모델 통합 기반 상대 차량 주행 경로 예측 알고리즘

Abstract

In this paper, the target vehicle trajectory prediction is proposed in order to improve the accuracy and robustness against the noise of the sensors. The physics-based model and maneuver-based model are composed and integrated to predict the trajectory accurately. First, the physics-based model is designed using the relative position information of the vehicle. Second, the maneuver-based model reflects the vehicle driving pattern, such as lane keeping or changing. The maneuver-based predicted model candidates are selected through a probabilistic approach. The integrated prediction algorithm is implemented through a physics-based model and a stochastically selected trajectory candidate. In particular, the proposed algorithm can only be designed by using the onboard sensor data, and it is validated by using the computer simulation software, CarSim and MATLAB/Simulink, and experimental test.