AGV Path Planning Based on Reinforcement Learning: A Review

The problem of path planning has always been one of the main problems of automatic guided vehicles to make autonomous decisions, in the intelligent production of automobiles, AGV can accurately transport the assembly parts of the car to the specified location, and can realize that the AGV and other working machines in the workshop do not conflict. At present, with the rapid development of the use of reinforcement learning to realize the autonomous decision-making of agents, how to apply the reinforcement learning algorithm to the path planning of AGV is a hot topic in today’s research. This paper mainly focuses on the combination of evolutionary algorithms and model-free reinforcement learning algorithms to solve path planning problems, first briefly introduces the application of evolutionary algorithms and reinforcement learning algorithms in path planning, and then elaborates how evolutionary algorithms are applied to different structures of reinforcement learning, and compares and analyzes these algorithms.