多类型障碍物环境下机器人机械臂避障轨迹自动规划研究  

Research on Automatic Planning of Obstacle A voidance Trajectory of Robot Manipulator in Multi-type Obstacle Environment

作　　者：黄骏 HUANG Jun(Wuhu Institute of Technology,Wuhu Anhui 241006,China)

机构地区：[1]芜湖职业技术学院,安徽芜湖241006

出　　处：《佳木斯大学学报（自然科学版）》2024年第7期73-76,共4页Journal of Jiamusi University：Natural Science Edition

基　　金：安徽省高等学校科学研究重点项目(2022AH040298);芜湖职业技术学院校级科学研究自然科学重点项目(wzyzrzd202409);安徽省省级质量工程课程思政示范课程(2021kcszsfkc452)。

摘　　要：针对机器人机械臂在多类型障碍物环境下的避障轨迹规划问题,提出一种通过深度学习进行机器人机械臂避障轨迹自动规划的方法。通过D-H参数法进行机器人机械臂的运动轨迹建模,并结合多障碍物环境进行碰撞检测分析。将多类型不规则障碍物归一化为圆球形状便于计算,通过马尔科夫决策以及Q算法进行深度强化学习,实现机器人机械臂避障轨迹的自动规划。实验结果表明,研究方法迭代收敛速度较快,自动规划避障轨迹步数更少。Aiming at the obstacle avoidance trajectory planning problem of the robot manipulator in the environment of multiple types of obstacles,a method for the automatic obstacle avoidance trajec-tory planning of the robot manipulator through deep learning is proposed.The motion trajectory of the robot manipulator was modeled by the D-H parameter method,and the collision detection analysis was carried out in combination with the multi-obstacle environment.Multiple types of irregular obstacles are normalized into spherical shapes for easy calculation,and deep reinforcement learning is carried out through Markov decision and Q algorithm to realize the automatic planning of the obstacle avoidance trajectory of the robot manipulator.Experimental results show that the iterative convergence speed of the research method is fast,and the number of steps of automatic obstacle avoidance trajectory is less.

关 键 词：六自由度机械臂 避障轨迹规划 D-H法建模 马尔科夫决策 避障距离 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置] TH122[自动化与计算机技术—控制科学与工程]