Autonomous acquisition of fuzzy rules for mobile robot control: first results from two evolutionary computation approaches

摘要

We describe two architectures that autonomously acquire fuzzy control rules to provide reactive behavioural competencies in a simulated mobile robotics application. One architecture is a Pittsburgh-style Fuzzy Classifier System (Pitt1). The other architecture is a Michigan-style Fuzzy Classifier System (Mich1). We tested the architectures on their ability to acquire an investigative obstacle avoidance competency. We found that Mich1 implemented a more local incremental search than the other architecture. In simpler environments Mich1 was typically able to find adequate solutions with significantly fewer fitness evaluations. Since fitness evaluation can be very time consuming in this application, it could be a strong positive factor. However, when the rule set must implement a competency in more complex environments, the situation is somewhat different. The superior ability of Pitt1 to retain a number of schema in the population during the process of optimisation, is then a crucial strength.