Learning algorithm of environmental recognition in driving vehicle

We consider the problem of recognizing driving environments of a vehicle by using the information obtained from some sensors of the vehicle. Previously, we presented recognition algorithms based on a usual method of pattern matching using the distance on a vector space and fuzzy reasoning. These algorithms can not be applied to meet the demands of nonstandard drivers and changes of vehicle properties, because the standard pattern or membership function for the pattern matching is always fixed. Thus to cover such weakness we present adaptive recognition algorithms with adaptive change of the standard pattern and membership function. In this work, we put forward a fuzzy supervisor in the learning process. Also, we present an algorithm into which a new learning method is introduced to improve the performance of the previous ones and to meet the above demands.<
>