农业装备与车辆工程2025，Vol.63Issue(5)：16-23,8.DOI:10.3969/j.issn.1673-3142.2025.05.003
基于视觉识别的水果采摘机器人关键技术研究进展与展望
Research progress and prospect of key technologies of fruit picking robot based on visual recognition
摘要
Abstract
To analyze the recognition ability and picking accuracy of the fruit picking robot based on visual recognition in the complex and changeable agricultural environment,the research progress of the key technologies of the visual picking robot was systematically sorted out.The recognition algorithms(threshold method,edge detection,deep learning)of the visual system,the positioning algorithms(sensor fusion,3D point cloud registration,visual SLAM),and the motion planning algorithms(path optimization,adaptive control)were focused on.Comparative analyses were carried out by combining typical fruit picking cases such as strawberries and apples.The research showed that the deep learning algorithm of the recognition technology could achieve a recognition accuracy rate of over 95%in complex environments,the multi-sensor fusion positioning technology could control the error within±5 mm,and the development of motion trajectory algorithms for different actuators could increase the average picking efficiency by 40%.In the future research,it is necessary to further improve the environmental adaptability of the visual system,develop lightweight visual models and universal end-effectors,and promote the development of fruit picking robots towards the directions of intelligence,high precision,and low cost.关键词
水果采摘机器人/视觉系统/识别算法/定位算法/运动轨迹算法Key words
fruit harvesting robots/vision systems/recognition algorithms/localization algorithms/motion trajectory algorithms分类
农业科技引用本文复制引用
欧骏,宁枫,程碧懿,王红军..基于视觉识别的水果采摘机器人关键技术研究进展与展望[J].农业装备与车辆工程,2025,63(5):16-23,8.基金项目
国家自然科学基金面上项目(32372001) （32372001）
广东省自然科学基金面上项目(2025A1515012307) （2025A1515012307）
广州市基础与应用基础研究专题青年博士"启航"项目(2025A04J1177) （2025A04J1177）
