Impedance Control of Space Manipulator Based on Deep Reinforcement Learning

The dynamic characteristics of space manipulator system are studied and the dynamic equations are given. The control method of the system is designed, and the double loop PID controller of the base and each joint and the feedforward force/moment controller of the base are given. To solve the problem of end-effector contact force control, a position-based impedance control method was proposed, and particle swarm optimization algorithm was introduced to optimize the impedance parameters, and deep reinforcement learning was combined to improve the impedance control effect. The rapid prototyping control is studied and a rapid prototyping control platform is built based on the air floating platform.