A path planning approach for multi-task AGV based on deep reinforcement learning and ant colony algorithm

With the increasing demand of distribution, it brings new challenges to the current algorithms of path optimization. Due to the effectiveness of the algorithm in complex scenes, there are still shortcomings in the logistics equipment path planning. In order to fill this gap, a negative reinforcement learning ant colony algorithm for multi-task AGV is proposed for path planning. Firstly, the path selection considering the material delivery time and AGV efficiency is analyzed. Secondly, the corresponding model assumptions are made for the path optimization problems under different tasks, and the negative reinforcement learning ant colony algorithm is used to make the optimal decision. Finally, according to the material distribution process of a cable workshop, the case is verified.