A Reinforcement Learning Method of Obstacle Avoidance for Industrial Mobile Vehicles in Unknown Environments Using Neural Network

摘要

This paper presents a reinforcement learning method for a mobile vehicle to navigate autonomously in an unknown environment. Q-learning algorithm is a model-free reinforcement learning technique and is applied to realize the robot self-learning ability. The state-action Q-values are traditionally stored in a Q table, which will decrease the learning speed when large storage memory is needed. The neural network has a strong ability to deal with large-scale state spaces. Therefore, the neural network is introduced to work with Q-learning to ensure the self-learning efficiency of avoiding obstacles for industrial vehicles in unpredictable environments. Experiment results show that an autonomous mobile vehicle using the proposed method can successfully navigate to the target place without colliding with obstacles, and hence prove the self-learning ability of navigation in an unknown environment.