SICE Annual Conference 2002, Osaka, Japan

Y.Fujiwara, T.Fujihira and S.Aadchi

Abstract

In order to improve the prevection of traffic accidents, it is necessary to add more intelligence to vehicles and to control the vehicles' mobility autonomously. However, to realize such functions, there are many problems to be solved, for example, recognition of vehicles' running environments. In this paper, the road shape is recognized by means of the feature values. The estimation method of the feature values based on sensor fusion is proposed, in which measurements from state-value sensors equipped in the vehicle and from GPS (gloval positioning system) are employed. As an estimation algorithm, H-infinity filter is applied. It is shown that the proposed method is effective than the conventional Kalman filter-based method from both computational and accuracy points of view.

Key words: Vehicle, active safety, H-infinity filter, sensor fusion.