동역학 모델을 이용한 차선 예측기 개발

Abstract

Autonomous driving vehicle control system uses a vision sensor to get road information. Such vision sensor may provide poor detection performance due to different line intensities, and characters or imperfect lane marks on the road. In this paper, we propose a new lane estimator which uses a vehicle kinematics model. The estimator can provide predictive virtual lane even when vision system cannot provides reliable lane information due to poorly detected road images. The performance of the proposed control method is validated via simulations implemented with CarSim and Matlab/Simulink.