10ITSWCPaperNaturalisticDrivingBehavioratIntersections1智能交通世界大会ITS智慧城市社区人工智能AI物联网IT报告课件教案.pdfVIP

22nd ITS World Congress, Bordeaux, France, 5–9 October 2015 Paper number ITS-2634 Naturalistic driving behaviour at inner-city intersections 1 1 2 2 T. Streubel *, L. Rittger , K. H. Hoffmann , J. F. Krems * thomas.streubel@, phone: 06142-773712 1. Adam Opel AG, IPC S4-01, D-65423 Ruesselsheim, Germany 2. Chemnitz University of Technology, D-09107 Chemnitz, Germany Abstract We report findings on driving behaviour in an exploratory naturalistic driving study that was conducted on urban roads in Ruesselsheim, Germany. Our goal was to infer driving behaviour when approaching intersections. Therefore, the start of braking and turn indicator activation was examined in relation to the driving direction (straight, left, right), intersection layout (T-junctions, X-intersections) and right of way rules (priority to right, yield way, stop). Our data showed that the brake onset for right turns is distinguishable from left turns and driving straight. In case of yield and stop intersections, we showed that brake onset could neither distinguish between directions nor right of way rules. However, natural braking behaviour was observable. The results will help to determine the number of probabilistic models that predict the driver behaviour in intersection scenarios. This enables driver assistance systems to estimate the criticality of upcoming driving situations. Keywords: naturalistic driving study, intersection behaviour. Introduction Intersections are a major hot spot for accidents on urban roads. The German traffic accident analysis states that about half (47.5%) of the traffic accidents with injuries in inner-cities take place at intersectio