후-측방충돌 회피를 위한 확률적 위험도 기반 점유영역 생성 알고리즘

Abstract

As Level 3 and higher autonomous vehicles become commercially availa ble, there is a growing interest in active safety systems that respond to u npredictable and dangerous situations. Therefore, based on this technologic al trend, this paper proposes a driving algorithm for responding to uncom mon but difficult-to-avoid side and rear collision situations. The proposed algorithm consists of three steps. First, the object is track ed through the Decoupled Tracking & Fusion method using mutual inform ation with the Ego vehicle. Second, to recognize the driving situation cons idering the path characteristics, the probabilistic coefficient of the direction al Time to Collision is applied. Third, generate an occupied region ahead based on the estimated collision risk to create a cubic path function based on the Risk Density. In the case of an abnormal rear-end merging of a neighboring vehicle, t he proposed algorithm generated risk-based occupancy zones and performe d evasive maneuvers. Therefore, we expect that the proposed algorithm pe rforms a reasonable risk assessment than previous one.