Smart Robotics in Viticulture: Enhancing Monitoring and Manipulation

Smart Robotics in Viticulture: Enhancing Monitoring and Manipulation Erfan Khiali1†, Gabriele Ribolla1,2†, Francesca Negrello1†, Fabio Previdi2†, and Manuel G. Catalano1,2† Abstract—To meet rising quality standards in plant production while addressing increasing environmental and societal concerns, the agricultural industry faces growing pressure to develop innovative solutions. Thus, novel approaches are essential to address sustainable practices, ongoing labor shortages and improve tasks throughout the life cycle of agricultural products. In this context, our project focuses on the design, control, and preliminary validation of a robotic system for precision agricultural processes, guided by stakeholder and end-user feedback. The platform integrates three modules: navigation, which employs path planning and obstacle avoidance; perception, to detect grapes and identify their positions within the vineyard; and manipulation, to perform tasks such as inspection and precise treatment application. Index Terms—Precision agriculture, vineyard monitoring and management, autonomous robots I. INTRODUCTION The agricultural sector is currently facing significant challenges [1], including the need to increase crop production to feed a growing global population, maintain sustainable agricultural practices despite labor shortages and fluctuating tariffs and simultaneously minimize environmental impacts [2]. Vineyards, as high-value perennial crop eco-systems, demand careful management throughout their growth cycle, incorporating tasks such as inspection, monitoring, and precise application of treatments like spraying [3]. In this context, end users have emphasized that traditional manual practices for these tasks are labor intensive, time-consuming and often rely on experience without proper data registration, making them neither easily replicable nor robust. Robotic systems can assist by collecting data and executing repetitive, datainformed actions, thereby improving resource efficiency, optimizing production, and supporting operator activities. On the other hand, a robotic platform tailored for vineyards faces several challenges including navigation in unstructured and often narrow environments [4], accurately detecting and localizing grape clusters [5], and executing delicate tasks like spraying or pruning without causing damage to the vines [6], all of which complicate autonomous navigation, perception and manipulation tasks. Recent advances have opened new avenues for developing autonomous platforms capable of performing complex operations such as an economical robotic research platform designed for autonomous vineyard inspection, equipped with sensors †All the authors are part of the JOiiNT LAB project 1Soft robotics for Human Cooperation and Rehabilitation, Istituto Italiano di Tecnologia,via Morego, 30, 16163 Genova, Italia 2Dipartimento di Ingegneria Gestionale, dell’Informazione e della Produzione, Universit` a di Bergamo, Bergamo, Italia Fig. 1. Proposed robotic platform performing: (1) grape inspection, (2) treatment application (3) grape geo-localization and labeling, and (4) navigation. for 3D mapping, grape detection and preliminary plant health evaluation [7], as well as a system that integrates computer vision for crop detection with a closed-loop pressure control mechanism for accurate pesticide application [8]. However, the complete and comprehensive integration of navigation, perception and manipulation remains an open challenge and human collaboration is still required, particularly for complex tasks and high-level decision-making. Moreover, the effective deployment of such platforms depends on the adoption of advanced learning methods and on achieving significant cost reductions before broader adoption becomes feasible, especially in smaller vineyards [1]. Conducted within JOiiNT LAB, in collaboration with SDF, an agricultural machinery manufacturer, and Le Corne, a winery that provided the end-user perspective, this study presents a compact and multifunctional robotic platform for precision vineyard tasks. A system that integrates diverse modules within a unified, user-friendly architecture enabling it to perform vineyard operations such as inspection and treatment. The platform was validated through trials in a controlled indoor synthetic vineyard, where it successfully navigated, detected, and interacted with grapevines, highlighting its potential to improve vineyard management efficiency. II. PROPOSED SOLUTION Considering the discussed open challenges, the proposed robotic platform (Fig. 1) comprises key components such as a mobile base, a collaborative arm, a depth camera along with