Energy-Aware Admittance Control for Human-Robot Interaction

Energy-Aware Admittance Control for Human-Robot Interaction 1st Matteo Nini DISMI UNIMORE Reggio Emilia, Italy [email protected] 2nd Andrea Pupa DISMI UNIMORE Reggio Emilia, Italy [email protected] 3rd Cristian Secchi DISMI UNIMORE Reggio Emilia, Italy [email protected] 4rd Cesare Fantuzzi DISMI UNIMORE Reggio Emilia, Italy [email protected] 5rd Federica Ferraguti DISMI UNIMORE Reggio Emilia, Italy [email protected] Abstract—Human–robot collaboration demands controllers that are both intuitive and safe. While admittance control ensures compliant interaction, it does not guarantee compliance with ISO/TS 15066 safety limits. We propose a framework that unifies stability and safety in variable admittance control. Passivity is maintained through an energy tank, and safety is enforced via a damping injection strategy that reduces kinetic energy when Power and Force Limiting (PFL) thresholds are exceeded. Experiments with a UR10e robot show that unsafe energy levels are rapidly dissipated while interaction remains natural, enabling safer and more effective collaborative robotics. I. INTRODUCTION The rise of collaborative robotics is transforming industrial workplaces into shared spaces where humans and robots can cooperate directly. This shift enables physical human–robot interaction (pHRI), where robots exchange forces with operators in tasks such as collaborative transportation [1] or lead-through programming [2]. Admittance control is widely used to provide intuitive, compliant behavior in pHRI, and several works have addressed stability and passivity [3]. More recent contributions also consider safety, e.g., via constraintbased controllers [4] or adaptive strategies [5], but these either neglect stability or fail to guarantee compliance with standards. Safety in HRC is defined by ISO/TS 15066 [6], which specifies admissible force, pressure, and energy limits. Among collaborative modes, Power and Force Limiting (PFL) is the only one allowing contact, provided collisions remain below thresholds. Existing PFL methods often enforce conservative velocity limits [7], integrate safety into trajectory planning [8], or adopt energy-based formulations [9], but most either reduce responsiveness or compromise stability. This work introduces a safe variable admittance framework explicitly integrating stability and ISO/TS 15066 safety. Building on [10], it employs an energy tank to maintain passivity and a damping injection strategy to dissipate excess kinetic energy when constraints are violated. Experiments on a UR10e collaborative robot confirm that safety is enforced without compromising natural interaction. II. PROPOSED FRAMEWORK The proposed scheme is designed to guarantee both safe and stable interaction during physical human–robot collaboration. The structure, illustrated in Fig. 1, integrates three modules that work in real time to regulate the robot dynamic behaviour. Fig. 1. Framework overview: monitoring, passivity (energy tank), and PFL constraints jointly adapt admittance parameters for safe pHRI. 1) Scene Monitoring. The robot’s velocity and external interaction forces are combined with motion capture data of the human body part most exposed to collision. This information provides the basis for real-time evaluation of both safety and stability conditions. 2) Passivity Constraint. To prevent instability, a virtual energy tank is introduced. This mechanism stores the energy dissipated by the system and regulates inertia variations, ensuring that the controller cannot inject energy beyond what is available in the tank. As a result, the system remains passive, preserving stability under unpredictable interaction conditions. 3) PFL Constraint. Compliance with ISO/TS 15066 is enforced by explicitly evaluating the admissible kinetic energy at collision. If the predicted energy transfer exceeds the limit defined for the monitored body part, the controller activates a dedicated safety mechanism. Adaptive Strategy. The controller operates in two modes: •Nominal Interaction: inertia and damping increase and decrease proportionally depending on the kinetic energy at collision. This preserves the ratio between the two parameters, ensuring that the operator perceives a consistent interaction. •Safety Activation: if the PFL constraint is violated, a damping injection is applied, while inertia values are kept constant to avoid unsafe increases in mass. This dissipates excess kinetic energy and reduces robot velocity until safe conditions are restored. Afterwards, damping gradually 2025 I-RIM Conference October 17-19, Rome, Italy ISBN: 9788894580570 10.5281/zenodo.17629658 73 Fig. 2. Experimental results for chest area (limit 1.6 J): inertia/damping adaptation, energy loss vs. threshold, human–robot distance, and energy tank dynamics. decays back to its nominal value, recovering the original interaction feeling. The framework unifies stability and safety into a single adaptive admittance