Automatic lane change data extraction from car data sequence

An automatic real driving data extraction method for lane change behavior is proposed in this paper which can efficiently detect the accurate start and end timestamp of lane change behaviors from long time driving data sequence. The objective of this work is to efficiently collect lane change data samples for behavior model building or intelligent ADAS system training. The proposed machine leaning based approach shows robustness against confusion from similar driving behaviors and results in highly accurate performance in extracting lane change behavior data segments in a fully automatic way.