Mobile robot path planning based on Q-learning algorithm

Nowadays, reinforcement learning is gaining more and more attention with AlphaGo's triumph. Path planning algorithm based on Q-learning, a model free reinforcement learning algorithm, was proposed for mobile robots. The algorithm translated sonar sensor information of the environment around the robot, the robot's pose, and the location of target points to finite states. Afterwards, a reasonable environment model and state space were constructed, with discrete rewarding functions being established. The experimental results validated the effectiveness of the proposed algorithm, as each action of the robot obtained the corresponding rewarding value, which improved the convergence efficiency of the proposed algorithm.