Road Departure Detection Using State Diagrams and Statistical Hypothesis Tests in Map Matching
Toshiyuki Aoki, Mikio Bando, Tomoaki Hiruta, Koichi Kato, Akihiro Kawabata, Zhixing Liu
Car-navigation systems display the car’s position on a link, which is a line representing a road in map data, when the car is running on a road and display its estimated position when it is running in a parking area. Therefore, it is necessary for map matching to determine whether a car is running on a road. The errors of the car’s estimated position and the link position cause incorrect detection when the car is running in a parking area parallel to the road. This paper proposes a road departure detection method using state diagrams and statistical hypothesis tests. The statistic hypothesis tests make detections considering the previous errors, and the state diagrams change detection conditions in each state. This method can therefore decrease the numberof incorrect detections that are made. The performance of this method was experimentally evaluated using observation data collected with a car, and the conditions under which this method could detect road departure were shown.