MRDDPG Algorithms for Path Planning of Free-Floating Space Robot

Abstract

In the path planning for manipulation of a free-floating space robot, the existing methods have difficulties in many constraints, poor adaptability, and other difficulties. For this reason, a path planning algorithm for free-floating space robots based on deep reinforcement learning is proposed. Firstly, considering the main constraints during path planning, three constraints, including safety performance, coupling disturbance and path length, are selected as multi-constraint criteria. And multi-constraint model for path planning is built. Then, the Multi-constrained Reward Function for processing the above constraints is designed, and the MRDDPG algorithm for solving the path planning model is proposed. Finally, the performance of the method is analyzed by simulation experiments. The result shows that the method can effectively complete the intelligent path planning task of free-floating space robot.