Automatic learning of Action Knowledge-Base for a mobile robot using genetic algorithms

Abstract

To design behaviors of a mobile robot for realizing given tasks, a designer has to make a set of rules which generates a proper action from a state of sensors. In general, however, it is difficult for the designer to make the complete set of rules since the number of rules is very large and the proper action for a given state of sensors is not clear. Therefore, the robot must learn and construct the knowledge base of actions by itself. This paper proposes a learning algorithm to construct the knowledge of action in order to achieve tasks that are given to the mobile robot. The action to achieve a task in an environment is generated by a genetic algorithm. It is also shown that repeating the knowledge extraction will make the construction of the Action Knowledge-Base possible, concerning the task in any situation.