Live State-of-Health Safety Monitoring for Safety-Critical Automotive Systems

Abstract

Autonomously driving vehicles require higher safety and reliability standards than traditional human-driven vehicles as they need to be able to handle safety-critical situations on their own. Therefore, these systems needs to demonstrate fail-operational behavior to ensure safety of the passengers by basic car controls. Especially silent failures of semiconductor devices can be critical from a safety point of view. Semiconductor devices fail abruptly and cannot be detected in advance. This paper presents a novel sensor approach to detect those kind of silent failures ahead of time and to ensure safety for future advanced driver-assistance systems (ADAS) such as LiDAR (Light Detection and Ranging). We have evaluated the design of our novel sensor concept in SystemC which will be implemented in a LiDAR system to mitigate silent failures as well as enable dynamic safety contracts.