ITS16922015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 ITS-1692 The detection of black-spot intersections by using event-data-recorder data Kentaro Kondo Fujitsu Ltd., Japan Abstract A large volume of data from event data recorders (EDRs) for motor vehicles was used to detect suspected black-spot intersections on roads. The data were analyzed to extract the braking positions and directions of travel when drivers braked suddenly. By combining them with the reliable positions of intersections obtained from road network dataset, it was possible to accurately reconstruct the geometry of the position and direction of travel of a vehicle in relation to an intersection when sudden braking occurred. The identification of black-spot intersections and the existence of blind corners were validated by visually checking the video records from EDR data. Keywords: event data recorder, black-spot intersection, accelerometer, road network dataset. Introduction Event data recorders (EDRs) for motor vehicles are installed in half of the approximately 250,000 taxis in Japan. The main purpose of EDRs is to record information when vehicle accidents occur. EDR data include measurements from accelerometers, GPS receivers, on-off signals from brake pedals, and the video recordings ahead of a vehicle. EDRs, however, record a large number of events of slight but sudden braking. These events are not regarded as accidents, but as a driver's automatic response to the unexpected emergence of objects on the road ahead. These events must imply the existence of a blind corner at the side of the road. Dangerous blind corners are often found at junctions known as black-spot intersections.