Autonomous Rolling Method of Intelligent Wheel Chair Using the Teaching and Q-learning

Abstract

In this paper, we propose an autonomous rolling method of intelligent wheel chair (IWC) using the teaching and Q-learning. In the proposed method, first, IWC learns a run path. Next, IWC corrects a position gap around the teaching point by using Q-learning. When detection of the teaching point goes wrong by the noise to motors, IWC cannot reach to the destination. The proposed method is applied to the task which runs a passage, and the simulation results show that IWC repeats a cutback and corrects a position gap. Furthermore, the simulation results show that IWC corrects a position gap more certainly by using the data learned in the environment with the noise to motors.