Tan Yarong, Liu Xin, Zhang Jinmeng, Wang Yigang, Hu Yanxiang
Institute of Data Science and Agricultural Economics, Beijing Academy of Agriculture and Forestry Sciences, Beijing 100097, China.
Sensors (Basel). 2025 Jun 12;25(12):3677. doi: 10.3390/s25123677.
Fruit and vegetable picking robots are considered an important way to promote agricultural modernization due to their high efficiency, precision, and intelligence. However, most of the existing research has sporadically involved single application areas, such as object detection, classification, and path planning, and has not yet comprehensively sorted out the core applications of deep learning technology in fruit and vegetable picking robots, the current technological bottlenecks faced, and future development directions. This review summarizes the key technologies and applications of deep learning in the visual perception and target recognition, path planning and motion control, and intelligent control of end effectors of fruit and vegetable picking robots. It focuses on the optimization strategies and common problems related to deep learning and explores the challenges and development trends of deep learning in improving the perception accuracy, multi-sensor collaboration, multimodal data fusion, adaptive control, and human-computer interaction of fruit and vegetable picking robots in the future. The aim is to provide theoretical support and practical guidance for the practical application of deep learning technology in fruit and vegetable picking robots.
果蔬采摘机器人因其高效、精准和智能，被视为推动农业现代化的重要途径。然而，现有的大多数研究只是零星地涉及单一应用领域，如目标检测、分类和路径规划，尚未全面梳理深度学习技术在果蔬采摘机器人中的核心应用、当前面临的技术瓶颈以及未来发展方向。本文综述了深度学习在果蔬采摘机器人视觉感知与目标识别、路径规划与运动控制以及末端执行器智能控制方面的关键技术与应用。重点关注与深度学习相关的优化策略和常见问题，并探讨深度学习在未来提高果蔬采摘机器人感知精度、多传感器协作、多模态数据融合、自适应控制和人机交互方面的挑战与发展趋势。目的是为深度学习技术在果蔬采摘机器人中的实际应用提供理论支持和实践指导。
