Autonomous Robotic Systems in Agriculture 5.0: A Review on Technical Architecture, Operational Applications, and Future Perspectives

Abstract

This comprehensive narrative review examines the current state and future development perspectives of autonomous robotic systems within the scope of the Agriculture 5.0 paradigm, which aims for sustainable and efficient agricultural production. In this paper, the cognitive and physical structure of robotic systems is addressed through the "Eye-Brain-Foot Hand" hierarchy, and the role of next-generation Vision-Language Action (VLA) models, such as RT-2 and OpenVLA, in contextual agronomic decision-making processes is analyzed. Within the scope of field applications, the SAM2-based SPAF framework, which offers 95.54% harvesting success rate in tomatoes, and spot-spraying technologies that reduce chemical input volume by 90% are evaluated using quantitative data. Furthermore, the synergy between heterogeneous UAV-UGV fleets, trust-building through Explainable Artificial Intelligence (XAI), and emerging advanced technological approaches such as Quantum Reinforcement Learning (QRL) are assessed. The results demonstrate that autonomous agriculture is not merely a tool to enhance technical efficiency, but is transforming into a cyber-physical production ecosystem where ethical responsibility, data sovereignty, and environmental sustainability converge.