부산대학교 도서관

For a swarm of mobile robots in an unknown environment, the most challenging task is to plan the optimal path and also to learn the environmental parameters. Machine learning is one of the solutions to this problem. This paper proposes a combination of particle swarm optimization (PSO) and Q-value based reinforcement learning (Q-Learning) for a swarm of mobile robots to find the optimal path in an unknown environment and to learn the environment. Q-learning combined with PSO enable the robots to learn the unknown environment with reward and action selection policy while reducing the time to find the optimal path in the environment using the iterative improvement method of PSO. The proposed algorithm is simulated and found performing faster than Q-learning and PSO performing alone. We also compared with some other established algorithm where the proposed algorithm outperforms each of them in accuracy and speed.