摘要
机械臂操作动态工件在分拣、上下料等作业场合有着广泛的需求。针对特定型号工件的移动机械臂动态抓取问题,提出一种基于特定工件ID编号识别和机械臂抓取位姿预测的运动工件鲁棒抓取方法。该方法选取AprilTag视觉标签来标记工件,通过图像采集、自适应阈值处理、连续边界分割、四边形拟合、快速解码等数据处理过程,获取局部视野下运动工件的ID、位置和姿态,实现工件的初次定位。结合移动机械臂的运动规划工艺策略,对特定型号目标工件的抓取位姿进行计算,实现抓取时刻的工件位姿预测。搭建了动态抓取实验平台进行了视觉标签ID识别、工件定位、动态目标抓取实验。结果表明,所提出的移动机械臂动态抓取方法,可以准确识别特定型号工件,并能够对不同运动速度的特定工件进行鲁棒抓取,抓取成功率平均值约为98%。该方法在智能产线的移动机械臂上下料等场景有着广泛的应用价值。
Performing moving workpiece operations with operator has a wide range of demands in sorting,loading/unloading operations.A robust gripping method for moving workpieces based on specific workpiece ID number recognition and operator grasping position prediction is proposed for the dynamic gripping problem of specific types of workpieces.This method selects the AprilTag visual label to mark the workpiece,and through data processing processes such as image acquisition,adaptive threshold processing,continuous boundary segmentation,quadrilateral fitting,and fast decoding,obtains the ID,position,and posture of the moving workpiece under local vision,realizing the initial positioning of the workpiece.Combined with the motion planning process strategy of the mobile manipulator,the grasping pose of a specific type of target workpiece is calculated to achieve the prediction of the workpiece position at the grasping time.A dynamic grabbing experimental platform was built for visual label ID recognition,workpiece positioning,and dynamic object grabbing experiments.The results show that the proposed dynamic gripping method for mobile robotic arms can accurately identify specific types of workpieces and perform robust gripping on specific workpieces with different motion speeds,with an average gripping success rate of about 98%.This method has broad application value in scenarios such as loading and unloading of mobile manipulators in intelligent production lines.
作者
刁世普
李兆澎
崔冠群
Diao Shipu;Li Zhaopeng;Cui Guanqun(School of Automation,Guangdong University of Technology,Guangzhou 510006,China;School of Mechatronic,Guangdong Polytechnic Normal University,Guangzhou 510665,China)
出处
《机电工程技术》
2023年第4期21-24,共4页
Mechanical & Electrical Engineering Technology
基金
广州市科技计划项目(编号:202102020508)
广东省基础与应用基础研究基金项目(编号:2021A1515110035)。
关键词
动态抓取
移动机械臂
局部视觉
特定工件
dynamic gripping
mobile manipulators
local vision
specific workpiece
作者简介
第一作者:刁世普 (1988-),男,河南邓州人,博士后, 研究领域为机器人视觉感知与运动规划,已发表论文 10余篇。
