soft cooperative ornithopter concept highlighted in the center of the figure. Chapter 2 sets the objectives wanted to replicate from nature and sets the limitation and initial conditions of the developments. Chapter 3 and 4 overcomes the limitations found in Chapter 2. These chapters develop and design the devices to improve the flight stage of the platform and provide manipulation capabilities. Chapter 5 evaluates the system’s behavior developed to reach the desired objectives. Finally, Chapter 6 presents the author’s conclusions about the work and developments performed during the investigation. At the end of this chapter, the author will present the future research lines that will improve the solutions carried out. 14 Chapter 1. Introduction Figure 1.13 Thesis Outline. 1.4 Contributions Five research works have been published in my Ph.D. The following list summarizes the main contributions of this dissertation. • Development of soft adaptable manipulators capable of interacting with humans and perching on various places. • Development of soft bioinspired morphing surfaces to improve the flight capabilities of ornithopters. •Study of low-weight soft technologies to be used in aerial robotics. •Study of the ornithopters perching to generate avian-inspired solutions. The following list organizes the publications chronologically starting from the last one and showing their title and abstracts. 1.4 Contributions 15 [ 72 ] High-Performance Morphing Wing for Large-Scale Bio-Inspired Unmanned Aerial Vehicles. IEEE Robotics and Automation Letters (June 2022). E. Savastano, V. Perez-Sanchez, B. Arrue, and A. Ollero. This letter proposes a novel bio-inspired wing design to improve some characteristics of Flapping Wing Unmanned Vehicles (FWUV) related to their potential applications such as payload capability, maneuverability, low injury risk, and energy improvement. The suggested solution takes advantage of a broadly based avian research, focusing on the integration of advanced bird-like features ranging from bird-like high-compliant airfoil to active morphing strategies to mimic the ways birds naturally manage flights. The proposed conceptual design is supported by Unsteady Vortex Latex Method simulations provided by a novel flapping-oriented open-source solver. Prototype validation relies on both test-bench and real-flights results. A motion capture system is used to validate the wing aero-elastic characteristics. The resulted improvements are highlighted by a comprehensive comparison with different similar-size prototypes. [ 73 ] Analysis of Forces Involved in the Perching Maneuver of Flapping-Wing Aerial Systems and Development of an Ultra-Lightweight Perching System 2021 International Conference on Unmanned Aircraft Systems (ICUAS) (July 2021). V. Pérez-Sánchez, A. Gómez-Tamm, F. García-Rubiales, B. Arrue, and A. Ollero Trying to optimize the design of aerial robotics systems, this work presents an optimized low-weight landing system for flapping-wing aerial robots. The design, based on the use of low-sized neodymium magnets, intends to provide that these aerial robots have the capability of landing in restricted areas by using the presented solution. This capacity will increase the application range of these robots. A study of this situation has been done to analyze the perching maneuver forces and evaluate the system. The solution presented is low-weight, low-sized, and also relatively inexpensive. Therefore, this solution may apply to most ornithopter robots. Design, analysis of the implied forces, development and experimental validation of the idea are presented in this work, demonstrating that the developed solution can overcome the ornithopter’s payload limitation providing an efficient and reliable solution. [ 74 ] Bio-inspired morphing tail for flapping-wings aerial robots using macro fiber composites Applied Sciences (March 2021). V. Perez-Sanchez, A. E. Gomez-Tamm, E. Savastano, B. C. Arrue, and A. Ollero The aim of this work is to present the development of a bio-inspired approach for a robotic tail using Macro Fiber Composites (MFC) as actuators. The use of this technology 16 Chapter 1. Introduction will allow achieving closer to the nature approach of the tail, aiming to mimic a bird tail behavior. The tail will change its shape, performing morphing, providing a new type of actuation methodology in flapping control systems. The work is intended as a first step for demonstrating the potential of these technologies for being applied in other parts of the aerials robotics systems. When compared with traditional actuation approaches, one key advantage that is given by the use of MFC is their ability to adapt to different flight conditions via geometric tailoring, imitating what birds do in nature. Theoretical explanations, design, and experimental validation of the developed concept using different methodologies will be presented in this paper. [ 75 ] SMA actuated low-weight bio-inspired claws for grasping and perching using flapping wing aerial systems 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (February 2021). A. E. Gomez-Tamm, V. Perez-Sanchez, B. C. Arrue, and A. Ollero, Taking inspiration from nature, the work presented in this paper aims to develop bioinspired claws to be used for grasping and perching in flapping-wing aerial systems. These claws can be 3D printed out of two different materials and will be capable of adapt to any shape. Also, they will be soft for avoiding undesired damages on the objects when performing manipulation. These claws will be actuated by shape memory alloys (SMA) springs to get rid of the weight of traditional servos. The design of all the components will be explained in this work. Also, the challenges of being able to control SMA using only a LiPo battery on an aerial vehicle will be exposed. The solutions applied and electronics used will be also described. Lastly, experiments made both in test bench as on flight will be summarized. [ 76 ] A linearized model for an ornithopter in gliding flight: Experiments and simulations 2020 IEEE International Conference on Robotics and Automation (ICRA) (September 2020) R. Lopez-Lopez, V. Perez-Sanchez et al. This work studies the accuracy of a simple but effective analytical model for a flappingwings UAV in longitudinal gliding flight configuration comparing it with experimental results of a real ornithopter. The aerodynamic forces are modeled following the linearized potential theory for a flat plate in gliding configuration, extended to flapping-wing episodes modeled also by the (now unsteady) linear potential theory, which are studied numerically. In the gliding configuration, the model reaches a steady-state descent at given terminal velocity and pitching and gliding angles, governed by the wings and tail position. In the flapping-wing configuration, it is noticed that the vehicle can increase its flight velocity 1.4 Contributions 17 and perform climbing episodes. A realistic simulation tool based on Unreal Engine 4 was developed to visualize the effect of the tail position and flapping frequencies and amplitudes on the ornithopter flight in real time. The paper also includes the experimental validation of the gliding flight and the data has been released for the community. 2 Bio-Inspired Soft Flapping Wing Aerial Robots(FWAR) This chapter summarizes the main concepts, limitations, and technologies that make it possible the development of the soft bioinspired cooperative Flapping Wing Aerial Robots (FWAR). This chapter aims to set the research base and the developments’ initial conditions. The chapter starts by analyzing the capabilities of birds in nature. The objective is to study the skills of the birds that the ornithopter has to replicate. Then the chapter continues analyzing the contour conditions of the development generated by the aerial platform. This part of the chapter is crucial to examine the limitations of the systems developed. In the end, the chapter presents the soft robotics technologies that can be used considering the platform’s restrictions. 2.1 Bioinspiration Nature is the world’s leading science, and evolution is its result. Evolving over millions of years, the systems shown in nature are almost perfect. This section focuses on the analysis of the solution achieved by nature regarding birds. Specifically, It evaluates the interaction capabilities of birds with the environment and other animals and the flight behavior that makes it possible. This section aims to develop a concept about the avian-inspired systems that ornithopters want to replicate. Causes evidence that bird inspires ornithopters. However, ornithopter capabilities are still far from birds’ ones. Nevertheless, this thesis aims to develop a platform as close as possible to birds solving the current robotics problems. Fortunately, the birds’ capabilities in nature and the current robotics problems are similar. They exhibit excellent interaction capabilities and dexterous manipulation skills. Furthermore, they are the world’s most excellent flight experts. This thesis aims to mimic these performance capabilities while replicating the bird systems to perform theirs. Evolving from dinosaurs, birds show an exceptional adaptation to the environment. In the beginning, birds’ ancestors used to walk on the ground. At that time, they were much larger than nowadays. However, they were smaller than their competitors, and they used to be the prey of giant carnivorous hunters. Evolution gave them feathers and light bones that were unuseful initially. Over the years, birds improved their flying ability. Consequently, their arboreal habits took prominence to avoid their ground vulnerability. 2.1 highlights the evolution of birds from the ground to the trees. Figure 2.1a shows the most ancient 19 20 Chapter 2. Bio-Inspired Soft Flapping Wing Aerial Robots(FWAR) (a) Deinonychus [78]. (b) Archaeopteryx [79]. Figure 2.1 Birds’ ancestors. non-avian ancestor and Figure 2.1b the first bird from the Late Jurassic Archaeopteryx at right [77]. The Figure shows the beginning of the evolution of some of the current characteristics of the birds in their ancestors. As shown in the figures, birds have feathers from the beginning unuseful for flight for Deinonychus and gave limited flying capabilities for Archaeopteryx. Another part of their bodies that realize their flight ability is the tail. However, their tail is not aerodynamic at all. They used the tail for balance and not for controlling their flight. Consequently, their flight was short at low speeds. Figure 2.1 also shows the first presence of a prominent beak. The birds used their beaks to tear the tough flesh of their prey. Everything surrounding them was aggressive. Consequently, the Figure shows huge claws prepared to work with their beaks to survive in that environment. In the beginning, the bird’s claws were the more prominent systems. The birds used their claws for locomotion and hunting. However, evolution made their flight ability increase step by step. Then, birds abandon the ground, and their arboreal habits incremented. Nowadays, birds live in a less aggressive environment. In the current world, everything is smaller than in the time of the presented birds’ ancestors. Their preys were smaller than before their feeding method also changed. At the same time, their flight abilities increased. Consequently, birds also reduced their size, and their wings and tail gained prominence on their bodies. The current flight abilities made the birds the reference for all the aerial systems developed. Birds show dexterous flight abilities. The combination of tail and wings made it possible to perform dexterous maneuvers in the air. Also, they increment the interaction with the environment making it possible to interact with them. However, the variety of species of birds in the world made it challenging to evaluate the specie that is wanted to replicate. In that way, this section analyzes the different species and the interaction capabilities of each one. The aim is to set a base for our developments and replicate them. The text divides the analysis into three. The first one is the manipulation capabilities. Depending on the specie, they demonstrated different adaptations to human objects or are used to living in different environments. 2.1 Bioinspiration 21 (a) Golden eagle hunting a rabbit [80]. (b) Ospreys hunting a fish [81]. (c) Parrot holding a pencil [82]. (d) Crow eating [83]. (e) Barn owl eating. (f) Secretary bird hunting. Figure 2.2 Birds’ manipulation skills. 2.1.1 Birds Manipulation Capabilities Birds show dexterous manipulation capabilities. Evolution improves the adaptability of the birds’ claws, enhancing their manipulation skills. Their main use in nature is for feeding. Hunting animals or helping tear the flesh with their beak is the primary utility. Nevertheless, the presence of humans in the environment is relaxing these skills. This presence reduced the difficulties of finding food. Mainly, vegetarian species have easier access to food from the expansion of agriculture. Their skills in the manipulation of objects come from their captive uses. They do not use to find these objects in nature. However, humans forced this situation over the years, improving their abilities. Figure 2.2 shows different examples of manipulation concerning a wide range of bird species and claws morphologies. The analysis begins with the group of the Accipitridae. Figures 2.2a and 2.2b show two examples of the manipulation skills of these birds. The specialty of these birds is the hunt. They can carry prey even larger than themselves. They used the hocked shapes of their claws to go inside the skin of their targets. It makes to reduce the adaptability of their claws. However, they distribute the stress along the foot and legs, taking advantage of their nails and the longitudinal stiffness of their toes. To absorb these stresses, the mass of the manipulation body takes almost 40% of the entire body mass. They combine these manipulation skills with their flying prowess. They can fly at high velocities with high maneuverability to reach their prey. They also show high-gliding abilities to control their target in the surrounding environment. Then, the lift and thrust generated by their wings can transport their prey to a place where they can eat safely. The second group analyzed is the Psittacidae. Figure2.2c show an example of their 28 Chapter 2. Bio-Inspired Soft Flapping Wing Aerial Robots(FWAR) behavior, and capabilities. In that sense, Table 2.1 summarizes the main characteristics of the platform flight. Table 2.1 E-Flap[68] main characteristic. Weight (Kg) Maximum Payload (kg) Maximum Speed (m/s) Flapping Frequency (Hz) b (m) Wing Section 0.51 0.520 6 5.5 1.5 Flat Plate / 1 4c This platform mimics the birds replicating the flapping motion and using the tail to control the trajectory. However, their aerodynamic characteristics limit their capabilities for flying. Payload is one of the most limited flight capabilities. The initial ornithopter can carry 520 g in this configuration, including the system needed for a basic flight. This reduced load limits possible applications and components used in the flight. However, the development of this dissertation improves this limitation. Focusing on the flight behavior, the average flight speed of the platform is 6 m/s at a flapping frequency of 6 Hz. These speeds and flapping frequencies are high for shortdistance tasks. Furthermore, observing nature, this behavior is typical of small birds. However, the ornithopter has a 1.5 m wingspan. The aerodynamic conditions of the wing cause this behavior. The wing has a flat-plate airfoil with a short chord. This airfoil increases the frequency required to fly, reducing lift and thrust generation. Nevertheless, they maintain stability and minimize flapping inertia. On the other hand, the airfoil’s flat shape reduces the platform’s glide capability. As a result, the glide phase with these airfoils is limited, and they cannot generate lift by varying the airflow. Consequently, it is not easy to use the platform for long-range flights. In addition, the glide with this profile involves high speeds losing more altitude. The platform controllability with this setup is also limited. The only control surface on it is the tail. The tail has little influence on the flight due to the disturbances generated in the airflow by the flutter. Instead, it increases the effect in the gliding phase. However, more actuation is needed to have the trajectories perfectly controlled. The text above analyses the current state of the platform and its behavior. The section now focuses on the influence of these characteristics in the proposed developments. This thesis overcomes the problem from two different sides. The first is developing a system that allows manipulating and perching the ornithopter in a safer interactive way. The second is to increase the flight capabilities considering the cooperation and contributing to the manipulation. The flight velocity and weight took special relevance in the development of the manipulator. Usually, in Robotics, perching and manipulation are two problems well distinguished. The traditional solution in Robotics is to isolate them and generate two different actuators. However, this thesis tries to achieve them with the same manipulator and controlling system. Birds can perform perching and manipulation tasks with the same actuator. Their perching is smooth: they stop in the air and prepare the whole body to perform the perching, as shown in Figure 2.3f. 2.2 Ornithopthers Flapping Wings Aerial Robots(FWAR) 29 However, ornithopters can not perform a controlled stop in the air before perching. Concerning ornithopters, perching is an impact on the manipulator in this case. The aerodynamic characteristics of the wing set the velocity of the perching. The lower speed at which the current setup still has control of the flight is around 3 m/s. This impact generates considerable energy that the manipulator has to adsorb. Additionally, the reaction time is extremely short. The combination of the high impact with the reduced reaction time makes the claw system very restrictive. This is improved by the development in this thesis of a new wing. In the future, considering the wing velocity could also reduce the impact as birds do. After the contact, the manipulator must guarantee a tight holding generating enough friction to maintain the balance. Usually, this problem is solved in traditional robotics using sharp spikes. However, the primary objective of the systems developed is to cooperate with humans. So in that way, these kinds of sharped solutions are not appropriate. Here takes particular relevance to the soft robotics concepts. Soft robotics is closer to the limitations of manipulation and cooperation with humans. However, the energy involved in the perching makes it challenging to use a soft system. Concerning manipulation, the platform’s main limits are weight and aerodynamic influence. The system developed also has to consider the load’s position during the flight. A change in the balance on the platform causes a disturbance on the flight. This disturbance has to be corrected by the control actuators. Therefore, reducing this influence on the flight will increase the performance. Concerning the weight of the manipulator, the aerial robotics platform’s payload limit is around half a kilogram with the actual setup. The payload restrictions are tough to achieve the desired results in the project providing manipulation, visual, and control capabilities. After analyzing the requirement of the different parts, the remaining load for the manipulation is 150 g, less than 15% of the weight of the aerial platform. In Section 2.1, the manipulation and perching system evaluation in high-size birds is around 30%. Then the system has to overcome the solutions found in nature. The reduced payload for the manipulation system makes it necessary to evaluate novelty actuation methods. These methods use smart novelty materials to increase the actuation capabilities while lowering the system’s weight. Concretely, it uses Shape Memory Alloys (SMA) to mimic the muscles of birds supplying that behavior. However, studying bioinspired aerodynamic surfaces of the platform relaxes the payload restriction. The wings can generate a greater thrust during the flapping, improving the gliding phase. The aim is to develop bird-shaped wings and abandon the concepts of flat wings traditionally used for ornithopters. The wings developed in this dissertation also have a soft structure improving the interaction skills of the platform. The design mimics the feathers of the birds. The wings also contribute to the control of the ornithopter. Traditionally, ornithopters only use their wings for lift and thrust generation. However, the new design can modify