자율주행 시스템 검증 시나리오를 위한 Reference Governor 기반의 차량 추종 모델 개발

Abstract

This paper represents a reference governor-based, car-following model for an autonomous driving system verification scenario. A reference governor(RG) is an add-on control scheme that enforces state and control constraints on stabilized systems by modifying the reference whenever the constraints are violated. Unlike existing controllers that do not handle states and input constraints, the proposed method handles constraints by using a maximal invariant that is set to consider actuator limits, ride quality, and vehicle safety with low computational effort. The proposed method was validated with simulation under various driving scenarios, such as acceleration, deceleration, and driving cycles. The RG-based, car-following model showed efficient computational performance while satisfying the defined constraints, and was able to simulate actual traffic flow.