A Research on Manipulator-Path Tracking Based on Deep Reinforcement Learning

The continuous path of a manipulator is often discretized into series independent action poses during tracking, and the inverse kinematic solution manipulator’s computationally challenging yields inconsistent results. This research suggests manipulator-route-tracking method employing deep-reinforcement-learning techniques to deal with this problem. paper takes an end-to-end-learning approach for closed-loop control eliminates process obtaining answer by converting path-tracking task sequence-decision issue. first explores feasibility deep reinforcement learning in tracking manipulator. After verifying feasibility, multi-degree-of-freedom (multi-DOF) was performed combining maximum-entropy algorithm. experimental findings demonstrate that performs well manipulator-path avoids need dynamics model, capable performing manipulator-tracking space. As result, proposes presented great significance on tracking.