A Study on Collision Prediction Methods of the Driving Assistance System of the Warning Type

摘要

This paper describes the influence of the collision prediction methods on the performance of the driving assistance system of the warning type. By using the autonomous microscopic traffic simulator constructed by authors, performance comparisons of the prediction method constraining the moving range in the lateral direction, and the prediction method that is not constrained are carried out; the performance of the direct collision judgment scheme and that of the successive collision judgment scheme are compared. It is shown that 1.5 seconds of prediction time, the constrained lateral moving range, and the direct judgment method are appropriate. Furthermore, the performance of the driving assistance system of the warning type using this prediction method and parameter described above have been evaluated by simulation on the total delay characteristics, the vehicle density characteristics, and the equipped ratio characteristics. The delay characteristics evaluation shows that this system improves the average accident interval about 5 or 6 times. The vehicle density characteristics evaluation clarifies that the effectiveness of this system increases with increasing accident frequency. The equipped ratio characteristics evaluation brings out that the average accident interval increases drastically in the region that the equipped ratio exceeds 70%.