Motion Planning In Human-Robot Collaboration And Adaptive Control For A Mobile Manipulator

Collaboration between robots and humans can give full play to their respective advantages and improve the work efficiency and comfort.At present,human-robot col-laboration has been widely applied in people's production and life,and the requirements for the coordination and safety of human-robot collaboration are increasing.Therefore,it is of great significance to study the motion generation and control of robots in human-robot collaboration.In this dissertation,appropriate robot motion planning methods in human-robot collaboration are proposed.The robot can generate corresponding motions according to human motions during the execution of tasks,so that the coordination ca-pability of the robot and the safety of human can be improved.An adaptive controller for the mobile manipulator is designed to control the robot to accurately track the de-sired trajectories and complete collaborative tasks.The main research contents of the dissertation are as follows:(1)Considering the uncertainty in human-robot collaboration,the Long Short-Term Memory network is used to model the overall system of the human and the robot.The training data is collected through human demonstrations,from which the model can learn the correlating information of human-robot motions.During the online phase,the learned network can be used to generate appropriate joint motions of the robot in real time according to the observed human motions.Compared with traditional pre-programming methods,this method is not task oriented and reduces the dependence on pre-programming.The adaptability of the robot's motions to human movements and the naturalness of the interactive motions are fully considered in the model,so that the autonomy and coordination capability of the robot in human-robot collaboration can be improved.(2)An adaptive control method for a mobile manipulator is designed to assure the tracking performance.This method is based on the dynamic model of the overall system consisted of the mobile platform and the manipulator.The neural network is adopted to approximate the unknown parameters of the model.And the Lyapunov direct method is used to analyze the stability of the system.(3)Aiming at the safety in human-robot collaboration,a motion planning method based on trajectory reshaping is proposed.The Gaussian Mixture Model is used to predict the human motions.Then the distance between the human and the robot is con-verted into a virtual force acting on the end of the manipulator.The collision-avoidance problem is converted into a problem about robot trajectory reshaping considering the interaction force.On the basis of energy optimization,the neural network is adopted to obtain the optimal trajectory.The proposed method can reduce the risk of collisions be-tween the human and the robot by replanning the trajectory segment where the collision may occur.The optimal trajectory can be acquired quickly.And the robot is allowed to determine the degree of trajectory deformation according to the actual.situation,then the energy consumption can be reduced in the process of motion generation.