An Automatic Labeling Scheme of Lane-Changing Maneuvers for Vehicle Trajectory Data Sets

Labeling out lane-changing maneuvers is a vital prerequisite to further potential research on vehicle behaviors in trajectories. However, there is no general and appropriate scheme to eliminate this obstacle. An automatic labeling scheme is, therefore, proposed in this paper to fill this gap, which can identify the starting and ending points of a consecutive lane-changing maneuver. This paper first distinguishes lane-changing scenarios with "single" and "multiple" and then designs a uniform algorithm for automatic labeling that merely relies on vehicle coordinates and time stamps. Furthermore, the optimal parameter group of the algorithm is found, and a comparative analysis is conducted to validate the applicability of the proposed scheme. The results show that there are no error labels in the proposed scheme, such as discontinuity, mislabeling, over-labeling, and missed labeling. The proposed automatic labeling scheme for lane-changing maneuvers has high accuracy.