MOTION PATH PLANNING OF MOBILE ROBOT USING REINFORCEMENT LEARNING ALGORITHM

Abstract

Path planning has applications in many areas, for example, industrial robotics, autonomous systems, virtual prototyping, and computer-aided drug design. Much research has been done on path planning especially in the robotic field. This thesis presents a new framework for path planning using the Soft Actor-Critic (SAC) algorithm of reinforcement learning. The Soft Actor Critics Algorithms is an offpolicy Reinforcement Learning which uses the policy gradient method that incorporates the entropy measure of the policy into the reward to encourage exploration. The agent learns policy that acts as randomly as possible while still succeeding in performing a task. CoppeliaSim (formerly known as VREP) was used to simulate the environment. The result shows that the proposed SAC-based motion planning method is feasible and practicable in path planning and obstacle avoidance. It also shows that it is more efficient than Deep Deterministic Policy Gradient (DDPG) algorithm using the same hyperparameter.