On Local Path Planning for the Mobile Robot based on QL Algorithm

This paper proposes a local path planning method for the mobile robot based on the Q-Learning (QL) algorithm to improve the common problems existed in the traditional method, such as the slow convergence rate, the dilemma between exploration and exploitation, and the complex obstacles in the workplace. First, the variables of state and action are designed and discretized according to the path planning task. Then a Q-value function matrix is used to store the reinforcement value, and a reward function is constructed according to the requirements of obstacle avoidance and shortest path. To solve the balance problem of exploration and exploitation and improve the convergence speed, the ε-balance strategies and the selection algorithm of actions are designed to improve the learning process of QL. After training of QL, the optimal pairs of state and action are obtained, further the optimal control rules are achieved and used to perform local path planning. In order to prevent the incomplete visiting problem for the pairs of state and action, the steering rules are appended to improve the efficiency of the path planning. Finally, the designed method has been simulated. The results show that the robot can plan an optimal or sub-optimal path while avoiding obstacles, even in a complicated environment.