关键词: 协作机器人, ＲＢＦ神经网络, 滑模控制, 误差补偿

Abstract: As intelligent operating assistants, collaborative robots have opened up a wide range of application scenarios in fields of industry, service and medical treatment. In order to explore the dynamic characteristics of collaborative robots and improve the accuracy of terminal positioning, this paper takes the KUKA LBR lightweight humanoid arm collaborative robot model as an example for research. Based on the RBF neural network and the synovial control algorithm, the dynamic control strategy of the collaborative robot is designed and the dynamic characteristics and end position error are analyzed. Parameter identification and error compensation of collaborative robots are carried out based on the Levenberg-Marquardt nonlinear damping least squares algorithm. The ADAMS-Matlab joint simulation shows that the dynamic control effect of a sliding mode controller based on the RBF neural network is better. The average terminal error under an extreme working condition is about 4.7 mm, which is mainly due to the influence of gravity load. After variable parameter error compensation, the average terminal error is less than 0.2 mm, which effectively improves the position accuracy and provides a theoretical basis for the research of collaborative robot control and error compensation.