auxiliary components including control and computational units, and batteries (providing about 4 hours of autonomy). The software architecture integrates three main modules: per2025 I-RIM Conference October 17-19, Rome, Italy ISBN: 9788894580570 10.5281/zenodo.17629728 103 Fig. 2. The sequence demonstrates the comprehensive testing process of the system, including (from left to right): operational map generation, spatial data collection phase, spray task assignment and execution via GUI and an autonomous inspection mission of the entire synthetic vineyard. ception, which leverages CNN-based object detection, crop estimation, and data management algorithms to support fruit detection, identification, tracking, and accurate geo-localization (mapping) of grape clusters; navigation, which manages autonomous movement and obstacle avoidance of the mobile base using sensor inputs; and manipulation, which utilizes data collected by the perception module to enable precise inspection and spraying tasks, ensuring targeted and accurate interaction with the grape clusters, all coordinated through a high-level, user-friendly interface enabling non-experts to plan missions while the robot autonomously executes operations. III. PRELIMINARY RESULTS As an initial evaluation, the platform was validated in a laboratory vineyard setup designed to replicate real vineyard conditions, with grapevines arranged in rows and grape clusters positioned at varying heights and locations to test the robot’s navigation, detection, tracking, and manipulation capabilities through the dedicated GUI. The validation process, illustrated in Fig. 2, began with environment map generation, enabling autonomous navigation by scanning the surroundings with LiDAR sensors to identify and distinguish obstacles. Subsequently, utilizing the user interface, the grape mapping phase initiated, where the system autonomously navigated through the synthetic vineyard rows and collected data such as the 3D positions of the grapes relative to the camera, the mobile base position with respect to its home location on the created map and assigned a unique identifier to each valid detection. Using this information, the robot successfully executed manipulation tasks on the identified grape clusters such as inspection, which provided non-expert users with the ability to remotely observe the grapes from multiple angles, and precision spraying, where a controlled amount of liquid treatment was applied directly onto the targeted clusters. Finally, as part of the robot’s autonomous mission, the platform was commanded via the user interface to execute operatorselected tasks such as inspection and spraying on either all or specific grape clusters without further human intervention. IV. CONCLUSIONS AND FUTURE PERSPECTIVES This paper presented a prototype robotic platform designed for precision vineyard operations, aiming to address key challenges in modern agriculture by integrating navigation, perception, and manipulation into a unified framework. The validation results highlight the potential of autonomous robotic solutions to support viticulture and contribute to more sustainable agricultural practices. Future work will focus on evaluating the system in real vineyard conditions, where the platform can be tested under diverse environmental factors. ACKNOWLEDGMENT We would like to express our gratitude to Vinicio Tincani and Alessandro Marchetti for their technical support. REFERENCES [1] M. Spagnuolo et al. Agricultural robotics: A technical review addressing challenges in sustainable crop production. Robotics, 14(2), 2025. [2] B. Ahmed et al. Smart agriculture: Current state, opportunities and challenges. IEEE Access, PP:1–1, 01 2024. [3] M. Ammoniaci et al. State of the art of monitoring technologies and data processing for precision viticulture. Agriculture, 11(3), 2021. [4] L. Wijayathunga et al. Challenges and solutions for autonomous ground robot scene understanding and navigation in unstructured outdoor environments: A review. Applied Sciences, 13(17), 2023. [5] Y. Tang et al. Recognition and localization methods for vision-based fruit picking robots: A review. Frontiers in Plant Science, 11:510, 2020. [6] T. Jin et al. Robotic arms in precision agriculture: A comprehensive review of the technologies, applications, challenges, and future prospects. Computers and Electronics in Agriculture, 221:108938, 2024. [7] S. Facenda et al. 3d robotics and lmm for vineyard inspection. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XLVIII-G-2025:431–438, 07 2025. [8] F. Nasir et al. Precision agricultural robotic sprayer with real-time tobacco recognition and spraying system based on deep learning. PLOS ONE, 18:e0283801, 03 2023. 104