their surface as birds do. This development generates actuation on the trajectory. Adding this control in the wing makes it possible to manage the payload added. The other control surface is the tail. This aerodynamic actuator is the only one in the initial setup. This tail uses the concepts learned from fixed wings. However, the behavior of the flight demonstrates that its actuation in Flapping Wing Aerial Robots (FWAR) is 30 Chapter 2. Bio-Inspired Soft Flapping Wing Aerial Robots(FWAR) limited under certain conditions. Therefore, this thesis analyzes to include bioinspired concepts in the tail. However, the tail and its position are restrictive. The distance between the tail and the Center of Gravity (CoG) of the platform is long. This longitude generates a moment that increases with the weight of the tail. This thesis analyses novelty morphing actuators to overcome this problem. These actuators will mimic the feather while reducing the total weight of the tail. The following list summarizes the main requirements generated by the aerial platform for the proposed developments. • Payload: This may be the biggest drawback to using a FWAR. The weight that an ornithopter can carry out is smaller than other traditional aerial platforms. The current solution is to use a flat plate wing section to make the payload smaller. This characteristic limits the payload and, as a consequence, the applications of the platform. This thesis overcomes this problem by designing bioinspired surfaces capable of loading more weight. • Controllability: The ornithopter’s current state shows limited trajectory control. The disturbances caused during the flapping limit the platform’s maneuverability. The proposed solution is to add more control degrees to the platform abandoning standards solutions. The aim is to mimic nature in all the control surfaces, considering the traditional aerodynamics but not replicating these systems. • Perching behavior: The state of the platform at the beginning of the thesis causes restrictive perching conditions. The contact velocity is bigger than the one found in the perching of birds. Therefore, It causes the developed perching mechanism to abandon some bioinspired concepts to focus on the contour conditions of the platform. • Cooperation behavior: Ornithopter shows excellent human cooperation skills. For example, they do not have sharp edges like propellers in traditional platforms. However, the proposed mechanisms improve its performance. They make the platform safer, making it possible to interact physically with humans. The manipulation skill was added to manage the contact. Moreover, the softer surfaces also reduce the damages caused in case of failure. Ultimately, this investigation goes through the current state of the ornithopters, considering the platform’s limitations. The aim is to obtain a platform safe for interaction and capable of cooperating with humans. In that way, bioinspiration is crucial in all the development presented. However, the platform flight contour conditions limit the bioinspiration of the developed systems. To reach complete bioinspiration, the knowledge about the controllability of the ornithopter must increase. Nevertheless, the development proposed made it possible to reach new horizons. They give ornithopters manipulations skills, better cooperation skills, and extended control capabilities. 2.3 Soft Robotics Technologies 31 2.3 Soft Robotics Technologies This work has been developed under the framework of the ERC Advanced Grant GRIFFIN [ 65 ]. The grant aims to provide ornithopters with dexterous manipulation skills and improved flight capabilities. E-Flap [ 68 ] is one of the platforms developed in this Advanced Grant. This thesis aims to transform E-Flap into a more interactive platform following soft robotics and bioinspiration principles. Ultimately, this platform is called Soft-Flap. Soft-Flap has manipulation skills and improved flight behavior. Soft-Flap follows the current trends in robotics, trying to develop safer interaction mechanisms adapted to aerial robotics. The goal is to mimic the behavior of birds in nature. The authors in [ 43 ] underline the necessity of using soft robotics technologies to replicate natural systems. In short, the aim is to mimic nature using soft robotics as the robotic technology that makes this possible. Figure 2.6 is an excellent exponent of the aim of this work. Figure 2.6 A Harris hawk about to land on a branch [98]. Figure 2.6 shows the perching approach maneuver of a hawk. The entire body of the hawk is working to reach the perching spots. The body could be divided into four parts: body, claws, tail, and wings. The tail and wings are the control surfaces. The bird has them fully opened to stop the airflow. The powerful legs and claws are preparing for the contact. All the systems shown in the figure are experimenting a considerable deformation. The feathers break the airflow, and the legs absorb the impact. The deformation experienced by these systems during the perching highlights their soft characteristics, This thesis studies the application of soft technologies in two kinds of systems, the wings and tail, as morphing actuators and in the claws and legs system. However, applying soft technologies in these applications is challenging due to the force needed to exert. Therefore, the system find an equilibrium between forces and deformations to reach the desired actuation. This dissertation studies and adapts soft technologies to aerial platforms. Nowadays, soft robotics is well-known in-ground robotics. However, there are few examples of these technologies in aerial robotics. Nevertheless, some soft technologies offer low-weighted developments with higher capabilities for aerial robotics. In that sense, this work analyses the use of soft technologies in aerial platforms to improve their interaction capabilities. That opens a new vision of soft robotics in aerial platforms with interaction capabilities. 32 Chapter 2. Bio-Inspired Soft Flapping Wing Aerial Robots(FWAR) Regarding manipulation, the goal is to develop an avian-inspired manipulation system for ornithopters. Soft technologies increase claw interaction capabilities overtaking the bird ones. One of the leading soft actuation technologies studied during the development of this thesis is tendon-driven actuation. This technology replicates the connection between the muscles and tendons in avian claws. However, traditional actuation methods cannot replicate the force exerted by animal muscles. In that way, using novel soft materials make replicating avian muscles’ behavior possible. The claw developed use SMAs as muscles and tendon-driven actuation to deliver the force to the toes. The claw keeps the soft characteristics also in its structure to enable adaptable manipulation and resolve the problem of perching. There are no record examples of systems combining SMAs materials for actuation and tendon-driven actuation in a gripper. This lack of developments is because of the recently discovered applications SMAs. In that way, the analysis presents both solutions separately. The evaluation starts with analyzing the oldest method. The first record use tendondriven actuation for a soft actuator was in 1978 in [ 44 ]. This conceptual work lays the foundation for tendon-driven actuation. However, the actuator’s capabilities were limited. In [ 99 ], the authors increased the complexity of the prototype by replicating a human hand with high payload capabilities. However, the development did not follow soft robotics basis. The author used hard material in the contact and a non-continuous structure. In [ 100 ], the authors developed continuous joints using deposition manufacturing and analyzed their behavior compared with traditional joints. This one could be one of the first examples of soft robotics as we know it today. This work analyzed continuous base structures. This concept is later followed by soft robotics. Figure 2.7a shows this gripper. The following is one of the first examples of functional soft robotics grippers using tendon-driven actuation as we know its today. In [ 101 ], the authors developed an anthropomorphic hand using flexible joints covered with Polydimethylsiloxane (PDMS). In addition, the author used capacitive sensors to measure the actuation. Figure 2.7b shows this development. In [ 102 ], the authors developed a gripper using a continuous structure of variable stiffness PDMS materials. This gripper showed dexterous adaptation capabilities and high friction generation. However, Its payload capabilities are low due to its highly deformable structure. The examples presented above comprise ground robotics mechanisms. There are no strictly soft developments used in aerial robotics. The absence of prototypes in aerial robotics is because of the uncertainties generated by the soft actuation. Their actuation is complex to model and must be revised to develop high-load demanding systems. However, there are grippers in aerial robotics using tendon-driven actuation without soft robotics characteristics. One of the leading examples is [103] shown in Figure 2.7c. The authors present a tendon-driven system to perch using a quadrotor. This work is similar to the one analyzed in this thesis. However, some differences increase the complexity of the work developed in this thesis. The first one is the use of a different aerial platform. The use of ornithopters generates a problem completely different, as Section 2.2 presents. The second one is the use of a soft claw. The claw shown in [ 103 ] relaxed requirements for the manipulation. This claw is more oriented toward smooth perching than manipulation. In this Ph.D. thesis, SMAs replicates the effect of the muscles in the legs of the birds. SMAs have been widely studied since the 90s. SMAs are an alloy of Nickel and Titanium. 2.3 Soft Robotics Technologies 33 (a) Gripper developed in [ 100 ] by the study of additive manufacturing of flexible joints. (b) Antrophomorophic soft gripper developed in [101]. (c) Bird inspired perching mechanism developed in [103]. Figure 2.7 Cable driven grippers examples. They can change their internal structure with the temperature. Therefore, They can recover the initial memorized shape. Thermal treatments performed in the alloy under certain stress conditions store this shape. Traditionally, SMAs are an alloy of Nickel and Titanium. However, other materials can be added in a minor composition to reach some specific properties. Two types of SMAs can be developed depending on the thermal treatment performed, one-way SMAs and two-way SMAs. The one-way SMAs have one shape memorized, and they recover it by increasing the temperature. The two-way SMAs have two memorized forms. They can change between them depending on the transition temperatures. The use SMAs in soft grippers can vary depending on the application. The first is SMAs working as force generators inside the structure, substituting the tendons. In works like [ 49 ], the authors developed a caterpillar-based robot using SMAs springs as internal tendons and muscles. In [ 50 ], the authors evaluate the use of SMAs as bending actuators for soft robotics applications. In this case, the actuator has the shape of a sheet that reduces the radio formed depending on the temperature. In [ 51 ], the authors develop a micro gripper using SMA wires. The works above show three different ways of using SMAs. The first is in a spring shape, the second is in the shape of a sheet, and the third is in a wire shape. This Ph.D. thesis uses SMAs with springs shape. The main reason is the linear force they can generate in a reduced space with a low weight. These capabilities make SMAs spring an excellent candidate for replicating avian muscles on ornithopters. The other proposal using soft technologies in this dissertation is their use to mimic birds’ flight mechanisms. The goal is to develop soft morphing aerodynamic surfaces to improve control capabilities and reduce risky situations in case of fault. [ 49 ] shows the soft morphing concept applied to ground locomotion. In this case, the authors used SMAs to vary the body of a caterpillar, generating locomotion on the ground. In this work, the environment is the air. So the necessity to create aerodynamic surfaces to control the trajectory. The literature shows examples of generating soft aerodynamic control surfaces in fixed wings platforms. The idea is to adapt this technology to a flapping wing device. Particularly, the interest of this Ph.D. thesis is to use MFCs. MFCs were developed by NASA in the 90s. 34 Chapter 2. Bio-Inspired Soft Flapping Wing Aerial Robots(FWAR) (a) Microgripper developed in [ 51 ] using SMAs. (b) Antrophomorophic gripper using SMAs bending actuators developed in [50]. Figure 2.8 SMAs grippers examples. Initially, they were conceived as sensors to measure small strains and displacement. This use reflects its capability to change its internal resistance depending on its deformation. This behavior generates works for structural applications like [ 104 ]. The authors developed a low-cost structural sensing actuator with low weight and precision in measuring small displacements and vibrations. However, researchers became aware of other properties that MFCs have. The first feature that focused the researchers’ interest of researchers in these actuators was their blocking force during the actuation. This behavior generated the first applications of MFCs in soft robotics. In works such as [ 56 , 57 , 58 , 59 , 62 ], the authors evaluated the use of MFC in underwater robots. They developed control surfaces for robots underwater by exploiting the isolation properties of the MFCs and blocking forces. These works point out the potential applications of these actuators to develop control surfaces. Concerning aerial applications of MFCs, some researchers started using them as aerodynamic control generation replacing traditional mechanisms. In works like [ 105 , 106 , 107 ], the authors developed ailerons for a fixed-wing using MFCs. These works aimed to control the aircraft trajectory with the developed actuator. They show the potential application of MFCs in aerial vehicles. Nevertheless, they did not follow the soft robotics basis, and the platform had different flight conditions. (a) Active wing and tail using MFCs actuators [106]. (b) Active wing using MFCs varying the dihedral angle from [107]. Figure 2.9 MFCs control surfaces generation. 2.3 Soft Robotics Technologies 35 The results of this thesis are reflected in the platform Soft-Flap.Soft-Flap is an ornithopter completely avian-inspired. Their wings are flexible and capable of actuating in the trajectory as birds do. The wings of Soft-Flap abandon the classical concept of the wing for ornithopters shown in state of the art. Their aerodynamic profile is avian-inspired, leaving the flat profiles presented usually in the ornithopters. It increases the capabilities of gliding of actual ornithopters. This profile mimics the deformation suffered in the feathers during the flapping to increase the lift and thrust generation. The deformation of the wing follows the soft robotics concept. The geometry of their wing is variable to generate actuation in their trajectory. The wings have a flexible tip made of elastomer material fabric, generating a differential actuation over the ROLL of the platform. These wings increase the control capabilities, payload, and safety during the operation from traditional ornithopters. Soft-Flap have an avian-inspired tail. The tail is made of MFCs following the concept learned from the literature. However, this tail has to overcome these developments performing major deformations and being capable of supporting the flapping disturbances. The tail mimics the behavior of the feathers of the avian tails. This inspiration increases the control capabilities of the current ornithopters. The claw of Soft-Flap provides manipulation skills. They are avian-inspired, following the concept of soft robotics. These claw aims to generate manipulation skills for ornithopters. The avian inspiration is crucial for their development as well as the contour conditions of the manipulation and perching of the platform. Ultimately, these claws must overcome the avian claws increasing their capabilities to interact with humans. Soft-Flap is a safer ornithopter with improved control and payload capabilities. Bioinspiration and soft robotics are crucial to developing this concept of platform. 3 Bio-Inspired Manipulators for Ornithopters This chapter focuses on the development of Soft Bio-Inspired manipulation systems for ornithopters. This development is motivated to provide manipulation capabilities to the ornithopter presented in Chapter 2.2. This kind of aerial platform is restrictive regarding payload and aerodynamic influence. Therefore, this development will consider these requirements to enhance flight capabilities. "Soft" and "Bioinspiration" are the two main characteristics of this work. A few years ago, robotics’s main problem was controlling the aerial platform. Nowadays, cooperation tasks are at the edge of robotics. Traditionally, systems in robotics were stiff and adapted to a specific application. It is challenging to cooperate between humans and robots using rigid actuators due to the high injury risk to humans. In that way, the solution is to develop systems adaptable to multiple shapes and safer for interaction. The system developed follows the concept of having an aerial platform adapted to the environment and safer for cooperation. Bioinspiration is the other main keyword of this work. The system sets aims to mimic the behavior and capabilities that the birds have with their claws. In that way, this work analyses different birds’ morphology while reaching the requirements of the aerial Platform [ 68 ] to achieve better results. Finally, after developing the thesis, the platform will be fully inspired by birds with soft capabilities to enhance their interaction skills. The system developed aims to contribute by adding capabilities to different tasks. The following list summarizes these tasks: • Human-Cooperation: The claw developed has to be safe for human cooperation following the requirements of the ERC Advanced Grant GRIFFIN [ 68 ]. Ornithopters are one of the most safer aerial platforms for human interaction. Its low flight velocities and the absence of sharp edges like propellers make it a good option. Therefore, the manipulation system must properly maintain these capabilities to enhance interaction with the environment and humans. In that way, developing a soft actuator supports these capabilities making it possible to interact with humans using the claws. • Manipulation Capabilities: Adaptability and soft systems usually are interrelated. The aim is to reach a system capable of manipulating multiple shapes, sizes, and objects. 37 44 Chapter 3. Bio-Inspired Manipulators for Ornithopters Figure 3.3 Actuation cycle parameters. is the shear modulus of the martensitic phase. The equation uses the effective diameter due to the reduction of the spring diameter during the motion. In Equation 3.5, GA is the shear modulus on the austenitic phase. Finally, The equation sets the displacement in terms of the force applied. In this work, the interest is to analyze the tension generated by an actuator with a specific deformation. Then, the aim is to control the system by using the equilibrium of two springs in opposite directions. The main problem with these actuators is that the provider doesn’t specify the properties of the actuator, so in that way, the study of these properties sets the behavior of the alloy. Assuming all the springs provided have the same properties and the alloy is the same. The studies of the material performed before contribute to determining that in the case of the same properties and shape. The force generated depends only on a constant. That simplifies the model, making it possible to achieve more accurate results. The test focus on obtaining this constant. 3.1.2 Modeling SMA Muscles Muscles dynamic model The first experiment performed evaluates two different springs. These tests aim to select one regarding maximizing the force/weight ratio. Table 3.1 shows the geometrical physic properties of these springs. The test analyzed the spring’s behavior, making its deformation stable and varying its temperature. Figure 3.4 shows the results. By analyzing the graphic, Spring 1 can exert forces up to 24 N with an elongation of 350% at 200 ◦ C. The Spring 2 muscle reaches up to 95 N with similar conditions. The weight of the second muscle is two times bigger than the first. However, the force reached is almost four times bigger, having a better Force/Weight ratio. As determined before, the strength exerted is crucial to replicate the birds’ muscles’ behavior. Also, the alloy should overcome the requirements of Section 3.3. 