The explication of characteristics of a driver is a basic theme in the field of vehicle engineering, and many studies have already been carried out. But characteristic of man is much more complicated than that of machine and characteristics described by one method could only express one aspect of a driver. Therefore, it is necessary not only to deepen our studies of the former methods but also to look for new ones.
In this study, Neural Network System(NNS) is applied to a driver model. NNS is a parallel processing system which was designed as one type of the models that simulate the information processing in the brain. One of the advantages of this model is that the system itself is generated automatically according to the combination of input and output data.
The purpose of this study is to propose an example for the basic method of the construction of a driver model using NNS and to show the application of it. First, a driver model is obtained from a simple driving maneuver and the analysis of it is tried. Next, this model is applied to the driving along general roads, and it is shown that the model has learned the essential characteristic which can be applied to other driving maneuvers. And it is proved that more applicable [Illegible]el can be obtained by learning from several driving [Illegible]rs together.
T. Fujioka, K. Nomura, N. Takubo, D. H. Chung
The University of Tokyo, JAPAN
International Pacific Conference On Automotive Engineering