A Study on Reinforcement Learning Simulation for Multi-Agent Pursuit Problem Based on Relative Representaion of Sensory Input States

Abstract

In order to realize intelligent agent such as autonomous mobile robots, Reinforcement Learning is one of necessary techniques in behavior control system. However, applying the reinforcement learning to actual sized problem, the "curse of dimensionality" problem in partition of sensory states should be avoided maintaining computational efficiency. In multi-agent reinforcement learning, the problem is emerged owing to the high dimensionality of each agent states. In this study, we evaluate the learning performance of agent that represents the input states as relative expressions through numerical experiments of pursuit problem.