【摘要】 针对变电站工程中的电缆敷设机械化程度低，人工敷设成本高等问题，提出一种基于视觉引导的电缆敷设抓取系统。首先，系统使用YOLOv8目标检测算法对电缆进行识别，并利用深度相机获取目标区域的局部点云数据，进一步地，利用滤波以及下采样算法对点云数据进行预处理，对预处理后的点云数据进行分割聚类以得到目标电缆点云，计算中心点位置作为目标电缆的三维坐标。然后，利用机器人手眼标定技术转换为机器人的抓取坐标，实现电缆的识别与定位，最后，通过串口通信，发送指令控制机械臂和夹爪完成目标电缆的抓取。为了验证所提出系统的可靠性，开展了验证实验，实验结果表明：抓取系统能够很好地完成电缆的识别定位与抓取，抓取成功率达到92%,研究成果可为电缆敷设机器人的后续研究提供技术支持。更多还原

【Abstract】 To address the problems of low mechanization degree and high manual laying cost in cable laying for substation projects, a vision-guided cable laying and grasping system was proposed.First, the system used the YOLOv8 object detection algorithm to identify cables, and utilized a depth camera to obtain partial point cloud data of the target area.Further, filtering and downsampling algorithms were used to preprocess the point cloud data.The preprocessed point cloud data underwent segmentation clustering to obtain the target cable point cloud, and the center point position was calculated as the 3D coordinate of the target cable.Then, the robot hand-eye calibration technology was utilized to convert it into the robot’s grasping coordinates, achieving cable identification and positioning.Finally, through serial communication, commands were sent to control the robotic arm and gripper to complete the target cable grasping.To verify the reliability of the proposed system, validation experiments were conducted.Experimental results indicate that the grasping system can very well complete cable identification, positioning and grasping, with a grasping success rate reaching 92%.The research results can provide technical supports for subsequent research on cable laying robots.更多还原