基于深度学习的清洁机器人机械手操作运动规划研究  

Research on Motion Planning of Cleaning Robot Manipulator Based on Deep Learning

作　　者：毕海婷[1] 梁入文 付龙海[1] BI Haiting;LIANG Ruwen;FU Longhai(Yantai Vocational College,Yantai Shandong 264670,China;Yantai University,Yantai Shandong 264005,China)

机构地区：[1]烟台职业学院,山东烟台264670 [2]烟台大学,山东烟台264005

出　　处：《长春工程学院学报（自然科学版）》2024年第4期84-91,共8页Journal of Changchun Institute of Technology：Natural Sciences Edition

基　　金：山东省教育教学改革研究项目(2015042)。

摘　　要：针对清洁机器人因机械手除尘动作规划不可预测而导致的机械臂运动规划困难的问题,采用高层次规划的方法对家庭环境中的表面灰尘进行清洁,提出了一种协作机械手操作清洁运动规划方法。借助机械手有效操作动作序列来进行时间规划,通过更新的信息重新规划,充分考虑不同擦布器或清洁工具在规划时间内的计划与质量。试验结果表明,深度学习有效改变了清洁机器人机械手的抓取方式与拖拽行为,使得清洁机器人机械手的操作运动规划性能更佳,尤其是在规划完成复杂清洁任务时更具优势。In order to solve the problem of the difficulty in the motion planning of the manipulator due to the unpredictable of the dust removal action planning of the cleaning robot,a high-level planning method is used to clean the surface dust at home,and a motion planning method for collaborative manipulator operation cleaning is proposed.By utilizing the effective operation sequence of the manipulator for time planning,it replannings with updated information,fully considering the planning and quality of different wipers or cleaning tools within the planned time.The experimental results show that deep learning effectively changes the grasping and dragging behavior of the cleaning robot manipulator,making the operation motion planning performance of the cleaning robot manipulator better,especially advantageous in planning and completing complex cleaning tasks.

关 键 词：机械手 表面清洁 深度学习 概率规划 清洁操作 

分 类 号：TP242.3[自动化与计算机技术—检测技术与自动化装置]