Multi-agent cooperative strategy learning method based on transfer Learning.

Path planning is of great significance for industrial production and daily life, where various intelligent methods have been applied to it. Reinforcement learning (RL), as a popular area of machine learning, can effectively find the optimal path by interacting with the environment. Q-learning is one of the most popular algorithms among the reinforcement learning methods. Although Q-learning is simple and effective, its training time will rise sharply when the environment to be explored becomes complex. To solve this problem, this paper proposes a multi-agent cooperative strategy learning method by combining transfer learning (TL) and traditional Q-learning. The proposed method is applied to the maze game and the experiments show that the proposed method outperforms the existing Q-learning methods in both iterations and performance.