一种分层协同进化学习方法及其在机器人追踪中的应用

A Hierarchical Co-evolutionary Learning Method and Its Application in Robot Tracking

本文在分析了密歇根方法和匹茨堡方法的基础上,提出了一种新型分层协同进化学习方法。该方法由上述两类种群构成,此两类种群分别属于不同的智能层次,进行协同进化来实现智能,种群内部各自独立地采取不同的遗传操作,种群之间使用交互算法进行交流。实验表明:该方法能改善分类器系统的短视特性并提高其智能。

A new Hierarchical co-evolutionary learning method is proposed based on Michigan approach and Pitt approach. The two populations belong to deferent levels of intelligence. Deferent genetic operators are used independently in Michigan population and Pitt population via co-evolutionary method. and the interacting genetic operators are used inter-populations. Simulation experiments show that the methods can amend the short sight characteristic of the classifier system and improve its intelligence.