摘要
针对人机共融场景中机器人任务快速变化、紧凑作业空间等导致机器人高性能运动控制困难的问题,提出一种基于环境吸引域的机器人运动控制与实时位型优化方法。为尽可能减小机器人本体的运动区域,采用两组关键点集对机器人本体与人类偏好的机器人作业区域进行抽象化表征,并提出基于点集距离的环境吸引域评价指标;设计笛卡尔空间运动控制器,并构造末端执行器跟踪误差收敛的关节角速度等式约束;结合关节角度、角速度等物理约束,建立基于约束-优化的机器人运动控制与位型优化问题模型。设计递归神经网络对机器人的角速度指令进行实时求解,并证明了系统的稳定性。最后通过四自由度机器人和七自由度Franka Panda机器人的仿真,验证了所提算法在保证机器人末端完成高精度运动控制的同时,能够使机器人的工作区域尽可能收缩到预设区域。
Aiming at the difficulties of high-performance robot motion control caused by the challenges such as rapid change of robot tasks and compact working space in man-machine integration scenarios,a robot motion control and real-time configuration opti-mization method based on environment attraction domain was proposed.In order to reduce the moving area of the robot body as much as possible,two sets of key points were used to characterize the robot body and the robot operating area preferred by humans,and an eval-uation index of the environment attraction area based on the distance of the point set was proposed.The Cartesian space motion control-ler was designed,and the joint angular velocity equality constraint of the end effector tracking error convergence was deduced.Com-bined with physical constraints such as joint angle and angular velocity,a constrained optimization model for robot motion control and configuration optimization was constructed.A recurrent neural network was designed to solve the angular velocity instruction of the robot in real time,and the stability of the system was proved.Finally,the simulation results on 7-DOF Franka Panda robot show that the proposed algorithm can ensure the convergence of the end-effector's tracking error,and shrinking the robot's working area as much as possible to the pre-defined area.
作者
吕晓静
徐智浩
徐恩华
LYU Xiaojing;XU Zhihao;XU Enhua(School of Aircraft Maintenance Engineering,Guangzhou Civil Aviation College,Guangzhou Guangdong 510403,China;Guangdong Key Laboratory of Modern Control Technology,Institute of Intelligence Manufacturing,GDAS,Guangzhou Guangdong 510070,China;School of Civil Aviation Management,Guangzhou Civil Aviation College,Guangzhou Guangdong 510403,China)
出处
《机床与液压》
北大核心
2023年第23期37-42,共6页
Machine Tool & Hydraulics
基金
国家自然科学基金项目(62003102)
广东省普通高校青年创新人才项目(2020KQNCX143
2019GKTSCX007)
广州市科技计划项目(202102020860)。
关键词
环境吸引域
模型驱动神经网络
运动控制
Environmental attractive domain
Model-driven neural network
Kinematic control