RL-ART2 Neural Network Based Mobile Robot Path Planning

The paper proposes a reinforcement learning based ART2 neural network (RL-ART2) and its learning algorithm. ART2 is used to store abundant classified patterns and state space. Facing large classified patterns, it´s hard to evaluate and select a classified pattern by hand, so the paper imports evaluating and selecting mechanism of reinforcement learning into ART2 for solving how to evaluate and select the classified pattern, and uses RL-ART2 to propose collision avoidance system RLART2-CAS in the research of path planning of mobile robot. The simulation experiment indicates that the collision times between robot and obstacle is effectively decreased. The RL-ART2 makes favorable result of path planning