Design Integration of Adaptive Safety Systems in Modern Passenger Cars

Abstract

Modern passenger cars are undergoing design integration of adaptive safety systems, which are redefining the nature of automotive safety as no longer being a series of individual instances of feature enhancement, but rather a context-based orchestration of safety. This paper summarizes the interrelationships between sensing and fusion, risk and severity prediction, decision logic, and coupled actuation between crash avoidance and injury mitigation. The focus is put on the effects that uncertainty-conscious triggering, driver and occupant state prediction, and the impact of pre-crash maneuver on the occupancy kinematics have on the effectiveness of adaptive restraint measures. It is also argued in the review that an evaluation method requires real-world effectiveness evidence coupled with scenario-based testing and injury-relevant measures, which require both consistency in the coverage of scenarios between traffic conflicts and occupant outcomes. Some of the challenges involve confirmation of AI-enabled perception when operational variability occurs, interaction of longitudinal/lateral control and restrain performance, and traceability of evidence between simulation and real crash results. The paper comes to a close and identifies future directions towards closed-loop co-design of safety with a dedicated emphasis on posture and occupant-aware protection and scalable validation methods offering plausible safety case to proceed with production deployment.