YOLO-Based Vision Perception and Automation for Agricultural Robots: A Comprehensive Survey of Algorithms,...

Deep learning-driven computer vision has become a fundamental enabling technology for agricultural robotics, providing reliable perception capabilities for precision farming and autonomous field operations. Among various object detection frameworks, the You Only Look Once (YOLO) family has attracted extensive attention due to its favorable balance between detection accuracy, computational efficiency, and real-time deployment capability in complex and dynamic agricultural environments. This paper presents a comprehensive review of YOLO-based vision perception and automation technologies for agricultural robotic systems. The evolution of the YOLO framework from YOLOv1 to recent YOLOv11 variants is systematically reviewed, with particular emphasis on architectural innovations, performance improvements, and adaptations to agriculture-specific challenges, including complex backgrounds, illumination variations, severe occlusion, and multi-scale small object detection. Beyond algorithm-level analysis, this survey investigates the integration of YOLO-based perception modules into robotic systems for practical applications, including autonomous navigation, precision spraying, crop monitoring, disease detection, and robotic harvesting. Furthermore, this review analyzes the complete deployment pipeline of YOLO-based agricultural robots, covering benchmark datasets, evaluation metrics, model optimization strategies, and hardware constraints for edge computing platforms. Emerging research directions are also discussed, including lightweight YOLO architectures, multimodal vision-language models, and cloud-edge collaborative intelligence for next-generation agricultural automation. By bridging advances in computer vision algorithms with robotic perception and execution, this survey provides a comprehensive perspective on the development of intelligent and efficient agricultural robotic systems.