我所殷辰堃老师课题组完成的论文《Localization of Occluded Target and Obstacle-Avoidance Path Planning for Fruit Harvesting Robots》于2025-05在《IEEE Data Driven Control and Learning Systems Conference (DDCLS)》正式发表。论文摘要如下：
Abstract: With the increasing labor demand of fruit harvesting in modern agriculture, the demand for automated orchard harvesting has become increasingly urgent. However, fruit occlusion presents significant challenges to traditional localization methods, resulting in insufficient accuracy and poor robustness in 3D coordinate estimation and grasping path planning. This paper proposes an occluded target localization method integrating YOLOv8 object detection, Canny edge detection, and depth information compensation. Combined with proposed obstacle 3D modeling algorithm, this approach enables 3D obstacle modeling in occlusion scenarios. The Damped Least Squares (DLS) method is employed to solve inverse kinematics, while the Expansive Space Trees (EST) algorithm is utilized for obstacle-avoidance path planning. A comprehensive method was developed for apple grasping, incorporating both apple localization and motion algorithms, with the use of a single-view depth camera. The feasibility of the proposed approach was validated in the CoppeliaSim simulation environment, demonstrating a comprehensive and practical solution for apple harvesting under occlusion conditions.
引用信息：YuXuan Xu, Chenkun Yin, Yuxiang Liu. Localization of Occluded Target and Obstacle-Avoidance Path Planning for Fruit Harvesting Robots. IEEE Data Driven Control and Learning Systems Conference (DDCLS), 2025-05. DOI: 10.1109/ddcls66240.2025.11065323