Automatic Behaviour Adaption for Mobile Robots with Diierent Kinematics

Industrial applications of mobile robots require the same behaviour for diierent types of autonomous robots. If this at a rst glance seems easy, it is in fact a rather complicated task as it means to adapt a system with a diierent mechanical model and thus diierent behaviours to a prescribed reaction to the environment. If in addition this goal should be reached automatically the complexity of the problem increases. In this paper we present an adaptive learning method-mother-child teach-in-through which from a given well tuned prototype a new rule base for a mobile platform with diierent kinematics is generated automatically. At the end of the teaching phase, the new platform shows the same behaviour as the prototype.