High Definition Map Assisted Vehicle Tracking with Lane Constraint

Vehicle target tracking is the key of auxiliary driving system and unmanned driving system. Restricted by traffic rules, vehicles on a multi-lane road always try to drive in the middle of the lane. In order to use this prior information to improve the performance of vehicle target tracking, a High Definition Map assisted vehicle tracking with lane constraint is presented. By the method, regarding the {Unction of lane as a proportional-derivative controller, a new vehicle motion model considering the affection of the lane is proposed. The measurement is transformed into the curvilinear-coordinate system, and the state estimation of the target is implemented in such coordinate system also. Furthermore, to evaluate the influence of adjacent lanes to the vehicle, a DS-VSMM filter is designed to identify whether the vehicle is changing lane and which lane the vehicle would change into. Simulation results show that the proposed algorithm improved the vehicle tracking accuracy, reduced the time consumption, and effectively identified the lane change behavior of target vehicle.