3.1 Shape Memory Alloy(SMA) Artificial Muscles 45 Table 3.1 Geometrical parameters physic of the Nitinol analyzed springs. The table shows the properties of the spring, focusing on their geometry and mass. Parameters Spring 1 Spring 2 Undeformed length (cm) 2 1.7 Spring Diameter (mm) 6 10 Wire Diameter (mm) 0.75 1.5 Weight (g) 3 6 (a) Spring 1: force in terms of the temperature and elongation. (b) Spring 2: force in terms of the temperature and elongation. Figure 3.4 Springs analysis. (a) shows the force that Spring 1 can exert in terms of the temperature( ◦ C) and elongation( △L ). (b) shows the force that Spring 2 can exert in terms of the temperature( ◦ C) and elongation( △L ). The legend shows in both graphics shows the deformation of each curve. After performing several tests, the spring experiments show some permanent deformation after overpassing the temperature of the 99% of the austenitic phase. Evaluating this temperature prevents the loss of properties during the actuation. By testing our alloy, the 46 Chapter 3. Bio-Inspired Manipulators for Ornithopters temperature is around 150 ◦ C, so during the experiment, the muscle had to work between ambient temperature and 140 ◦ C. This condition avoids permanent deformations, which is one of the major problems of this alloy. The temperature also corresponds with the gradient stabilization shown in Figure 3.4. Regarding the study of the behavior of the SMA Alloy presented in [ 120 ], the transformation between the martensite phase to the austenite is linear in terms of temperature. Then the linear part of the graphic describes the conversion from martensite to austenite. Looking at Figure 3.4, the alloy analyzed in this work starts at 22 ◦ C with 1% of austenite, and the austenite reaches 99% at 140 ◦ C. The conclusion is that the behavior of the SMA is linear between 20 ◦ C and 145 ◦ C. Therefore, the force exerted by the SMA is linear too in this range. The following equation shows the developed model taking into account these characteristics. F(T,x) = FPD +K(x−x0)(T−T0)(3.6) where F is the force exerted, T is the temperature of the spring, x−x0 is the elongation of the spring. FPD is the force needed to perform a plastic deformation on the martensite phase, K is a constant identified from the experiments that set the relationship between temperature increment and the elongation, and T0 is the temperature that sets the martensite composition at 99%. The temperature and elongation define the behavior of the muscles. This model describes the behavior of the actuator between 22 ◦ C and 145 ◦ C. The analysis of the experiment identifies the parameter K of the model. These experiments try to analyze the force exerted by the spring at a specific temperature and elongation. Firstly the study evaluates the gradient in the curve between temperature/force at a particular deformation. Then, this gradient is related to the curve deformation/force at a stationary temperature. The accuracy of the model increases with the amount of data. FPD is the constant value obtained by reaching a permanent deformation on the martensitic phase. The model set it as the value obtained by applying a continuous deformation against a load cell in the martensitic phase. FPD is easier to get and reflects the alloy’s state. After the analysis of the muscle is essential to define the actuation method used on the aerial platform to heat the springs. The electrical model determines the power consumed by the spring, and the energy model analyzes the exchanges with the air to evaluate the temperature of the spring. Muscles electric and energy model The spring’s power consumption models the energy exchanges with the environment. The electrical model set the power consumption of the spring. Therefore, the first step is to analyze how to plug the muscles into the power supply to develop the electrical model. Due to the difficulties of soldering this alloy, this is a challenging task. The alloy doesn’t admit traditional soldering and high temperatures cause fatigue in the material. The study of the three methods set the best one reducing the energy consumption in the connection and guaranteeing conductivity. The first is to make a knot with the cable around 3.1 Shape Memory Alloy(SMA) Artificial Muscles 47 the SMA. Although, unfortunately, this one is not reliable due to the own motion of the spring, the knot loosens a couple of tests. The following method tested maintains the knot using a tin solder. This one shows a better performance than the previous one. However, the knot loses some of the tin due to the actuation temperature and the low melting temperature. Ultimately, a silver-based epoxy provides all the requested requirements, high electrical and heat conduction, high-temperature resistance, and stability during the actuation. Figure 3.5 presents the electrical analysis of the SMA. The figure shows the results of a wide range of experiments. The procurement of the tests was: First, a high-current power supply powers the spring. The Pt100 integrated into the knot then measures the temperature, and the power system software stores the current supplied to the muscle. Figure 3.5 SMA Electrical analysis. The figure on the left side shows the temperature( ◦ C) in terms of the electrical current(A) and time(s). The right shows applied Power(W) during the experiments. The legends show the applied electric current in different colors.. This graphic shows significant results in using the SMA actuated by electricity. The following list is going to summarize the conclusion of the experiments. • Dependence on electric current: The graphic on the left shows the temperature SMA can reach in terms of the current and the time. This temperature has a great dependency on the current supply. Looking at the graphic, the slope of the curve increases with the current reaching higher temperatures in less time. Therefore, with the current increase, the time to reach the temperature is shorter, but the instant consumed power is higher than lower currents. In that way, the system has to be analyzed to evaluate which situation is more favorable regarding energy consumption to reach a specific temperature. 48 Chapter 3. Bio-Inspired Manipulators for Ornithopters • Temperature stabilization: The curves show a stabilized temperature value after a time that increases with the electric current. This result is interesting because there is a balance in the energy exchange with the environment. It makes it possible to estimate the instant energy delivered to the environment. The model must consider all the systems connected to the spring during the actuation because they exchange energy with the environment. This energy is not negligible. This analysis aims to obtain the electrical model of the muscle, including the systems involved in the actuation. The graphic on the right shows the system’s power consumption to analyze the system electrically. One of the main characteristics of this graphic is that the power is stable with the temperature. So the system could be modeled as a constant resistor. After several tests using different temperatures and currents, the mean value of the resistor is 0.23 Ω. The electrical analysis presented in Figure 3.5 opens two possibilities. The first is to use less current, having more time to heat the spring but wasting less instant energy. The second one is to use a higher current, implying less time to heat the spring. Of course, reducing the reaction time is crucial in our system. In addition, however, the system must save as much energy as possible to minimize the battery weight. The battery is one of the heaviest systems in the ornithopter, so reducing energy consumption is one of the ways to reduce the weight. The energy model analyzes the way of actuation to reduce energy consumption. The first step to developing the energy model is to analyze the balance in the energy exchange found in Figure 3.5. Figure 3.6 shows the instant energy lost during the heating process of the spring at 0 m/s of wind velocity. Figure 3.6 Power absorbed by the environment. This graphic represents in blue points the energy delivered to the environment at each temperature. The red line shows the theoretical approach to these energy losses. 3.1 Shape Memory Alloy(SMA) Artificial Muscles 49 The analysis of temperature stabilization during power supply concludes that the environment receives all the energy supplied to the spring during this stabilization. Therefore, the graphic presented in Figure 3.6 shows the power transmitted to the heat reservoir regarding temperature. The literature models of these systems as electrical resistors exchanging with the air. The following equation presents the model: PL=hA(T−T0),(3.7) where PL is the power delivered to the air, h is the heat transfer coefficient, A is the contact area with the air, T is the temperature of the spring and T0 is the enviroment temperature. In this equation, the term hA is a constant determined analytically from Figure 3.6 analysis. The heat exchange area is difficult to estimate due to the number of systems involved in the actuation. The components involved are the longitude of the cables, the control system, and the spring surface. In that way, the analytical approach is better than the theoretical one. Furthermore, h depends on the heat exchange condition and the fluid that wraps the system. In the end, h changes regarding the kind of heat exchange. However, the relationship of this term with other conditions is well-known. After having the energy consumption of the environment related to the temperature and having the time and power consumption to reach this temperature, the following equation shows how to calculate the energy. ETotal =Zt(T) t(To) PTotaldt PTotal =I2R ELost =Zt(T) t(To) PLdt ESpring =ETotal −Elost (3.8) where Etotal is the total energy consumed in Joules, t(To) is the starting point of the calculation, t(T) is the time when the spring reaches T Temperature, Elost is the energy lost during the warming process, and ESpring is the useful energy absorbed by the spring to increase its temperature. Figure 3.7 evaluates the energy losses depending on the temperature and electrical current applied. This figure shows six curves, five corresponding to the energy consumed using different electrical powers, and the dashed line is the energy absorbed by the spring. The blue-shaded area is the useful energy consumed by the muscle. Analyzing the graphic in Figure 3.7 is interesting to focus on the energy losses at a specific temperature by applying different currents, for example, at 80 ◦ C. The blue curve shows the energy absorbed by the spring to increase its temperature. The green curve corresponds to the waste of power at 9 A. The curve closer to the blue one is the 9 A current supplied; its energy loss is less than the rest. So the system loses less energy applying higher currents. Also, the system has to reach the operational limits of working between 50 Chapter 3. Bio-Inspired Manipulators for Ornithopters Figure 3.7 Spring power consumption. The figure shows in continuous lines the electrical power supply to the muscle in terms of the electrical current, and in dashed line, the energy used to warm the spring . 20 and 145 ◦ C. These limits set the minimal power to 8.5 A in wind-stationary conditions. So our lower limit is 8.5 A. The energy losses depend directly on the spring’s exposition time and the energy exchange with the air. As higher the exposition time, more energy goes into the air, and less energy is used to warm the spring. In addition, the forced ventilation the flight generates increases the power delivered to the environment. With these equations, the actuation is modeled on Simulink to analyze the behavior of different muscles configuration. Regarding this thesis, the simulation evaluates the antagonist muscles configuration closing the control loop. The main conclusions of this section are that the spring could be modeled electrically as a resistor of 0.23 Ω. Concerning the energy model, a higher current during the actuation generates lower energy losses reducing the heating time. 3.1.3 Developing SMA Muscles The pandemic situation creates difficulties in obtaining these types of smart materials. The usual suppliers ran out of stock of this alloy, especially in spring shape. However, they do have SMA wires. Therefore, we had to develop a fabrication method to study how to communicate the spring shape to non-deformed SMA. The aim was to develop a custom-made SMA spring with the same properties as the one studied before. The spring has the geometry previously studied, and the provider guarantees that the wire’s alloy is the same. The main problem is communicating the shape and adequately performing the thermal treatment. 3.2 Analysing the Forces Involved in the Perching 51 Firstly, the manufacturing focuses on developing the spring shape with the wire. Consequently, the alloy has to reach martensitic plastification in the desired shape. However, due to the high elastic limit of NiTinol, it is impossible to perform this operation in the lab. Therefore, the solution was to send the material to a spring factory in which they deform the wire to a spring shape. The next step was to analyze the thermal treatment. This treatment sets the undeformed shape of the muscle. However, if the temperature of the process surpasses the limit, the alloy will be damaged, obtaining inclusions inside as a result. Therefore, the state-of-the- art contemplates less aggressive treatment maintaining more time the temperature. The thermal treatment follows this option. Due to the knowledge extracted, the initial test started with a thermal treatment at 250 ◦ C for 2 hours. Unfortunately, the results didn’t achieve the expected ones. Then, an increment of the temperature in a slope of 50 ◦ C until it reaches 450 ◦ C was tested. However, the material doesn’t have the same properties. The limit found in the literature was 500 ◦ C, so the thermal treatment can not reach this temperature. After the treatment, the material showed two main problems: The elastic limit on the martensite phase was smaller than the one evaluated in the previous section. Also, the color of the alloy was blue. Concerning Titanium alloys, the color is crucial; it indicates the oxidation state. The solution was to perform a hardening process after the thermal treatment. Instead of cooling down the alloy with the ambient temperature, the spring was with water generating a thermal shock. The analysis of the martensite phase provides similar results to the previous muscle. Then the muscle test in the austenite phase provided the same results as the previous alloy. 3.2 Analysing the Forces Involved in the Perching Before starting with the design part, this section is based on a work developed during this thesis and presented to International Conference on Unmanned Aircraft Systems (ICUAS) 2021 [ 73 ]. First, this work presents a theoretical analysis of the forces involved in the perching. Then, the paper presents a simple perching mechanism using magnetic forces. Developing this work is a good starting point for analyzing the perching in ornithopters. The objective of this section is to have a first approach to the forces involved in the perching. Therefore, the first step is to analyze the contour conditions of the perching; Figure 3.8 evaluates the flight stages involved in the maneuver. The figure highlights three phases before the perching: • Normal flight: In this phase, the ornithopter is on a stable flight. The flight velocities are from 3 m/s up to 6 m/s[ 68 ]. The control flight system can control and change the trajectory of the flight. • Approach flight: In this stage, the contact is close, and the platform prepares the contact. The ornithopter velocity is still enough to perform smooth changes in the trajectory. This phase aims to reduce the platform’s speed to the lower control limit, around 3 m/s. 52 Chapter 3. Bio-Inspired Manipulators for Ornithopters Figure 3.8 Flight phases. The figure shows the identified flight stages before the perching. • Contact phase: The contact is imminent in this phase. The intention is to reach the perching spot with the lower velocity possible. However, the influence of the aerodynamic surfaces on the flight during this stage is limited due to the reduced speed. After analyzing the stages of the flight before the perching conditions defining its velocities, the study sets the contour conditions. Next, the study evaluates the worst inputs for a conservative approximation of the maneuver. In this case, the weight considered is 1 kg. This is the usual flying mass of E-Flap. The impact velocity is the lower limit in which the ornithopter still has control flight capabilities. Finally, the perching mechanism considered has no absorptive parts. All these conditions are unrealistic. However, they set a good starting point for the design. The text above evaluates the parameters that cause influence in the problem. Now, the study analyzes the perching system as an impact using the one-degree-of-freedom vibration equation for an undamped system. [ 121 , 122 ] is the base for these calculations. Equation 3.9 represents that equation. uc(t) = (Acosωnt)(3.9) where ωn=pk/m , m is the mass system, k is the stiffness, u is the displacement and A is the initial displacement u0. The Equation 3.9 assumes that the perching system can’t absorb any impact and is attached to the body of the ornithopter. Then, the study sets the collision between the platform and the landing spot. In this case, It assumes an almost perfect elastic collision with a coefficient of restitution of 0.9. This condition means the aerial platform, claws, and perching place could absorb some energy during the impact. However, it is unrealistic to assume that the collision is purely elastic, meaning that the stiffness of the materials is infinite. Finally, the following equations calculate the forces implied in the perching. 3.2 Analysing the Forces Involved in the Perching 53        mo·(v+ o−v− o) = Ic mpp ·(v+ pp −v− pp) = −Ic e=−v+ o−v+ pp v− o−v− pp (3.10) where mo is the mass of the ornithopter and the perching mechanism. v+ o is the velocity of the ornithopter after the impact and v− o before the impact. Ic is the momentum. mpp is the mass of the perching place. In this case, this mass is assumed to be much larger than the mass of the aerial platform. v+ p is the velocity of the perching place after the impact and v− pp before the collision. Finally, eis the restitution coefficient. The study assumes that the mass of the perching spot is much bigger than the mass of the aerial platform. This condition makes the perching spot steady during the perching maneuver. So, the ornithopter and the claws will manage all the energy generated during the impact. The system evaluates the ornithopter’s velocities after and before the collision. The last step is calculating approximately the system’s stiffness to obtain a magnitude order. After defining the design, the materials used set the stiffness of the mechanism. However, the perching mechanism was unknown at this moment. The system was supposed to be a Carbon Fiber (CF) bar. The following equation calculates its stiffness. k=E·A L(3.11) where E is the Young module, A is the area of a fiber carbon tube, and L is the length. Finally, by balancing the system during the collision, the study obtains the forces the mechanism absorbs. Figure 3.9 shows the preliminary pressures the developed system has to support. Figure 3.9 Impact results. The figure shows the force generated in case of perching with an undamped system. 