controller. By combining energy-aware passivity preservation with ISO-compliant safety enforcement, the approach enables robots to remain responsive and intuitive while guaranteeing that any collision remains harmless. III. PRELIMINARY RESULTS The framework was validated in an experimental setup involving a UR10e collaborative manipulator and a human operator. The robot was controlled through ROS2 at 500Hz, while the operator’s motion was monitored using an OptiTrack motion capture system. A reflective marker was placed near the operator’s chest, corresponding to a body region for which ISO/TS15066 prescribes a maximum admissible impact energy of Emax = 1.6J. During the experiments, the operator manually guided the robot and intentionally moved towards the monitored body region to provoke situations in which the safety threshold could be exceeded. The system’s response was evaluated under two operating regimes: Nominal interaction. When the predicted energy loss ∆Ke remained below the ISO-defined threshold, the controller adapted the inertia and damping values proportionally. This resulted in smooth variations of the robot’s dynamics, preserving both passivity and an intuitive feeling for the human operator. Safety activation. When ∆Keexceeded the admissible limit, the PFL constraint triggered the damping injection mechanism. In this phase, inertia was frozen while damping was rapidly increased, leading to an immediate reduction in robot velocity and kinetic energy. Immediately, the energy transferred in a potential collision was brought back below the safety limit. Subsequently, damping values gradually decayed to restore the nominal interaction ratio. Key observations. i.) The system reacted selectively: adaptations were triggered only when the operator and robot moved towards each other, as required by the safety model. ii.) Passivity was preserved throughout the experiments, as confirmed by the bounded dynamics of the energy tank. iii.) The operator perceived the robot as responsive and predictable, with no sudden or destabilizing changes in behaviour. These preliminary results in Fig. 2 confirm that the proposed framework can effectively enforce ISO/TS 15066 constraints while maintaining stable and natural physical interaction. The experimental validation provides a strong basis for extending the method to more complex collaborative tasks in industrial environments. IV. CONCLUSION AND OUTLOOK Preliminary experiments with a UR10e collaborative manipulator confirmed the effectiveness of the approach. When interaction remained within safe limits, the system adapted inertia and damping proportionally, maintaining responsiveness and transparency. When unsafe energy levels were detected, the damping injection strategy quickly dissipated excess kinetic energy, restoring compliance with the ISO thresholds immediately. REFERENCES [1] J. E. Solanes, L. Gracia, P. Munoz-Benavent, J. V. Miro, M. G. Carmichael, and J. Tornero, “Human–robot collaboration for safe object transportation using force feedback,” Robotics and Autonomous Systems, vol. 107, pp. 196–208, 2018. [2] M. Ragaglia, A. M. Zanchettin, L. Bascetta, and P. Rocco, “Accurate sensorless lead-through programming for lightweight robots in structured environments,” Robotics and Computer-Integrated Manufacturing, vol. 39, pp. 9–21, 2016. [3] A. Q. Keemink, H. Van der Kooij, and A. H. Stienen, “Admittance control for physical human–robot interaction,” The International Journal of Robotics Research, vol. 37, no. 11, pp. 1421–1444, 2018. [4] B. Muraccioli, C. Mathieu, B. Mehdi, and V. Gentiane, “Demonstrating a control framework for physical human-robot interaction toward industrial applications,” arXiv preprint arXiv:2502.02967, 2025. [5] J. Chen, Z.-Y. Dong, S. Shi, Y. Wei, X. Yu, and L. Ou, “Adaptiveconstrained admittance control for physical human–robot interaction,” Transactions of the Institute of Measurement and Control, p. 01423312241298351, 2025. [6] ISO/TS 15066:2016(E). Robots and robotic devices–Collaborative robots, International Organization for Standardization Technical Specification, Feb. 2016. [7] A. Palleschi, M. Hamad, S. Abdolshah, M. Garabini, S. Haddadin, and L. Pallottino, “Fast and safe trajectory planning: Solving the cobot performance/safety trade-off in human-robot shared environments,” IEEE Robotics and Automation Letters, vol. 6, no. 3, pp. 5445–5452, 2021. [8] A. Pupa, M. Minelli, and C. Secchi, “A time-optimal energy planner for safe human-robot collaboration,” in 2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 17 373–17 379. [9] F. Benzi, F. Ferraguti, and C. Secchi, “Energy tank-based control framework for satisfying the iso/ts 15066 constraint,” arXiv preprint arXiv:2304.14059, 2023. [10] F. Ferraguti, C. Talignani Landi, L. Sabattini, M. Bonfe, C. Fantuzzi, and C. Secchi, “A variable admittance control strategy for stable physical human–robot interaction,” The International Journal of Robotics Research, vol. 38, no. 6, pp. 747–765, 2019. 74