内部动机驱动的机器人未知环境在线自主学习

Internal motivation driven robot online and autonomous learning of unknown environment

针对目前智能移动机器人在未知环境中学习遇到的如学习主动性、实时性差,无法在线积累学习的知识和经验等问题,受心理学中内部动机的启发,提出一种内部动机驱动的移动机器人未知环境在线自主学习方法,在一定程度上弥补目前该领域存在的一些问题。该方法通过在移动机器人Q学习的框架下,将奖励机制用基于心理学启发的内部动机取代,提高其对于未知环境的学习主动性,同时,采用增量自组织神经网络代替经典Q学习中的查找表,实现输入输出空间的映射,使得机器人能够在线增量地学习未知环境。实验结果表明,通过内部动机驱动的方法,移动机器人对于未知环境的学习主动性得到了提高,智能程度有了明显改进。

In order to solve the problems in intelligent mobile robot learning in unknown environment such as low activity and real-time performance, unable to accumulate learning experience. Inspired by internal motivation in psychology, a method called internal motivation driven mobile robot unknown environment learning algorithm is proposed to make up the problems above to some extent. The proposed method architecture is Q-learning, and the reward is modified by the internal motivation inspired by psychology, to raise the performance of learning activity. The look-up table in Q-leaning is also replaced by incremental self-organized network, to realize the map from input to output. So the robot is able to incre-mentally learning unknown environment in a online manner. The experiment result shows that by the internal motivation driven method, the learning activity for learning the unknown environment is raised, the intelligence of the robot is also improved.