Robot Reinforcement Learning for Automatically Avoiding a Dynamic Obstacle in a Virtual Environment

Abstract

In a virtual environment, a robot can serve people by bringing things to them. However, when a robot moves within a house, it collides with a dynamic obstacle. These collisions make it difficult for a robot to complete its mission. We therefore apply reinforcement learning to the robot to make it more intelligent. Consequently, the robot can automatically move to avoid the dynamic obstacle in order to successfully complete its mission.