60 Chapter 3. Bio-Inspired Manipulators for Ornithopters between the phalanxes and the joint must be fixed and stable. However, the different elasticity of the material makes it difficult to reach this requirement. At that point, the manufacturing method contemplated generating welding between the phalanx and the joints using the melting point of the plastic. However, it modifies the properties of the joint, making it impossible to analyze what was happening inside. In the second case shown in Figure 3.13b, the structure is continuous of the same material having and structural joint between the phalanx and the joints. The absence of external manufacturing methods that modifies the material’s properties makes the system and the manufacturing more stable. The study of this configuration contemplates using a cantilever approach to model the flexion of the joints. Figure 3.14 shows the parameters of each joint to model the system. The dimensional problem contemplates the variation of these parameters to achieve the design conditions. The dimensional parameters are the joint thickness e, the joint longitude L, and the phalanx thickness l. Each joint has different dimensions configurations to perform the appropriate holding movement as proposed in Figure 3.12. Figure 3.14 Joint parameters representation. The figure represents the last joint dimensioning of the joints. To analyze the dimensional problem of the joints, Figure 3.15 shows the stress study performed on each joint. The procedure is to examine each joint as a cantilever beam. Prepresents the finger’s weight from the joint to the tip of the finger. Trepresents the force the tendons deliver to each joint. This study analyzed the undeformed shape. The objective is to maintain the form while the tendons actuation is not applied and start the movement when the tension is applied. In that way, the study analyzes the joint’s critical flexion. Figure 3.15 highlight two scenarios: • Scenario 1(represented in red in Figure 3.15): In this case of study, the tendons do not apply force to the claw, only with the influence of its weight. Psymbolizes the weight in the diagram. The joint has to maintain its shape during this scenario. • Scenario 2(represented in green in the Figure 3.15): In the second state, the tendon’s tension Tactuates the claw. The objective is to reduce the tension needed to close the claw. In that way, the study analyzes the momentum diagram looking for the critical flexion point. 3.3 Claws Morphology 61 Figure 3.15 Joint Stress Diagram. The figure shows the study performed in the finger joints to resolve the dimensional problem. It highlights two scenarios. The first one with no tension applied and the second one in which the tendon delivers the force to the fingers, represented in red and green, respectively. In the end, it shows the stress diagram in both scenarios. The tendon’s tension is a couple of magnitude orders bigger than the weight of the rest of the claw. Then by looking at the stress analysis in 3.15, the tension is the main force. The application of the tension generates the motion of the fingers. The objective is to reduce the tension needed to close the claw applying, delivering a significant part of it to the grasping. As shown in the figure, the tension of the tendons generates an axial effort. The distance from the center of the joint to the point of application of the tension causes a moment on the joint. To maximize the force applied to the objects grasped, this distance d has to be maximized. However, greater distance means more material and greater weight. The formulation presented in Equations 3.13 analyzes the previous problem. The structure has to support Scenario 1 and collapse in Scenario 2. The following equations analyze the critical point of the shear modulus and the momentum. Specifically, the critical shear strength of the phalanx must be greater than the weight it has to support ( Vcr >P ), and the moment resistance must be less than the force exerted (Mcr <Td). 62 Chapter 3. Bio-Inspired Manipulators for Ornithopters I=1 12bh3=1 12Le3 Wz=Iz yz =1 6e3 Ncr(e,L) = π2EI (Lcr)2=π2E β212 e3 L Vcr(e,l) = Av fy √3 γMo = el E √3 γMo Mcr(e,l) = C1 π2EIz L2sIwL2GIt Izπ2EIz (3.13) where I is the section inertia, L is the distance between the phalanx, e thickness of the TPU joint, l is the thickness of the phalanx, Wz is the section modulus, Ncr is the ultimate axial load limit, Lcr is the length in the bending plane considered, βis the bending coefficient, Vcr is the ultimate shear stress, Av is the cross section area, γMo is the calculation safety factor, E is Young’s modulus, Mcr is the ultimate moment supported by the structure, C1 is the coefficient that regulate the bending conditions, G is the shear modulus of the material where G=E 2(1+ν) and ν is the Poisson’s ratio, It is the torsional constant and Iw is the warping constant. Finally, the analysis shows that the length between phalanges is directly related to bending conditions. Therefore, the structure has to support the shear force V generated by the weight directly related to the cross-section area. In the end, it has to collapse whit the tension of the tendons. The material properties take particular relevance at this moment of the analysis. As explained before, the claw structure is entirely 3D printed. In this case, the phalanx and the joints are a continuous structure made of TPU. The properties of this material are E=26 MPa, G=78.7 MPA, and elongation before break 580%. The theoretical study sets a preliminary result with these properties. However, analytical research selects the final geometry to reach the desired results. In addition, the manufacturing process sets the printing infill to 100%, which means the joint is solid without air inside. This configuration sets the make that the joint properties are the same as the material. It also reduces the complexity of the study of the material. The following text analyzes the behavior of the ways to generate the closing movements. The first one is to use the same dimension in all the joints. Then, the tendon’s insertion points cause a differential actuation to perform the bioinspired closing motion. Nature also inspired this configuration. The difference is that in nature, the bird has a muscle capable of exerting great forces managing each toe. However, in our design, only one actuator drives all the tendons and toes. A phalanx with two joints was analyzed to evaluate the behavior of the differential moment generation method for the closing. The technique aims to generate a more significant moment around the first joint than the second. In that way, it models the insertion points of tendons in the birds’ bones. Figure 3.16 shows this method of actuation. 3.3 Claws Morphology 63 Figure 3.16 Differential moment generation actuation method. The figure represents the differential moment generation actuation method. The pink color represents the position of the tendon inside the toe. The figure also shows the parameters of the actuation method design. The following equation calculates the moment generated around the first and the second joint depending on the angle of the tendon and the insertion point using the parameters shown in Figure 3.16. M1=cos(φ1)·T·D1 M2=cos(φ2)·T·D2(3.14) where M is the moment generated around the joint, φ is the angle of direction of the tensor, T is the tension of the tensor, and D is the distance between the insertion point and the insertion point of the tendon and the center of the joint. Analyzing the equations and the motion’s requirements, M1 has to be bigger than M2 . This means that cos(φ1)·D1>cos(φ2)·D2 . Analyzing this result to take the major advantage of the tension force φ1 has to be as smaller as possible and D1 as bigger as possible. The limit for the angle of φ is the friction generated. If φ1 is very small, the friction between the tendon and the phalanx increases, absorbing much of the force. Regarding the distance between the center of the joint and the insertion point of the tensor, the limit is the distance that makes φ=0 . Figure 3.17 shows several test results to analyze the joints’ characteristics, longitude, and thickness. Figure 3.17 shows the force needed to close the claw using different joint configurations. The longitude Land the joint ethickness have been analyzed. It is important to remember that the system has to provide the desired differential actuation to maintain the best configuration for the holding movements. The experimental results show that the fingers achieved the objective of having a differential actuation between the phalanx. The primary dependence to close the joint is the thickness of the joint, as shown in Equations 3.13. However, the force needed to close the fingers using this actuation is around 1 kg/finger, meaning the muscle had to perform 4 kg to close the claw. Analyzing the system, the major disadvantage of this kind of actuation is the friction generated between the tendons and the walls. A significant part of the force is lost on these walls increasing the tension needed to close the finger. However, the study continues performing additional tests analyzing the strength to open the finger after the holding motion. The objective is to evaluate if the 64 Chapter 3. Bio-Inspired Manipulators for Ornithopters Figure 3.17 Analisis of the joints’ characteristics Configuration 1. The figure analyzes the tendon tension needed to close the fingers depending on the configuration of the joints. The blue color represents the force to close the first phalanx, and the orange the second phalanx. friction that causes difficulties in closing the claw contributes positively, making it more difficult to discharge the mechanism contributing to the holding. In that way, the toe tendons were pre-stressed at 2.5 kg of tension. Then, a dynamometer measures the resultant external force needed to open each phalanx. Figure 3.18 shows the results of these experiments. The results conclude that friction positively affects the force to open the claw. Also, in these results, a dependence of this force on the dimensions of the joints was found, reaching a maximum on the resistance force at 1.5 mm of joint thickness. However, the tendon tension needed to close all the fingers of the claw is too big. This configuration complicates the actuation needing up to 4 kg to close the claw’s fingers entirely. Therefore, this design requirement discards this actuation configuration. The second configuration to actuate the fingers is a longitudinal tendon parallel to the central axis of the finger. This tensor goes as near the palm as possible to generate the maximum possible moment. Then to generate the desired movement, the dimensions of the joint are modified. The idea was to increase resistance from the first joint to the claw’s tip. The resistance of the joints increases with the thickness of its. Figure 3.19 represents this actuation method. This configuration aims to generate the maximum possible moment in all the joints. Then by modifying the resistance of the joint, obtain the motion wanted. According to Equations 3.13, to set the joints’ desired resistance, the parameters are: e(joint thickness), L(joint longitude), and lphalanx thickness. The decision was to vary the thickness, maintaining the dimensions of the rest of the parameters. As shown in Figure 3.19, the thickness of the first joint is smaller than the rest of the joints. Then Equations 3.13 were used to calculate the minimum thickness in Scenario 1 described above (no tension in the tendon). The analysis of the results sets a joint 0.7 mm in thickness and 3 mm in longitude was selected. Finally, the design establishes a first joint of 1 mm to absorb the inertia 3.3 Claws Morphology 65 Figure 3.18 External force needed to open the joints in Configuration 1. The orange color represents the force resultant in kg needed to open the first joint, and in blue, the second joint. Figure 3.19 Finger longitudinal tensor actuation method. The figure represents in pink the position of the tendon inside the finger. It highlights the differential thickness of each joint. generated by the flapping and mass. Then the rest of the joints increased linearly by 0.25 mm for each joint. After the joints between the phalanxes are analyzed, the claw’s structure study starts. As presented before, looking at nature, the idea is to develop an anisodactyl claw structure inspired by eagles and falcons due to their sizes and capabilities. The next step is to create the finger structure. Several approaches evaluate the finger structure to select the size number and the size of the phalanxes. The goal is to develop a claw structure similar to that of eagles, improving the adaptability for grasping objects inspired by parrots. Figure 3.20 shows an image of the structure that inspired the development of the claw. Figure 3.20 shows the bone structure of the eagles’ claws. The Hallux, Digit 1, or back finger, is the principal one generating the force during the grasping. This one, combined with the front central one, provides the major part of the force when holding objects. 66 Chapter 3. Bio-Inspired Manipulators for Ornithopters Figure 3.20 Eagle claw [130]. Morphology that inspired the development of the manipulator. Front-side fingers generate stability in the grip and perch. The size of the fingers changes depending on the different species of eagles. Some species have a smaller, stiffer hallux to generate greater force. The capabilities this toe supports are similar to the ones provided by the thumb finger in human hands. The development of the claw study is focused on the development of two kinds of fingers: the main finger and the equilibrium fingers. The main toe is the one that can exert the greatest forces. This finger supplies the capabilities supported by the Middle talon and hallux. The equilibrium finger provides the lateral balance during the perching and grasping. It mimics the abilities of the Outer and Inner talons. The first approach presented in [ 75 ] has the same size in all the toes of the phalanx and the joints. The variation between fingers was the number of phalanxes; the front central and back fingers were one phalanx bigger than the other. This condition provided good adaptability to objects. However, it was impossible to hold things autonomously. This behavior is due to the absence of the desired closing mechanism. In addition, the actuation of the Middle toe and Hallux did not match the behavior shown in nature. After an exhaustive analysis of the functionalities and shape of the fingers, this study highlights various concepts about birds’ claws summarized on two central ones. The first one is that no one species of bird have the same claw morphology as other species, including species of the same family. Of course, the differences in the claw of different families of species are more significant. However, the size of the foot varies in the same family, depending on where the animals live and the type of diet. For example, claws prepared for hunting bigger prey show sharp nails and smaller stiff hallux talons capable of exerting greater forces. Conversely, if they hunt smaller prey, their nails are usually smaller, which reduces the injury risk. The stiffness of the fingers took particular relevance when it wanted to exert greater force. The first development of the claw reached some limits. One principal limit was the weight the claw could carry out. During this last part of development, one of the main properties found in the feet of birds that want to carry heavier weights is the rigidity of their hallux, which makes this toe capable of withstanding greater forces. 3.3 Claws Morphology 67 Figure 3.21 Middle and Hallux talons evolution. The figure shows a scaled comparative of the different toes developed for the claw’s Hallux and Middle talons positions. The arrow sets the direction of the evolution from the first version to the last version. The legend shows the scale of the figure. The color for the parts of the claw made of PLA is blue, and the gray for the ones made of TPU. The conclusion is that the birds’ claws found in nature have shown an exceptional adaptation to where they live. However, developing a universal claw capable of holding and perching on a wide range of objects and places is more challenging. 3.3.3 Claws Design Evolution The document focuses now on the iterations performed to develop the structure of the claw. After analyzing the nature and several iterations, the decision was to create a claw with two kinds of fingers. One can exert a greater force on the front center and back fingers(Hallux and Middle Talon). The other is more adaptable and appropriately used in the front lateral fingers(Inner and Outer Talons). The front center finger and back finger are the main of our claw, and the lateral fingers provide equilibrium to the grasping. Starting with the development of the main fingers(Hallux and Middle Talon), Figure 3.21 shows the evolution during the development. The following list summarizes the evolution of the main developed robotic toes replicating the Hallux and Middle talons. 68 Chapter 3. Bio-Inspired Manipulators for Ornithopters • Finger I: It was the first developed. It comprised three phalanxes and nails made of PLA. The joints were made of TPU, having a constant thickness of 1 mm. The nail made of PLA replicates the nail of a small bird. The distribution of the tendon tension was longitudinal and parallel to the central axis of the finger. The study of the finger also evaluated to vary the number of phalanxes. The tests concluded that the stiffness of the finger decreases with the number of phalanxes. That increases the adaptability reducing the capability to exert forces with the toe. • Finger II: It was the first finger that tried to develop specific closing movements. It used the differential tendon actuation method presented in Figure 3.16. In this finger, the stiffness increases, increasing the capability to exert force. In this case, the phalanx material is PLA, and the joint material is TPU, both 3D printed. The joints had the same thickness due to the requirements of the differential actuation. The main problem found was adaptability. The stiffness of the finger was high, difficulting their adaptation capabilities. • Finger III: The main innovation of this finger was the development of a continuous structure made of TPU having a structural joint between the joints and phalanx. This finger reduced the stiffness of the whole finger using the flexibility of the TPU in the phalanx. However, using the TPU combined with the differential moment generation system increases the friction between the tensor and the phalanx. Analyzing this scenario, the force needed to close the finger grows, significantly reducing the generation grasping pressure. Another innovation presented on this finger is the linear reduction of the thickness of the phalanx from the beginning up to the nail. That reduces the mass of the finger, making it possible to reduce the thickness of the joints. Another skill presented on this finger is using hard edges in the palm of the finger to contribute to friction generation. This idea tried to emulate Hard Robotics mechanisms, mimicking spikes to go inside the material. Our idea was to have harder edges using more concentration of TPU on it. These soft-hard edges looked for the object’s edges to go inside, generating more friction. • Finger IV:Finger IV is the last finger-developed version. It includes the longitudinal tendon actuation presented in Figure 3.19 and explained before. As shown in Figure 3.21, the system has two phalanxes and three joints made in a continuous structure of TPU. The thickness of the joints increases from the first one up to the nail, performing the desired closing motion. A study performed on the nail sets tendon’s position to generate the greatest moment. Finally, the nails used in this distribution perfectly replicate the shape of the nail of an eagle. The development of the front lateral fingers(Inner and Outer talons) presents a similar evolution to the main fingers. The difference between the main actuation fingers and the lateral ones is their stiffness, which also translates into the number of phalanxes. In addition, the lateral finger is a supporting finger that is more adaptable and used to provide equilibrium to the system during the holding wrapping around the objects. Figure 3.22 summarizes this development. 3.3 Claws Morphology 69 Figure 3.22 Lateral fingers(Inner and Outer talons) evolution. The figure compares the toes developed for the claw’s Inner and Outer talons positions. The arrow sets the direction of the evolution from the first version to the last version. The legend shows the scale of the figure. The color for the parts of the claw made of PLA is blue, and the gray for the ones made of TPU. The following text highlight the main differences between the finger versions. The evolution is similar to the one shown on the central actuation finger. However, the requirements of these fingers are different, varying their geometry. • Finger I: This toe is the first version developed of the lateral fingers. This finger was similar to the first version of the other actuation fingers. An analysis of nature looking for birds inspired claws founds one eagle specie with smaller lateral fingers. The finger developed had TPU joints and PLA phalanxes for these birds. This finger was one phalanx smaller than the main actuation fingers Version I of the Middle talon development. The actuation method used was longitudinal tensor actuation. • Finger II: This is the second version of the lateral fingers. This version has similar concepts to the first one. The finger was smaller than the main one to provide equilibrium to the perching and the grasping. The actuation method applied was the differential tendons actuation. 76 Chapter 3. Bio-Inspired Manipulators for Ornithopters (a) Leg V3 integrated into GRIFFIN’s[ 65 ] platform E-Flap[68]. (b) Leg V3 renderization. (c) Schematic representation of the leg including the main parameters. Figure 3.27 Leg third version.