基于神经网络的Q学习在Khepera Ⅱ机器人避障中的应用

Neural Network-based Q-learning Applied to Obstacle Avoidance in Khepera Ⅱ Mobile Robots

为了提高移动机器人的自主学习能力,在传统的机器人行为控制结构基础上设计了智能控制结构,同时引入了基于神经网络的Q学习模块算法,克服了传统算法只能应用到离散状态中的不足。移动机器人的避障实验结果表明,该方法能够使移动机器人通过自学习实现自主避障。

An intelligent control architecture was designed based on a behavior architecture to improve the learning ability of mobile robots. At the same time, a Q-learning algorithm based on a neural network was led into the intelligent control architecture,while normal algorithm can only be applied to discrete states. Experiments of obstacle avoidance show that the mobile robot can learn to avoid obstacles with the algo- rithm.