An approach to managing the operational safety of autonomous vehicle trials

Abstract

This paper summarises the methodologies that HORIBA MIRA have developed to ensure acceptable safety mitigations are identified and applied within research projects involving the physical trialling of autonomous vehicles. Two complimentary approaches to risk assessment are advocated: a risk matrix to provide a qualitative scoring of hazards, which has been developed specifically for the challenges of autonomous vehicle trials and therefore differs significantly from ‘traditional’ risk matrix approaches, and a GAMAB (Globalement Au Moins Aussi Bon, or ‘globally at least as good’) approach, which compares how the most prevalent hazards for manually-driven vehicles will be affected by the presence of an autonomous vehicle. In addition to being applicable to autonomous vehicle research projects, the methodologies presented are applicable to Advanced Driver Assistance Systems on production vehicles, and indeed to any risk assessments concerned with changes to safety-critical systems that involve human interaction.