A robot path planning algorithm based on reinforcement learning is proposed. The algorithm discretizes the information of obstacles around the mobile robot and the direction information of target points obtained by LiDAR into finite states, then reasonably designs the number of environment model and state space, and designs a continuous reward function, so that each action of the robot can be rewarded accordingly, which improves the algorithm and improves the training efficiency of the algorithm. Finally, the agent training simulation environment is built on gazebo, and the training results verify the effectiveness of the algorithm. At the same time, the navigation experiment is carried out on the actual robot. The experimental results show that the algorithm can also complete the navigation task in the real environment.
Loading....