(a) Shows the legs integrated into GRIFFIN’s[ 65 ] platform E-Flap[ 68 ]. (b) Shows the rendering of the system, highlighting the material composition. (c) Shows the principal geometrical parameters of the system. The angle β represented in the Figure 3.27c reduces the legs’ frontal surface, maintaining the distance of actuation to minimize the aerodynamic influence. The legs conduct the energy to the same dumper to better deliver the impact forces using an angle α to guarantee the distance between them. In this version, compared to the previous version, the dumper was made of TPU with a triangle shape to absorb the energy. As a result, this dumper avoids problems of stability found with the spring dumper presented in the previous version. In addition, this 3D-made dumper can easily modify its infill properties by adjusting the dumping ratio to the shock of the platform. Version V4 presented in figure 3.28 was the most complex leg. These legs combine all the concepts presented in the previous version and resolve most of them. The following list enumerates the concepts for the development of the legs. • Impact absorbance: The mechanism to contribute to the perching passively and to absorb the impact energy are directly related. The idea is to use the energy absorbed from the impact to contribute to the perching transferring this force to the claw. The claws absorb the impact by using tendons and the scissor mechanism. 3.4 Leg Morphology 77 (a) Leg V4 integrated into the aerial platform.. (b) Leg V4 renderization. (c) Schematic representation of the leg including the main parameters. Figure 3.28 Leg fourth version.(a) Shows the legs integrated into GRIFFIN’s[ 65 ] platform E-Flap[ 68 ]. (b) Shows the rendering of the system, highlighting the material composition. (c) Shows the principal geometrical parameters of the system. • Passive perching contribution: The passive damping mechanism developed within this version contributes to the perching. This mechanism uses the impact to contribute to the perching. In addition, this bird-based mechanism uses the tendons inside the legs to contribute to the perching. After the shock, this mechanism takes advantage of the mass of the ornithopter to maintain the perching. Then, the active actuation combined with this one generates enough actuation to keep the perching. • Low aerodynamic influence: As shown in Version V2, the aerodynamical influence of the claw is important. This development implements two ways to reduce aerodynamic influence. The first one is to use the same actuation configuration presented in V2.V4 solves the problems found in V2 with this mechanism by using Titanium screws, a bearing, and C.F. layer-composed pulleys. The second one is related to the position of the claw during the flight. The objective is to fly with the claw backward, reducing the aerodynamic perturbances. Before the perching, the hip joint moves the claw forward, oriented to the perching spot. In nature, birds fly with their claws back and then change them forward for perching. • Possitioning control: The versions presented before do not have the option to control the position of the claw. It is impossible to reach equilibrium in the perching by 78 Chapter 3. Bio-Inspired Manipulators for Ornithopters balancing the claws. This version uses a hip joint actuated by SMA muscles to control the position o the claw reducing the torque needed to keep the perching and manage its position during the flight. The list presented before enumerates the main concepts achieved in developing Version 4of the leg. The text continues focusing on this development and the way to achieve these results. As shown in the figure, this leg is one of the most complex in development. It is only possible to understand this development by looking at the previous ones. CF is this leg’s main component, increasing the fabrication process’s complexity. This development shows the integration of two central systems for the first time: the antagonist muscles configuration and the system that absorbs the impact contributing to the perching. The passive perching/impact absorbance and leg positioning control systems are presented first. Figure 3.29 shows this system. Figure 3.29 Leg dumping system representation. The figure shows in yellow the tendon position inside the legs and highlights the main geometrical parameters. The system comprises eight transmitters in the shape of a scissors mechanism. Then, a prismatic joint of one degree of freedom connects the leg and the hip joint. There are two points of view concerning passive perching, the mechanism, and the tendons. Focusing on the mechanism: Initially, the claws receive the impact shock. Then, the legs transmit this force to the mechanism. In the end, the prismatic part converts this force to one direction. It closes the scissor mechanism vertically and expands it horizontally. The tendons absorb the horizontal expansion. They connect the claw and the scissor mechanism. The force generated in the lateral expansion is transmitted through the tendons to the claws. The fact that the force needed to close the claws increases proportionally with the closing motion makes it possible to generate a dumper. In the end, the claws themselves are our system’s dumper. The system design generates as much force as possible to contribute with the SMA actuators to the holding. Therefore, the mechanism must cause tension to close the claw during impact. The geometry of the scissor mechanism models the quantity of force 3.4 Leg Morphology 79 delivered to the claw. Making a balance and a graphical analysis, the equations to obtain the longitude of the transmitters are the following: 4T=Fimpact ·cosα △Ltendon =2·L(cos(αmin)−cos(αmax)) (3.15) where T is the tension of the tendon. Fimpact is the force of the impact, α is the parameter that defines the position of the scissor mechanism, △Ltendon is the increment of the tendon longitude, L is the longitude of the transmitter of the tendon mechanism, αmin is the parameter that defines the final position of the mechanism, and αmax is the parameter that defines the initial position of the mechanism. Fimpact is obtained theoretically in Section 3.2 to evaluate these equations. △Ltendon is two centimeters, the longitude necessary to close the claw. αmin is set to 15 ◦ due to the geometrical limit. Finally, by resolving the system, the longitude obtained of the transmitter is L=46.2 mm. This system replicates the behavior of birds’ legs while perching. However, this solution is adapted to the perching contour conditions of the ornithopters. Two muscles in the antagonist configuration manage the positioning system of the legs. A pulley placed on the hip handles the tension generated by the SMA, increasing the moment generated to control the position. Figure 3.30 synthesizes the hip joint positioning control system. Figure 3.30 Claw positioning control. The figure represents the main geometrical parameters of the system. It also describes the way to connect the antagonist muscles configuration to the joint using a yellow line to highlight the position of the tendons. The system has to maintain the leg back during the flight, reducing the system’s aerodynamic influence, and then forward to prepare the perching. During the perching, this system must maintain the ornithopter’s position. SMA 1 actuator moves the claw forward, and SMA 2 moves it backward. Section 3.5 presents a control system to close the loop of the position with the temperature of the SMAs. 80 Chapter 3. Bio-Inspired Manipulators for Ornithopters By applying Euler-Lagrange, the dynamics of the hip joint are analyzed. (Izz +MeqR2 CoG)¨ θ(t)−MeqRCoGgsinθ(t) = ∑FSMA(t)r−c˙ θ(t),(3.16) where Meq(kg) is the equivalent mass of the system in the CoG, Izz(kg.m2) is the moment of inertia of the leg, R(m) is the distance from the rotational point to the CoG, g=9.81m/s2 is the gravity constant, θ(t)(rad) is the generalized coordinate of the system, ∑F(N) is the resultant of the forces of the SMA spring in antagonist configuration, r(m) is the internal radius of the pulley and c(kg.m/s)is the dynamic friction coefficient. The text focuses now on the manufacturing method and final prototype. The leg is entirely made of CF plates. It is composed of 30 different parts to reach the desired specifications. First, the part has been designed layer by layer with a CAD program. Then, a Computer Numerical Control (CNC) cuts the layer from CF plates of 1.5 mm or 2mm in thickness, depending on the component. Lastly, the parts are glued together using Epoxy resin and a vacuum system. Figure 3.31 details the integration process. (a) Layers cut with a CNC. (b) Layer’s mechanical integration. Figure 3.31 Leg Layer integration process. (a) shows the layers cut from the CNC machine. (b) shows the mechanical integration of the layers glued with Epoxy and using CF bars to generate a mechanical union between its. Figure 3.31 details the layer integration process. First, the layers were glued using Epoxy; screws were used to guarantee the position during drying. After that, the screws are substituted by CF bars to make a permanent mechanical joint between layers. In the end, everything is sanded, providing the finishing touch. All the concepts learned in Leg V4 have been applied to a two-legged version of this development shown in Figure 3.32. The figure shows the integration of the mechanism in a two-leg system. Figure 3.32a shows the leg position oriented to the perching spot. Figure 3.32b shows the legs’ role during the flight. The inclusion of two legs simplified the lateral equilibrium. Their position backward reduces the aerodynamic influence of the manipulation system as birds do in nature. Finally, Figure 3.32c shows the integration into the GRIFFIN platform. 3.4 Leg Morphology 81 (a) Schematic representation of the legs’ position before perching. (b) Schematic representation of the legs’ position during the flight. (c) Leg’s integration into E-Flap[68]. Figure 3.32 Two legs integration. (a) and (b) shows a schematic representation of the system in the positions for perching and during the flight, respectively. (c) displays the system integrated into E-Flap[68] . Figure 3.33 shows a dynamic analysis to evaluate if the system can support the perching shock. This analysis used the Stress Analysis environment of Inventor ® software. The idea is to simulate the forces involved in the perching, analyzing the weak points of the structure. Two forces of 125 N were applied in the center of the palm of each claw to emulate the perching stresses. The analysis considers the scissor mechanism stuck in the position in which it transmits all the tendon tension to the claw. The stress results conclude that all the deformations made during the impact are elastics. We found some weak points in the scissor mechanism’s transmitters, but they can still support the stress generated. Finally, 20% of tolerance was considered to prevent fatigue damage in the different parts of the mechanism. The system developed can maintain the geometry of the legs during the perching contributing to it. The force generated by the tendons taking advantage of the energy of the perching is enough to close the claw. It also contributes to the system’s reaction time, helping the SMAs’ warming time. This system makes it possible to use the claws in ornithopters reducing the weight and increasing the operational area. It is only understandable to have one system with the other. 82 Chapter 3. Bio-Inspired Manipulators for Ornithopters Figure 3.33 Legs stress analysis. The figure represents the legs mapped in colors analyzing the stress in each part of the system. The color coding of the map is on the left of the figure. The yellow arrows represent the force location and direction. This section shows the evolution of the leg concept during this work. The legs were taking more relevance and complexity during the development. Ultimately, leg and claw work together, keeping the bioinspiration. This result also highlights the personal evolution of the researcher during this period. The following section will focus on the actuation method implemented on the ornithopter. The novelty method presented will overcome the payload limitations. 3.5 Actuation Method and Integration This chapter describes integrating the different actuation methods developed during this thesis. Therefore, It discusses the physical aspect of integrating the system’s electronics. At the end of this chapter, the whole manipulation system is ready to work together. 3.5.1 SMAs Actuation Configuration This section focuses on the solutions to integrate the SMAs into the manipulation system presented before. It resolves two main requirements: the position and configuration of the SMAs to actuate the claw and the electronic developed to control the muscles. The section starts analyzing the position and the configuration of the actuation. As shown in the legs designs of Section 3.4, two methods exist to place the muscles in the leg, both based on the antagonist muscle configuration. These methods use two SMA muscles. One is used to apply the force directly to the tendon; the other muscle is opposed to this, reducing the tension applied to the tendon. Figure 3.34 presents two methods to place the SMA using this actuation method. Figure 3.34a presents the linear position of the antagonist muscle actuation method. The active muscle is directly connected to the other using a tendon. The place of both 3.5 Actuation Method and Integration 83 (a) Linear muscles antagonist configuration. (b) Parallel muscles antagonist configuration. Figure 3.34 Antagonist muscles positioning method. (a) Shows a schematization of the positioning method in which both muscles are in the same face of the leg. (b) Presents the configuration that reduces the space needed to place the SMAs in opposites faces of the leg.. actuators is the frontal part of the leg. The main advantage of this actuation method is its reduced complexity. Therefore, no elements between the muscles reduce the possible losses in the actuation. The main disadvantage of this actuation configuration is the length needed to actuate, making the leg larger. As presented in the previous section, a larger limb causes the claw to need to deliver more torque to compensate for the mass in balance during the perch. Figure 3.34b presents a configuration that reduces the length needed for the actuation. In this solution, one muscle is in front of the other. The system transmits the tension from one to the other using a tendon to connect both muscles and a pulley to reduce the tension loss at the edges. The pulley works by reducing friction with the axle. This solution divides the length needed to place the muscles by two. This configuration makes it possible to reduce the size contributing to the manipulation system’s perching, mass reduction, and aerodynamic influence. The configuration used in the end was the second one. This configuration reduces its negative aspects by reducing the friction in the transmission. The comparison did not find a negative part of this one compared to the other. 3.5.2 Tendon Driven Actuation This section focuses on how to transmit the force from the SMA springs to the fingers of the claw. The following test analyzed the material and the integration of the tendons in the manipulation system. 84 Chapter 3. Bio-Inspired Manipulators for Ornithopters Tendons connect the claw and muscles. They have special requirements due to the operating conditions of the SMA. They must withstand high temperatures up to 145°C combined with high forces. Some traditional materials, such as nylon, show elastic deformations due to the applied stress. This deformation absorbs a part of the energy reducing the actuation. In applications such as fishing, this property reduces the impact force received by the fisherman. The other problem with connecting nylon directly to SMAs is that its melting temperature is lower than the actuation temperature. The melting temperature makes it impossible to use nylon for this application. However, nylon has excellent friction properties due to its polished surface. The text above highlights that nylon is not a possible solution to be connected directly to the muscles. Evaluating the possible materials with high-stress resistance and temperature tolerance, Kevlar appears as potential material. Regarding its applications, Kevlar was combined with CF in some fabrics to avoid scratches. Traditionally Kevlar was used in the military industry to make bulletproof vests. Looking at the second application implies temperature resistance and high-stress resistance. Analyzing the market, we select Kevlar in threads and cords. Table 3.2 compares both materials concerning the properties wanted to evaluate. Table 3.2 Tendon material comparison. The table compares Nylon and Kevlar for their use as tendons. Material Tensile strength (MPa) Elongation before break (%) Supported temperature (◦C) Nylon [136] up to 607 up to 200 up to 120 Kevlar [137] up to 3450 3-4 up to 425 The table focuses on the properties of the material. The analysis focused on three commercial products, nylon fishing lines, kevlar threads, and kevlar micro cords. Lowtemperature resistance and elongation before break eliminated nylon fishing lines from the selection. Concerning Kevlar, its high friction with the claw’s plastic parts makes it difficult to use in our design. In addition, the friction and the presence of only longitudinal fibers in the thread sometimes cause breaks during experiments. Therefore, Kevlar micro cords were analyzed. The main difference between the Kevlar thread and Kevlar micro cords is the presence of fibers in all directions. That generates a more homogeneous surface reducing the friction coefficient between the tendon and the parts of the claw. However, the friction coefficient is still high for use inside the fingers. So, the final design uses two materials. Nylon in the system’s part in which the tension and temperature requirements are relaxed, and Kevlar micro cord in the other places. [ 138 ] shows the Kevlar micro cords used with a working temperature of up to 400 ◦ C and a tensile Strenght of 50 kg. [ 139 ] shows the properties of the nylon used capable of resisting up to 14.6 kg of longitudinal tension. Another critical topic in this section is how to deliver the force to the fingers. In the first version of the claw presented in [ 75 ], the system has a manually adjusted tensor mechanism that manages a fixed elongation of each tendon. Then, before the object experimentation, 3.5 Actuation Method and Integration 85 (a) Configuration 1. (b) Configuration 2. Figure 3.35 Tendon force distribution. The blue lines represent the tendons, the blue circles the sliding nodes, and the text highlight the position of the muscles and toes. we adapt the dimensions of each tendon. In this version of the claw system, we develop a differential tendon system that autonomously manages the force delivered to each finger. One of the main advantages of this system is its capability to adapt to a wide range of objects. In contrast, the previous system has manually adapted to each object’s shape. This system can update the tendon elongation during the holding and adjust the claw to the object. The system is based on balancing the tension of the fingers’ tendons. Each finger has a tendon, and they are interconnected using this system. The idea is to make an unstable equilibrium. For example, a finger stuck transmits its tendon tension to the other interconnected fingers. That makes it possible to continue wrapping around objects with heterogeneous shapes. This section presents the study of two tendon configurations. In the first one, based on nature, the front central finger and the back finger have more force than the rest of the finger. Therefore, the front middle talon doubles the pressure exerted by the front lateral fingers. On the other hand, the back finger has the same force as the three front fingers. In the second configuration, the four fingers can exert the same force. Figure 3.35 summarises these concepts. Equations 3.17 and Equations 3.18 evaluate the system’s behavior depending on the tendon distribution set. Configuration 1: FCentral =2·FLateral FFront =FCentral +2·FLateral FBack =FFront FSMA =FFront +FBack (3.17) 92 Chapter 4. Bio-Inspired Morphing Surfaces morphing actuation methods like piezoelectrics are studied. Their way of actuation is to deform the host surface generating a controlled deformation on it. On the contrary, servos and other traditional methods only transmit a punctual force. Birds’ are our reference to take advantage of years of evolution. The objective of the wings is to increase the control inputs by controlling the wing’s surface. At the same time, the materials used are softer, mimicking the birds’ wings. So the idea is to vary asymmetrically the surface of the wings generating actuation on the ROLL angle of the aerial robot. The cover is made of elastomer materials to improve the human cooperation capabilities of the aerial platform. Studying a bird-based wing profile enhances the gliding capabilities and maintains the flapping behavior. This chapter is structured as follows. Section 4.1 is focalized on the development of the morphing tail studying the actuation method, morphology, and ways to control it onboard. Section 4.2 focuses on the development of the wings. Mainly, this section focuses on constructing it to support the forces generated by the flapping. Finally, Chapter 5 presents the experimental validation of the system. 