System health indicators in mixed criticality e/e systems in automated driving context

Abstract

One problem standing in the way of fully automated vehicles is the question of how to ensure vehicle safety and the safety of all traffic participants. Standards like ISO 26262 and ISO/PAS 21448 tackle those issues from different viewpoints by defining safety measures and mechanisms. While ISO 26262 focuses on safety hazards arising from malfunctioning of E/E systems, ISO/PAS 21448 stresses hazards due to technological limitations. However, it is an open challenge how system-wide safety can be monitored and validated at run-time. To complement those safety specifications we propose a system-wide run-time safety analysis. Our System Health Management concept is based on so-called Health Indicators (HIs) to propagate knowledge about detected errors and trigger appropriate error reactions. We analyze probable information sources to define meaningful HIs in automated driving context and investigate influence factors, of both ISO 26262 and ISO/PAS 21448. We apply our approach to a case study demonstrating its applicability in an automated driving scenario.