Scenario Understanding of the Drive Recorder Data by use of Speed and Acceleration Profiles
Rihito KIKUCHI, Yuki UENISHI, Kota Kikuchi, Tetsushi Mimuro · The Proceedings of the Transportation and Logistics Conference · 2016
Analysis using near-miss database, which is composed of drive recorder data, is expected to promote active safety researches to reduce traffic accidents. Automatic classification techniques for such big data have been eagerly studied. In the most of the preceding researches, they took notice of acceleration variation near the trigger time. For the better scenario understanding, it is important to observe the data of several seconds before the trigger. In this paper, two approaches are reported; firstly, smooth/rough road classification by power spectrum of vehicle’s vertical acceleration, and secondly, vehicle behavior pattern classification by vehicle’s speed profiles.