4.1 Morphing Tail This section focuses on the development of the tail. The objective is to develop a bird-based tail proper for ornithopters. Traditionally, as shown in [ 140 ], they use plain rigid surfaces actuated by servos. However, birds’ tails in nature are composed of feathers moving in all space directions. Furthermore, the actuation is not linear. Instead, the actuation describes an exponential function ending in the tip. Figure 4.2 shows a natural bird tail to explain the concepts. Figure 4.2 Peregrine Falcon [141]. The figure shows the size and shape of the bird’s tail. Concretely, its focuses on the peregrine falcon. The falcon is performing an approach maneuver to the target. The V-Tail is widely open, increasing the control action of the tail. The surface of the tail open has the same magnitude order as the rest of the body. This actuation is impossible to perform with traditional actuators. The weight and inability to generate deformations in an area discharge servos form the group of possible actuators. Instead of servos Smart Materials like Macro Fiber 4.1 Morphing Tail 93 Composite ™ (MFC) show better capabilities to perform these tasks. This material can perform deformation in the whole surface of the actuator. Also, they are capable of exerting high blocking forces with reduced weight. However, before this work, they were used as sensors to measure small deformations. Usually, they are glued to other surfaces to generate deformation. This work studies a new way to use MFC to develop the tail’s structure, mimicking birds’ feathers. The tail will be able to replicate the movement and the actuation of a natural bird tail. Additionally, to increase the actuation, the control surface is augmented. The combination of the bird-inspired motion and the major surface makes it possible to increase the control capabilities of the aerial platform. Ultimately, it will increase the control capabilities making it possible to load more weight. In addition, other applications like manipulation will take advantage of weight reduction. The section is structured as follows. First, it starts with studying the actuators used in the tail. Then, it continues with the development. In the end, Chapter 5 presents the experimental validation. 4.1.1 MFC NASA developed MFC in the early ’90s. Initially, they create MFC to work as sensors to measure small deformations due to the sensibility of their internal resistor to the displacement. Then, researchers discovered the high locking force exerted by these under certain conditions. The capabilities of MFC to perform actuation in an area makes them a good candidate for morphing applications. Figure 4.3 represents its shape and flexibility. (a) MFC. (b) MFC flexibility [142]. Figure 4.3 MFC actuator. (a) Shows the geometry of the MFC. (b) Shows the flexibility and thickness of the actuator. MFCs are light weighted and thin actuators that cannot perform displacements. The displacement’s capabilities depend mainly on the material of the host lamina. Therefore, the structure of MFC layers is crucial to understand its behavior. They are a piezoelectric material sandwiched with electrodes capable of generating a specific electric field to modify the shape of the piezoelectric. The rest of the layers guarantee the isolation of the electric field. Figure 4.4 shows an example of the layer structure of the MFC. Specifically, the figure shows the ones used in this work. 94 Chapter 4. Bio-Inspired Morphing Surfaces Figure 4.4 MFC layer structure. The figure shows the layer structure. The arrows around the figure highlight the material of each layer. The layers are symmetric from the Piezoelectric material layer . The layer in charge of the field generation is the layer of copper electrodes. In this case, this layered design set the actuation to work between 0V and 1500V. The isolation layers are made of acrylic and Kapton to avoid short circuits. The piezoelectric material in the central layer expands depending on the field generated. The deformation performed by the electric field in the piezoelectric generates the deformation. The piezoelectric material is a foam-based material that produces an expansion with the electrical field. The layer in charge of the field generation is the electrode copper layer. The actuator itself is capable of generating minimal deformations in the longitudinal direction. However, the force delivered in this direction is high. Therefore, the objective of the tail is to convert its deformation to the orthogonal direction. Different researchers evaluate the work of these actuators. Due to its isolation capabilities, underwater applications like [ 56 , 57 , 58 , 59 , 62 ] use this actuator. Then after these approaches, other researchers studied the use of MFC in fixed wings in works like [ 105 , 106 , 107 ]. The work developed in this thesis is a combination of both works. The main difference between underwater works and the work developed in this thesis are the environment and the locomotion generation. In this thesis, the domain is the air, which isn’t wanted to generate locomotion. However, the displacement needed to control the trajectory of the ornithopter is significant as the one performed on underwater applications. The density of the water is also bigger than the air, making it necessary to generate stronger movements. Ultimately, to overcome the contour conditions of the platform, the tail developed in this thesis wants to create fast, long, and strong motions as the works of underwater applications. Concerning the works in fixed wings, they use these actuators to modify the wing’s surface. These publications show two kinds of results. The first uses the MFC actuators to alter the camber line of the wing profile, and the second employs these actuators as flaps. The objective is to add more input to control the trajectory. The displacements wanted on these publications are smaller than those wished to perform. 4.1 Morphing Tail 95 Additionally, they use hard surfaces to glue the actuators. Usually, they utilize metal sheets and Carbon Fiber (CF) sheets. These surfaces make it possible to generate a higher blocking force losing in the way most of the displacement. Using these actuators is desired to develop an entire tail that mimics the behavior of the bird’s tail. The structure is the MFC combined with other materials to generate enough actuation mimicking the birds’ feathers. As birds’ tails are soft and adaptable, the one developed is soft and capable of increasing safety during the interaction. Developing a soft tail made of MFC for ornithopters is not a direct integration. Different design requirements define the way to integrate these actuators into the ornithopter. • Soft integration: Usually, researchers employ hard materials to glue the actuator blocking the expansion of the actuator, reaching the displacement a reduced displacement in the orthogonal direction. Contrary to this, high displacement is wanted. Using these actuators is wanted to replicate the feathers of the tail of a bird. So the expansion has to be blocked by using other materials that maintain the soft capabilities of the system. The use of more flexible materials enables the generation of bigger deformations. Ultimately, the actuator must combine orthogonal flexibility and longitudinal stiffness. • Flapping frequency: The flapping frequency of the ornithopters goes through the entire body of the ornithopter generating a high inertia even on stiff material. The actuator’s stiffness is crucial to reduce the perturbances on the tail. • Weight: Weight is one of the main limitations of ornithopters. Compared with other systems on the ornithopter, the weight of the tail is more critical even due to its distance to the CoG. It could generate a passive moment that makes the platform unstable. The objective is to replace the two servos and the tail’s structure for an entire morphing surface made of MFC, reducing that weight and generating a bird-inspired actuation. • Bimorph actuation: The tail has to act in the actuator’s orthogonal directions. MFC can only actuate in one way, so the idea is to combine it to have actuation in all the ways. The soft MFC’s integration aims to block the expansion in the longitudinal direction. The actuator stops the growth by using CF threads and epoxy. This approach also keeps the orthogonal flexibility. The following section theoretically models the behavior of the MFC to work as soft morphing actuators. Modeling MFC To analyze the movement that MFC can develop, the Rayleigh-Ritz method was applied, following the developments performed in [143]. As was highlighted before, the MFCs are composed of layers of different materials. Therefore, analyzing the layers and host lamina is crucial to obtain the properties of the actuator. In that way, works like [ 143 ] use the Classical Lamination Theory (CLT) to estimate and compare the properties of the materials with those given by the manufacturer. Then, our work uses the properties estimated in this work. 96 Chapter 4. Bio-Inspired Morphing Surfaces However, it is necessary to understand the mechanical properties of the MFCs to analyze the actuation. Mechanically, the MFCs are composed of two layers of isotropic materials and another two of orthotropic materials. Isotropic materials have the same composition in all directions and orientations, so their mechanical properties are the exact in-plane directions. So by definition, Kapton and epoxy layers are isotropic layers. Orthotropic materials have different properties depending on the orientation. In the case of MFCs, the cooper electrodes layer has a direction of 90 ◦ , and the piezoelectric material has a frontage of 0 ◦ . Figure 4.3b is an excellent example to highlight these concepts. The figure highlights the orthotropic properties of the whole actuator. It is shown the longitudinal flexibility compared with the orthogonal stiffness. Before starting the calculation, it is necessary to define the reference system(see Figure 4.5). In this thesis, the host lamina and the MFC have the exact longitude in the XY plane. However, the thickness of the MFC layer and the host lamina differ. Therefore, Lx and Ly define the longitude in the XY plane, and HMFC and HHOST determine the thickness. The figure also represents the MFC geometry in the reference system. Figure 4.5 Actuator geometry parameters. The figure represents the geometry parameters for calculating the properties of the MFC. It shows the axis of the calculation defined in black arrows with the name of each axis. The dark yellow represents the MFC, and the light yellow is the actuator’s host lamina. Following the approximation developed by Schultz in [ 144 , 145 ], the midplane’s fourterm displacement equation must be assumed to apply the Rayleigh-Ritz method. The four unknown coefficients c1,c2,c3,c4define the equation. 4.1 Morphing Tail 97 u0(x,y) = c3x−c2 1x3 6−c1c2xy2 4 v0(x,y) = c4y−c2 2y3 6−c1c2x2y 4 w0(x,y) = 1 2(c1x2+c2y2) (4.1) where u,v,w are the displacements in the midplane in the axis X,Y,Z respectively, c1c2c3c4 are the unknown coefficients that have to be calculated and x,y is the position of the point in which the displacement is calculated. The equation of the displacement is defined. Then, the Rayleigh-Ritz method evaluates the potential energy of the complete system. There are two situations to calculate the system’s potential energy, the non-actuated and the actuated state. The objective is to determine the potential energy of the MFC in terms of the unknown coefficients c1,c2,c3,c4 . Finally, by analyzing the values of the coefficients that minimize the total potential energy, c1,c2,c3,c4 could be obtained. After all, the displacement is evaluated by plugin these coefficients into Equation 4.1. The development starts calculating the potential energy in the host lamina. The host lamina has the same properties in both situations. So the equation that evaluates the potential energy is the same in both states. However, the coefficients change between the two states. Therefore, the following equation calculates the potential energy of the host lamina: UHOST =1 2ZLx 2 −Lx 2ZLy 2 −Ly 2ZHHOST 2 −HHOST 2 [σHOST,xεHOST,x+σHOST,yεHOST,y +σHOST,xyγHOST,xy]dzdydx (4.2) where σ are the stresses in the directions of the plane, ε and γ are the strains in the different directions of the plane. The nonlinear strain-displacement relationship for minor strains and rotations in terms of the displacements sets the strain in the mid-plane. ε0 x=∂u0 ∂x+1 2∂w0 ∂x2 ε0 y=∂v0 ∂y+1 2∂w0 ∂y2 γ0 xy =∂u0 ∂y+∂v0 ∂x+∂w0 ∂x∂w0 ∂y (4.3) 98 Chapter 4. Bio-Inspired Morphing Surfaces The following equation defines the curvature in the midplane of the layer to obtain the strain in all the sections: k0 x=−∂2w0 ∂2x k0 y=−∂2w0 ∂2y k0 xy =−2∂2w0 ∂x∂y (4.4) The coming equation defines the strains in the volume by applying the Kirchoff hypothesis:    εx εy γxy   =    ε0 x ε0 y γ0 xy     +Z    k0 x k0 y k0 xy     (4.5) c1,c2,c3,c4 evaluates the strain by substituting the four-term displacement equations in all the formulas. Then, the reduced stiffness matrix Qlinks the strain and stress. σ=Q·ε(4.6) where σ are the stresses in all directions, Q is the reduced stiffness matrix of the system and εare the strains. Regarding the materials used as host lamina, the following equation defines Q: Q=    E1 1−ν12ν21 ν12E1 1−ν12ν21 0 ν12E1 1−ν12ν21 E2 1−ν12ν21 0 0 0 G12     (4.7) where E is Young’s modulus of the material, ν is Poisson’s ratio, G is the Shear modulus. Finally, the unknown coefficients c1,c2,c3,c4 determine the potential energy equation of the host lamina. Concerning the MFC, there are two states: the non-actuated state with no voltage applied and the actuated state when the voltage is applied. In the non-actuated state, the behavior of the MFCs is similar to the host material. However, there are different considerations due to its layer structure. The following equation expresses the potential energy of the MFCs: 4.1 Morphing Tail 99 UMFC =1 2ZLx 2 −Lx 2ZLy 2 −Ly 2ZHHOST 2+HMFC HHOST 2 [σMFC,xεMFC,x+σMFC,yεMFC,y +σMFC,xyγMFC,xy]dzdydx (4.8) The relationship between the curvature and the strains follows the assumptions made before in the host lamina. The formulation assumes that the curvature of the midplane of the MFC and the host is the same. So the strain in all the surfaces of the MFC could be expressed as follows:    εx εy γxy   =Z−HMFC +HHOST 2    k0 x k0 y k0 xy     (4.9) Finally, the potential energy of the MFC could be in the non-actuated state. The following equation defines the system’s potential energy in the non-actuated form. UTOTAL =UHOST(c1,c2,c3,c4)+UMFC(c1,c2,c3,c4)(4.10) After all, as exposed in the Rayleigh-Ritz method, the coefficients are set by minimizing the potential energy function: ∂UTOTAL ∂ci =0i=1,2,3,4 (4.11) At this point, the formulations determine the system under non-actuated conditions. All the coefficients are equal to zero, highlighting that the system has no displacement. The result in the coefficients is an obvious solution. However, it is a good starting point to continue analyzing the system. At this point, the formulation evaluates the displacements by applying a voltage. Therefore, the potential energy on the host lamina does not vary between the actuated and non-actuated states. However, due to the voltage applied, the MFC changes its potential energy. The following equation defines it during the actuation. UMFC,Actuated =1 2ZLx 2 −Lx 2ZLy 2 −Ly 2ZHHOST 2+HMFC HHOST 2 [(σMFC,actuated,x)εMFC,x +(σMFC,actuated,y)εMFC,y+(σMFC,actuated,xy)γMFC,xy)]dxdydz (4.12) where σMFC,actuated are the stresses in the MFC without the stresses induced by the piezoelectric material. It could be expressed as: 100 Chapter 4. Bio-Inspired Morphing Surfaces θMFC,actuated =θMFC −θPiezoelectric (4.13) where σpiezoelectric is: σpiezoelectric =QMFC △V △x   d11 d12 0  (4.14) Finally, c1,c2,c3,c4 solve the potential energy equation, and the Rayleigh-Ritz method determines these coefficients. ∂UTOTAL ∂ci =0i=1,2,3,4 (4.15) The equation 4.1 determines the approximate displacement equation by replacing the coefficients. Figure 4.6 shows the displacement results of the theoretical model. Figure 4.6 Displacements theoretical analysis. The figure shows the displacements in the Z axis of the MFC mapped in color on its geometry. The scale of the colors and the value that corresponds to each color is on the figure’s right side. The figure evaluates the displacements of the actuator developed. The host lamina of the model is another MFC (MFC-2). The actuation on the graphic is asymmetrical, meaning that when one MFC is activated and the other is not. Therefore, if MFC-2 is deactivated, MFC-1 performs the actuation. On the other hand, when MFC-2 is activated, 4.1 Morphing Tail 101 the displacement of the actuator will be in the opposite direction. The model is an impressive starting point to evaluate the technology developed for the tail. However, the construction and experimentation of the system are crucial because the model is only a theoretical approximation. The following section focuses on developing the bio-inspired tail using these actuators. 