Motion Control of a Cable-Driven Parallel Robot Using Reinforcement Learning Deep Deterministic Policy Gradient Multi - Agent

Abstract

Machine learning has demonstrated a great capacity to find solutions for various research in the control and robotic realm. Reinforcement learning is a branch of machine learning which can interact with the environment while searching for desirable scenarios and eliminating the undesirable ones in controlling the robot. Though many studies have shown the significant performance of machine learning over the classic control methods, there are many unexplored strategies for improving the performance of the overall controller in machine learning, especially reinforcement learning. This paper provides a case study on the efficient control of a cable-driven parallel robot using the deep deterministic policy gradient multiagent and the possibility of transfer learning in this context.