4.1.2 Tail development The aim is to replicate the behavior of a bird’s tail using the actuators presented before. The following list summarizes the main concepts of tail development, highlighting the objectives. • Soft actuation: In nature, the birds’ tails are composed of feathers. Their tail is adaptable to most conditions. In a crash, their soft structure avoids damage to the body and the environment. The objective is to replicate this capability by developing a soft tail using the MFCs as a structure. • Turning velocity: Looking at the birds’ flight, they exhibit an exceptional turning velocity of the tail. MFCs are one of the faster-known actuators. This velocity increases the control skills of the platform. This situation is analyzed using a Motion Capture system in the experimental part of this development. • Weight distribution: The central part of the birds’ weight is around the CoG. Regarding ornithopters, a usual problem is the weight of the servos to control the tail. The size of the servos needed to maintain it is significant due to the high aerodynamic forces generated. On the other hand, MFCs exhibit high blocking capabilities with a much lower weight. However, the electronics to control MFC are more complex. This part is tiny compared to traditional servos, and the location is close to the CoG of the bird, reducing the influence on the trajectory. Traditionally ornithopters’ tails use servos as actuators. Table 4.1 compares servos and MFC, focusing on the concepts and requirements previously highlighted. Table 4.1 Comparison Servomotors and MFC Actuators. Parameters Servomotor MFC Dimensions (mm) 22.8x12x29.4 101x20x0.5 Weight (g) 15.8 2 Blocking Force (N) 58.5 152 Power consumption (W) 2 ∼0 Energy Efficiency 85 100 The table shows servos are heavier than MFCs. Their blocking force generated is lower than MFCs. Servos perform only punctual actuation, and their isolation from the environment is lower. The power consumption of servos is also bigger. The blocking force generated by MFCs improves the actuation of the tail, increasing the control on the trajectory. The actuation in the surface of the MFCs make it possible to develop morphing 108 Chapter 4. Bio-Inspired Morphing Surfaces Table 4.2 Tails comparison: The table shows the comparison between two V-tails and the prototype developed in this thesis. Parameters Commercial V-Tail [133] V-Tail [140] V-Tail developed Actuation method Servos Servos MFCs Actuator weight(g) 15,8 9 2 Number of actuators 2 2 10 Actuator blocking force (N) 58.5 38.5 152 Active surface (cm2)462 740 1190 Total weight (g) 67.1 108 51.8 Active surface/weight Ratio 6.89 6.85 22.97 Speed Medium Medium High as the less-weighted servo system. The blocking force is four times bigger than this system and three times bigger than the other. It made it possible to increase the surface of the actuator. In the end, the isolation properties of the MFCs made it possible to work in more complex environments. Servos show isolation problems in dusty environments. Regarding our experience with these actuators, the dust comes inside the gears damaging them. The impact resistance of servos is also smaller than MFCs. The fact that MFCs does not have mechanical connection reduces the damages generated in case of a crash. Focusing on the whole tail, the actuator developed is less weighted than the others. This difference gains prominence by locking at the active surface. The active surface is two times bigger than the rest of the tails, generating more than two times actuation due to the bio-inspired actuation. The velocity to change the tail’s position is more significant than the rest of the systems. This transition velocity is due to the capabilities of the MFCs to shift the force generated in terms of voltage. The resistance to the impact of the tail is improving the testing capabilities shown in the system presented in the state of the art. The experimental validation of the tail will be presented in Chapter 5 together with the other system highlighting the interrelationship between them. 4.2 Morphing Wing Traditionally, the wings of ornithopters use plain stiff, aerodynamic surfaces. These profiles are one of the most simple in engineering. Their main benefit is the reduced weight of the wing. The absence of free volume makes it possible to avoid filling materials. However, the capabilities of these profiles could be higher. They are not able to have lift generation 4.2 Morphing Wing 109 themselves. Instead, the flapping and the surface of the wing generate the lift. In other words, the ornithopter has low gliding capabilities. However, the gliding phase of the flight is crucial. It makes it possible to reduce energy consumption and increase fault tolerance. During the gliding phase, the energy consumption of the ornithopter is near zero, and the flight is stable. The platform’s response to the control inputs significantly reduces the actuation needed. The second concern focuses on possible flight faults. Increasing the gliding capabilities of the ornithopter makes the gliding velocity smaller. As a result, the ornithopter can have controlled flights at lower speeds. The ornithopter can smoothly land in the case where the flapping mechanism breaks suddenly, avoiding risky situations. Concerning the perching, using this kind of profile also help the perching task by reducing the energy of the impact. To improve the control of the trajectory, birds have actuators in the tail and wing. Ornithopters use to control it with the rudders in the back part of the platform. However, due to the reduced control actuation during the flapping is interesting to analyze to include control on the wings. This work proposed looking at nature mimicking birds’ mechanisms to glide and control their trajectories. Birds can change between gliding and flapping without difficulties. Migratory birds are used to flap to the appropriate wind gust or thermal current. Then they can spend several hours in the flow. These birds can perform long flights looking for a place to live during challenging seasons. The aim is to develop a bird-based wing taking advantage of the geometries and controls ways learned from nature. Nevertheless, the final objective of the aerial platform is to improve the interaction skills of the aerial system. Figure 4.12 shows the conceptual design of the wing. Figure 4.12a shows an eagle taking advantage of the wind gusts to maintain the flight. Figure 4.12b shows the structure of the feathers of a wing. Figure 4.12c summarized the primary mechanism to control the wing’s trajectory, distribution, and geometry. Figure 4.12d shows a schematization of the development proposed. Finally, Figure 4.12e shows the wing developed during this work integrated into the platform created in the project GRIFFIN [65]. The objective is to replicate birds’ wing tip control to improve the control skills, as shown in 4.12c. They use their feathers and the last joint to modify the surface of the wings. This differential actuation generates a moment around the ROLL of the aerial platform, adding another degree of freedom to control the trajectory. To reduce the damage in case of impact. The birds’ wings are flexible and composed of feathers and light bones. The aim is to reduce the stiffness of traditional ornithopter wings, making it possible to maintain flight behavior. The bird-based aerodynamic profile mimics the birds’ gliding behavior. Furthermore, this profile also mimics the movements performed by the feathers during the flapping. The thickness of the profile varies between the leading and trailing edges. The leading edge is much thicker than the trailing edge, making it possible to replicate the wing feathers motion presented in Figure 4.12b. The following sections will focus on the design and manufacturing process of the wing, and Section 5 evaluates its behavior. 110 Chapter 4. Bio-Inspired Morphing Surfaces (a) Eagle flying [147]. (b) Wing feather topology from [148]. (c) Sketch of the mechanism. (d) Sketch of the purposed development. (e) Developed wing integrated into E-Flap[68]. Figure 4.12 Wing conceptual design. The figure shows the concepts learned from nature on the top (a) and (b). The middle part of the figure offers the transcription of these concepts to a sketch of the bird’s wing (c) and a soft folding mechanism (d). In the end, (e) shows the integration of the soft structure with the aerodynamic profile in an ornithopter. 4.2.1 Design This section focuses on the development of the wing. This development’s starting point is analyzing the birds’ wings dimensions. The system developed should overcome the gliding capabilities of a traditional ornithopter, making the platform safer for interaction with humans. However, the flapping flight is the main characteristic of the ornithopter. So the wing developed must have an equilibrium between the flapping and gliding capabilities while improving the interaction and maneuverability skills. By looking at the nature and research works like [ 149 ], birds of prey support most of the inputs for the design. The development focuses on the morphology of a bald eagle. These birds show dexterous control skills having high maneuverability to follow their prey. 4.2 Morphing Wing 111 Figure 4.13 Wing conceptual design from [ 72 ]. The figure summarized the main characteristics and systems developed in the wing. The figure represents the different parts of the wing mapped with a pattern. The mapping legend is on the bottom right side. The top left shows the structure of the folding mechanism, and the bottom left shows the aerodynamic profile of the wing. Another of its main characteristics is its high payload. They can load a weight similar to their mass. A large wingspan and a mid-range aspect ratio characterize their wing geometry. A high lift generation at low flapping frequencies depicts these wings. Compared with other species, the bigger chord size makes it possible to improve the gliding capabilities. Figure 4.13 shows the dimensions of the proposed wing. Figure 4.13 synthesizes development. It includes summarizing all the systems developed in the wing and its main characteristics. The wing geometry replicates bald eagles. The wing span is b=2 m, and the chord c=0.33 m. Then an active tip was designed to mimic the movement’s primary and secondary feathers, having a maximum surface on the unfolded configuration of 0.498 m 2 and a minimum of 0.420 m 2 . The control of the wing’s area generates actuation on the ROLL angle of the aerial platform. These characteristics define the planar structure of the wing. It mimics the feature of the bald eagle’s wings, mainly focused at this point on the Aspect Ratio (AR). Theoretically, this structure maintains these birds’ flapping/gliding capabilities. The next step to continue with the design of the wing is the selection of the aerodynamic profile. The objective is to use an aerodynamic profile that emulates the wing’s feathers during the flapping. Therefore, AS-series from [ 150 ] was evaluated. In this paper, the authors evaluate the design of bird-inspired profiles by using the multipoint inverse airfoil design. Particularly in this work, the profile AS-6091 can maintain the flapping while keeping the gliding capabilities. Figure 4.13.II shows this profile. This profile has a bulbous leading edge mimicking the leading edge of the bird’s wings with a thing tip emulating their feathers. The section of the profile emulates the section of an eagle wing. One of the main novelties of the work is the use of flexible fabric. The objective is to maintain the aerodynamic surface during the tip movement as birds do with their feathers. 112 Chapter 4. Bio-Inspired Morphing Surfaces Table 4.3 Elastomer materials tested. Properties [151] Mixed Viscosity (Pa ·s) Specific Volume (m3/Kg) Pot Life (min) Tensile Strength (KPa) Elongation at break (%) Dragon Skin™10 Slow 23 9.32·10−445 3275 1000 Dragon Skin™30 Slow 20 9.28·10−445 3447 364 Ecoflex™ 00-30 3 9.4·10−445 1378 900 Ecoflex™ 00-35 Fast 3.5 9.4·10−42.5 1378 900 Ecoflex™ 00-50 8 9.32·10−418 2172 980 A wide range of variables takes relevance in the development of the fabric. The fabric has to be as thick as possible while maintaining the aerodynamic surface during the flapping and gliding. Different elastomer products from the company Smooth-On [ 151 ] were evaluated, focusing on Ecoflex ™ Series and Dragon Skin Series ™ . Table 4.3 shows the main properties of the material tested for this development. It is needed a fabric capable of supporting high efforts and high deformations. Concerning this concept, the values of the Dragon Skin ™ Series concretely Dragon Skin ™ 10 Slow show the best values. The properties that set this material as the best from the mechanical point of view are Tensile Strength and Elongation at Break. This material is in the middle of the way between Dragon Skin ™ Series and Ecoflex Series.Dragon Skin ™ 10 Slow have values similar to Ecoflex ™ Series in elongation maintaining the values of resistance to deformation of Dragon Skin ™ Series. However, all the materials on the table support the forces generated during the flapping and gliding phases; the need to have thick low-weighted fabric difficult the manufacturing process. The fabric is handmade due to its novelty development. The fabrication process determines the material used. The selected material will be exposed in the next section when the manufacturing process is determined. Nylon ripstop fabric composed the surface of the non-active part of the wing. Usually, this kind of platform uses this fabric. Its main characteristics are its low weight and resistance to tearing and ripping. Designing the frame of the wings, the concepts learned from the literature, and our knowledge from other wing developments are crucial. The wings are long (2 meters of wing span) and must deliver high forces due to their increased payload capabilities. Therefore, we developed a review of the material proper for the wing structure using 4.2 Morphing Wing 113 (a) Four ribs analysis. (b) Five ribs analysis. Figure 4.14 Ribs displacement/Stress analysis under maximum load conditions: (a) Shows the displacements in [mm] mapped in colors on a wing with four ribs. (b) Shows the displacements generated in a wing with five ribs. Both figures represent the color scale on the left side. Ansys ™ Granta Edupack. Lightweight composites that were stiff enough focused the review. However, there were no surprises. The best material for the wing frame was CF in different compositions. The main reason to select CF is the easy way to find this material on the market in various configurations. The wing has three main parts to be studied separately: the aerodynamic profiles, the folding mechanism, and the transmission part of the wing. The study begins with the aerodynamic profiles. The selected aerodynamic shapes can mimic the movements of the feathers during the flapping. However, the analysis must first set the number and thickness of the profiles to maintain these capabilities in the whole wing. The designing process sets the lift generation to 1.4 kg. The wings absorb this force during the flapping; each half-wing has to absorb 0.7 kg. Therefore, each rib has to drink a proportional part of the force. As a result, the rib integrated into the folding 114 Chapter 4. Bio-Inspired Morphing Surfaces Figure 4.15 Folding mechanism structural study. The figure analyses the displacement in mm performed in the tip with a force of 2.5 N. The figure shows the color scale of the displacements on the left side . mechanism does not have an aerodynamic profile. In the end, the total number of ribs is one more than those with an aerodynamic shape, as shown in Figure 4.13. The analysis evaluated each rib’s displacements separately to analyze the behavior of the ribs that set the wing’s aerodynamic profile. The goal is to maintain the desired deformations to mimic the feathers. To evaluate the deformation Inventor ® Nastran ® was used. Figure 4.14 evaluates the displacement behavior of the profiles under fully loaded. The further analysis presented in this work only shows the displacement results. The aim is to evaluate if the wing developed can perform the desired bioinspired movements. However, the stress analysis evaluates if the mechanism can support the efforts generated. The stress analysis sets the limits and the material used in the wing. The evaluation of the efforts set the thickness of the profile to 1.5 mm. The material selected was CF plates with ply orientation (woven 0/90 ◦ ). The configuration with three ribs was discharged for being close to the break limit of the CF, generating fatigue in the material. Five ribs set up (shown in Figure4.14b) was dismissed for maximizing the deformation of the wing, minimizing the mass. The deformation achieved with four ribs was closer to the expected one, having 2.12 mm of deformation on the tip as shown in Figure 4.14a. The structure can support the stresses generated at 2 Hz of flapping frequency. The security margin in the strains using this configuration is 30 %. The transmission part of the wing is closest to the flapping mechanism and must be stiffer to transmit the flapping frequency. The wing’s folding tip must be flexible enough to perform the folding movements. A CF tube is the material of the transmitter, and the folding mechanism uses a CF rod. The rod is glued inside the transmitter to reduce the weight of the joint. Figure 4.15 shows the study performed in Inventor ® Nastran ® to select the size of the tubes and bars. The bar selected for the tip has 4 mm in diameter with unidirectional fibers to facilitate 4.2 Morphing Wing 115 the folding. The longitude of the bar is 300 mm to set the folding area as designed before. The figure shows the displacement performed with 2.5 N of pull. A servo placed in the body of the ornithopter generates the force, and tendons transmit it inside the transmitter tube till the tip of the folding mechanism, as shown in Figure 4.13. The geometry selected for the transmitter is a 6 mm in diameter tube. The longitude of the bar is 700 mm to generate the required wing span. The rod has 1 mm of thickness and two layers of compressed bidirectional CF fabric (ply woven 0/90 ◦ ). This part of the wing implies high-effort transmission. In addition, the orientation of the fabric’s fibers guarantees the bar’s stiffness in all directions mimicking the wing bones of the eagles. The displacement reached in the folding mechanism was 175 mm. The study presented in Figure 4.15 also shows that the transmitter is stiff enough to maintain a significant part of the deformations in the tip. Furthermore, the stress analysis guarantees the system can support the forces generated. The next stage generates the wing model with all the components. Figure 4.16 shows its displacement and behavior during flapping. First, the selected profiles must support the stresses generated without plastic deformation. The reported geometries achieve this requirement. Then the displacements were analyzed. The goal o this wing design is to mimic the deformation performed by an eagle wing. During the upstroke, the deformations of the profiles (shown in Figure 4.16b) make the area more minor, reducing the resistance to perform the motion. The upstroke generates force opposite to the lift. This area reduction makes it possible to reduce these forces making the flight more stable. Concerning the downstroke, the deformation of the profiles maximizes the wing’s area to increase lift generation. This behavior is possible due to the profiles selected. It perfectly replicates the motion of a falcon’s wings. Theoretically, as shown in the studies, the wing design can mimic the deformations of birds’ wings. This condition increases the behavior of a regular one. However, it is only possible if the manufacturing process maintains the design parameters. Therefore, the following section focuses on the construction process. 4.2.2 Manufacturing Process Figure 4.17 summarizes the construction process. For a clear explanation, the following text divided the description into the parts shown in the list: • Aerodynamic ribs manufacturing: This part is related to the Figure 4.17.1.a. It focuses on obtaining the ribs used to generate the wing profile. • Mechanical parts integration: Shown in Figure 4.17.1.b. It is related to integrating all the parts of the wing’s frame. In this part, the filling material also is developed and integrated. •Ristop fabric unification: This part of the manufacturing process shows comprises the integration of the fabric on the wing structure. Figure 4.17.1.c. shows this process. • Elastomer fabric development: It is focused on generating the fabric. Figure 4.17.2.a shows that this part of manufacturing is one of the most important. 116 Chapter 4. Bio-Inspired Morphing Surfaces (a) Downstroke. (b) Upstroke. Figure 4.16 Wing structural study. The figures show the deformation performed on the wing in the worst load cases possible. (a) During downstroke. (b) During upstroke. The color scale of the displacements is represented on the left side of the figures. • Elastomer fabric integration: This part of the process focuses on developing the folding mechanism with the elastomer fabric, Figure 4.17.2.b. 4.2 Morphing Wing 117 Figure 4.17 Wing manufacturing process. This figure divides the manufacturing process into two. Yellow dashed lines surround the fabrication of the wing’s main structure, and green dashed lines rounded the wing tip’s development. The arrows highlight the order of the processes to develop the wing. The explanation of the manufacturing process follows the order set in the list presented before. The ribs with the desired aerodynamic profile are the first on the list. A CNC machine cut the desired profile from a CF plate. The sketch created in Inventor ® has the geometrical data of [ 150 ]. The software Vcarve translated the sketch to G-Code. Finally, the post-processor WinPC-NC commands the work orders to the machine. The machine used was HIGH-Z S-1400/T CNC ROUTER from the company CNC-STEP. The milling cutter used had 2mm in diameter, specific for CF milling with a diamond structure. The next step is to integrate the developed profiles into their assigned position. As shown in the previous study, these profiles must support high efforts during the flapping. Different methods evaluate how to have a solid joint between the profiles and the transmitter tube. The first idea was to develop a mechanical union, discharged due to the complexity added. The cylindrical shape of the rod makes it difficult to place the joint exactly in the same place. In the end, the tests conducted cause the loss of material. The solution was to glue the profiles to the rod. Nevertheless, this union has to support all the pressure of the wing. Starting from the traditional epoxies and evaluating fast and slow epoxies, the evaluation performed evaluates its capabilities for the application. In the end, Metalset ™ A4 is used. This epoxy has metal particles inside, making a mechanical joint. However, the main capability added to our system by this epoxy is the possibility to increase the contact area. The contact area can distribute the pressure on a bigger surface by making it bigger, reducing the punctual efforts. It replicates perfectly the constraints set in the study; however, it increases the mass of the wing. A custom-made stencil guarantees the place of the profiles in the desired position and 124 Chapter 5. Experimental Validation (a) Tendon Configuration 1 from Figure 3.35a. (b) Tendon Configuration 2 from Figure 3.35b. Figure 5.3 Tendon’s configurations experimental results. (a) Shows the results for Configuration 1 and (b) for Configuration 2. Each level offers four graphs, each applying a different tension to the main tendon, shown at the top of the graphs. The graphs classified the objects in terms of their weight and reliability level. The dashed lines show the borders between the layers shown in the list above. The graph shows the claw’s adaptation results. As expected, adaptation capability increases with the applied tension. However, the charts also show less weighted objects with challenging graspings. These objects used to have polished surfaces and varying shapes. However, the claw proves to guarantee adaptation to these objects. It showed a tight grasp with all the objects measured. Concerning the analysis of the configurations, the behavior of both is close. However, Configuration 2, in which all the fingers have the same tension in the tendons, shows better results for grasping objects. The graphs prove this configuration requires less force to hold the same things. The chart with 8 kg of tension applied shows that the gripper can firmly and forcefully grasp all tested objects. The rest of the graphs also agree with this trend, 5.1 Bio-Inspired Claws Experiments 125 showing better results for Configuration 2 than Configuration 1. It is not a good analysis to measure the difficulties of grasping, considering only its weight. Also, the holding volume is significant and the geometry of the objects held. The following figures show images of experiments with different things that highlight the capabilities of the claw to adapt to multiple geometries. The figures have been classified by the tension applied to the main tendon. Figures 5.4a,b,c,d show the results for 2 kg of force applied, Figure 5.4e,f,g,h for 4 kg, and Figure 5.4i,j,k,l for 8 kg. The figures show the adaptation skills of the claw. Analyzing the experiment performed using 2 kg of tension, which implies applying 0,5 kg in the tendon of each finger, it starts to show tangible results. Concerning adaptation, the image shown in Figure 5.4d demonstrated the skills of the claw to hold firmly spherical shapes. This shape is challenging to keep and usually implies having independent control of each finger. That means the tendons configuration correctly balances the system to reach a tight grasp. Figure 5.4a shows the adaptation to short cylindrical volumes. All the fingers are in contact with the object keeping the object held with this geometry. Figure 5.4b shows the adaptability of the claw using low tension on small holding volumes. In the end, for the configuration analysis using 2 kg of pressure, Figure 5.4b shows the adaptation to an aluminum bar. The edges of this bar profile make it possible for the nail of the claw to go inside, simplifying the grasping. The analysis continues with the following level of the figures, the experiments performed with 4 kg of force applied to the tendon. The graphs continue showing the adaptation of the system to daily objects. Figure 5.4e demonstrates adaptability to pliers. This grasping is interesting to analyze for cooperation work with workers delivering tools in difficult access areas. Figure 5.4f demonstrated the ability of the claw to hold varying shaped objects with high masses. The bottle in the figure has an irregular conical-shaped polished surface. The claw still has some difficulties maintaining this object with 4 kg. However, the grasping is suitable for flight but improves with more tension, as is shown later. Figure 5.4g is interesting. The object is thin, but the claw can maintain the holding autonomously, adapting to its surface. Ultimately of this tension, Figure 5.4h demonstrated the adaptation of the claw to cylindrical polished high-weighted geometries. The claw can maintain the grasping firmly, generating grip enough with the contact surfaces. The final part of the analysis comprises applying 8 kg of tension to the tendon. This force is perfectly reachable by the muscles designed. The objective of analyzing this force is to test the adaptability and evaluate the possible damages the claw can exert. Figure 5.4i shows the capabilities of the claw with a crystal bottle. The bottle has a challenging surface with many polished edges but can still hold the object autonomously. It was able to preserve the bottle without scratching or damage. Figure 5.4j analyzes heavy-weighted conical shapes. The claw has demonstrated that it could generate enough grip to maintain this kind of object held from the middle, taking advantage of edges to increase the grip blocking the motion of the object. The grasping is a balance between grip and adaptation. Ultimately Figures 5.4k and 5.4l analyze the cooperation with humans. These tests examine the damage the claw can exert using heavy load on the tendon. The adaptability shown in Figure 5.4k is remarkable. However, as the holding volume is small, the force applied is also smaller. The test conducted in Figure 5.4l with a human analyzes the possibility of damaging a human. This holding volume is major, so the forces applied are 126 Chapter 5. Experimental Validation (a) Tape 95.4 g. (b) Brush 75.6 g. (c) Square bar 200.5 g. (d) Tennis ball 58 g. (e) Pliers 298.9 g. (f) Glue bottle 281.3 g. (g) Aluminum T-shaped plate 57.2 g. (h) Spray paint bottle 147.3 g. (i) Crystal Liquor bottle 450 g. (j) Glue bottle 281.3 g. (k) Human Finger. (l) Human arm. Figure 5.4 Configuration 2 analysis varying the holding volume. (a)(b)(c)(d) Shows the results using 2 kg of tension in the tendon, (e)(f)(g)(h) uses 4 kg, and (i)(j)(k)(l) 8 kg. Each figure labels the weight of the object held at the bottom, and the image shows its geometry. also significant. This test concludes with little marks on the arm that disappear 1 min after the trial, not even a scratch or hematoma. The setup experiments show excellent results demonstrating that the geometry and internal force distributions perform according to the design specifications. The following 5.1 Bio-Inspired Claws Experiments 127 tests integrate the claw with the actuation of the SMA muscles to further analyze the gripping capabilities. The experiments to analyze the claw’s behavior with the developed artificial muscles evaluate the system holding different objects. The objective is to replicate the system performance shown in the previous experiments. The system setup for this test includes a Thermal/RGB camera focused on the actuation to evaluate the holding during the actuation. This camera was the WIRIS Pro from the company Workswell ® [ 152 ], specially designed for thermal inspection in aerial robots. It is a radiometric inspection thermal camera proper for drones due to its lightweight. The resolution of the thermogram is 1266×1010 px reaching up to 1200 ◦ C, and the resolution of the RGB camera is Full HD (1920 x 1080 px). The accuracy of the thermal measurement is ±2% . The characteristic of the camera is enough to provide a visualization of the system’s behavior. The final actuation control uses Pt100 sensors to reduce the weight. Figure5.5 shows the visualization of the test to extract the conclusions. The figure focus on two experiments. The first one, concerning Figures 5.5a and 5.5b show the sequence of holding a spherical shape. The first shows the RGB view, and the second is the thermal view. This geometry is one of the most challenging to maintain the hold. Other grippers can not retain it because they need to generate more friction with the surface, not to slide. However, our claw can accommodate its geometry to the object generating enough gripping force. This behavior is due to its material composition and the tendons’ tensions distribution. Everything is contributing to this grasping. First, the nails o PLA are stiff enough to generate friction with the surface and little edges. Second, the palms of Ecoflex adapt to the phalanxes developing more grip. Third, the Carbon Fiber structure maintains the longitudinal delivery of forces. Finally, most important, the structure of Thermoplastic Polyurethane (TPU) makes it possible to adapt the shape generating the autonomous holding motion. The sequences expose the evolution of the system during the grasping process. They show that the claw can hold the spherical shape object autonomously in less than two seconds. The first figure is the system’s state before the actuation. The figure in the center shows when we leave the object in hand. The last one on the left of each sequence shows moments before the holding. Figure 5.5b shows that the temperature in the muscle is less than 50 ◦ C to maintain the object. The maximum actuation temperature desired in the actuator is 120 ◦C. So it can still load bigger things. The test shown in Figure 5.5c and Figure 5.5d evaluate the system’s limits. It consists of using a reduced holding volume with a huge load. These experiments are challenging because of the soft capabilities of the claw. Usually, it is impossible to use this kind of robotics system for heavy objects. This behavior is because they used to be deformed by the object, making it impossible to maintain the grip. However, our system’s structure can maintain soft capabilities while exerting considerable force compared with its weight. The figure shows this behavior with a briefcase handle of 2 kg. The final mean temperature of the SMA to maintain the object hold is around 70 ◦ C. Therefore, the actuator is still on elastic deformation, being possible to use it for heavier objects. Furthermore, the claw structure keeps the object weighting only 40 g, making a significant relationship between the force exerted and the system’s weight. 128 Chapter 5. Experimental Validation (a) RGB view holding a spherical shape of 55.4 g. (b) Thermal view holding a spherical shape of 55.4 g. (c) RGB view holding the camera case of 2 kg. (d) Thermal view holding the camera case of 2 kg. Figure 5.5 Thermal view of the actuation. (a) and (b) are the RGB and thermal cameras of the sequence holding a spherical shape of 95.4 g. (c) and (d) are, respectively, the RGB and thermal camera of the grasping of the suitcase of 2 kg. The scale of the thermal camera comes from 10 ◦ C to 100 ◦ C from cold to warm colors. On the bottom side of the thermal figures, the label shows the average temperature of the spring( TmeanS ) in ( ◦ C) and the time of the sequence (t) in (s) . Ultimately, the load was limited to 2 kg for this experiment. However, the behavior shown makes it possible to increase the weight tested, although the capabilities of the ornithopters would need to improve to be tested realistically. Figure 5.5 shows only two experiments in detail. However, these tests with the actuation included a wide range of objects. Figure 5.6 summarizes these experiments. The number of objects held during the experiments was large. The figure shows their geometry and weight. It exposes the quantity and differences in their shapes, which 5.1 Bio-Inspired Claws Experiments 129 Figure 5.6 Holding Experiments Summary. This figure shows a summary of the experiments that analyze the adaptability of the claw to a wide range of objects changing its size and shape. differ in the holding volumes. The following text highlights the behavior of the claw in particular cases. The behavior shown in the experiments with the 2 kg case highlights the improvement of the ratio (Object weight / Handling system weight), but also that the system can hold variable shapes with a high weight, such as the yellow glue bottle at the bottom of the second one starting from the left. The case of the reduced screwdriver’s volume on the top right side is also important. Also, the cylindrical shapes on the top of the five figures start from the left with different masses and diameters, it also interesting. The figure on the middle bottom side shows the claw holding a whiteboard eraser in which the gripper looks autonomously for the edges, maintaining the grip. In the end, the claw uses its soft characteristics to be adaptable to multiple shapes, the force generation of the muscles to apply enough force to the tendons, and the longitudinal stiffness to maintain the object hold. This behavior is a good representation of the birds’ claws. In the end, it is difficult to measure this behavior objectively. Figure 5.7 do it by measuring the objects’ weight and estimating the grasping volume. The figure classified the grasping volume into five groups. The explanation follows the order in the legend of the chart. The first one is Cylindrical volumes. This volume refers to the case of the paint bottle shown in Figure 5.4h. The next is the Circular volumes. It refers, for example, to the tape shown in Figure 5.4a. The third is Spherical volumes. This one comprises the cases of the different balls tested. The fourth is Rectangular volumes. In this case, the grasping volume is a parallelogram. It is shown in Figure 5.4a. The fifth is Varying volumes. It refers to grasping volumes in which the volume varies inside the claw, for example, Figure 5.4j. (Note: The use of these figures is only for clarification; the experiments shown on this chart also implied the SMA actuation). The experimental results before this summary conclude that the claw can manage various shapes. However, Figure 5.7 objectively summarizes these experiments. First, it classifies 130 Chapter 5. Experimental Validation Figure 5.7 Holding analysis. The figure summarizes the objects experimented with in terms of their holding shape, weight, and grasping volume. The legend shows the five grasping volumes classification evaluated, represented with different marks. The chart uses these marks to develop the representation. the object by its grasping volume in m3 , weight in g, and the classification of the grasping volume showed before. These parameters are the parameter that our research identifies as the most important ones regarding manipulation. Then the chart shows the range of shapes and weights the claw can hold classified into those parameters. Concerning grasping volumes, the manipulator can hold objects a 500% bigger than the smaller ones. The soft adaptability of the claw provides this behavior, but also, the longitudinal stiffness of the toes takes relevance, focalizing the pressure of the actuation in the object. These results are novel for soft robotics manipulators and difficult to see in other actuators with the same characteristics. Concerning the weight of the objects, an actuator with less than 40 g can manage up to 2 kg. These results also highlight the resistance of the actuator to deformation. In some way, the claws fuse their palms with the objects avoiding damage. The longitudinal stiffness and the force actuation of the developed artificial muscles make this behavior possible. This behavior is also a novelty for soft actuators. Usually, they only can manage small loads to maintain their characteristics. Ultimately, one of the most challenging situations is grasping varying volumes. It implies autonomously adapting the tension in each finger’s tendons to generate enough grip. Regarding these results, studying the tendon distribution of forces was crucial. The behavior demonstrated by the robotic claw goes through the state of the art of soft robotics manipulators. The gripper is capable of mimicking birds’ claws, improving their adaptability. Sometimes, the birds’ feet are challenging to cooperate with, and the operators use protective gloves to deal with them. This scenario is the case of the falconry. 5.1 Bio-Inspired Claws Experiments 131 The development of the claw takes inspiration from falcons and eagles. However, It improves the adaptability to different shapes with a softer structure maintaining the natural characteristic. The following experiments analyze the claw’s behavior for object delivery. The objective is to test the claw concept and the antagonist muscles’ configuration controlling the release of objects during the flight. (a) Delivery of medical resources. (b) Delivery of a security vest. (c) Delivery of a Micro Aerial Vehicles (MAV). Figure 5.8 Delivering objects analysis. The figure shows the ornithopter transporting objects to specific locations. The figure’s blue arrows and red circles point out the things transported. These tests were possible within the first version of the claw. In the first version, the number of objects was more limited. However, it demonstrates that the system can deliver things in a controlled way. The last version of the claw can manage more objects more nimbly. Figure 5.8a shows the ornithopter transporting a bandage. The tests show that the ornithopter can transport small medical resources for a person in need. In addition, this platform increases the safety provided by other aerials platforms avoiding propellers and reducing the impact energy. These characteristics make ornithopters one of the safer 132 Chapter 5. Experimental Validation platforms for human interaction. Figure 5.8b demonstrates the ability to support significant loads. In this figure, the ornithopter is transporting a life vest. In this case, the object’s aerodynamic influence is more considerable, causing the platform’s control difficulties. This situation improved by adding the soft aerodynamic control surfaces studied, making it possible to increase the control capabilities in these situations. Concerning the grasping and release, the claws showed good behavior delivering the object to the desired point. Figure 5.8c shows a Micro Aerial Vehicles (MAV) release to inspect specific places. The objective is that the reduced consumption of the ornithopter makes it possible to reach the farthest locations. Then, the MAV is released to perform the inspection. After finishing the review, it returns to the claw, and both return to the base. This experimental validation realizes the capabilities of the claw to manipulate and possible applications of ornithopters. The following section focus on the analysis of the perching skills of the system. 5.1.2 Perching Experimentation This section focuses on the evaluation of the system’s perching behavior. The perching involves the claw experimented in the previous section and the legs. The design process sets the legs’ functionalities. They have to support the impacts contributing to the perching and control the position of the actuator. The section starts by defining the experimentation setup. One of the most critical parameters in ornithopters is the weight of the systems. Therefore, Figure 5.9 shows the weight of the different parts of it. Figure 5.9 Experimental setup weight. The figure shows the weight of the different parts of the manipulation system. The representation rounds with a circle, label the component, and an arrow connects the elements with their corresponding part of the graphic. 5.1 Bio-Inspired Claws Experiments 133 The figure divides the manipulation system into three main parts. The first one is electronics. This part involves the muscles, control electronic and actuation electronics presented in the design process. Its weight is 50 g, 27% of the system’s total weight. The second part is the legs. The legs are composed of their structure and the damping system. Its weight is 44 g, 24% of the total weight. The last part, and the most important, is the claws. It constitutes 49% of the total weight of the system. Their contribution is essential to keep the perching balanced. Finally, the total weight of the manipulation system is around 15%. When compared with birds, the percentage of the mass of this system is smaller than the examples found in nature. In nature, this percentage rounds the 30% reaching the 40% bird’s weight. The first analysis developed during the experimentation focuses on the claw’s position control. This system manages the positioning of the claw during the flight and keeps the platform’s balance during the perching. Figure 5.10 presents the setup for these experiments. Figure 5.10 Leg positioning control experimental setup. The figure divides the configuration for these experiments into two groups: the onboard and the offboard setup. The onboard setup is on the right side, including all the components inside the ornithopter during the experimentation. The offboard setup is on the right side and shows the representation setup and data used. The figure highlights two kinds of setups for this experiment. The first one on the left side is the offboard setup. The figure shows all the signals the processor board receives for the representation. In this case, the board was Arduino. This board analyzes the actuation data generated by the controller and the different signals of the sensors onboard. Arduino processes the data and sends it through the serial port to a computer for online representation. The computer processes the data in Matlab ® , making it possible to analyze the behavior online. The online experimental control is crucial during the first trials because the system has high instant energy consumption, and the battery can suffer damages with inadequate actuation.