Safety Standards for Autonomous Vehicles: Challenges and Way Forward

Abstract

This paper examines current autonomous vehicle (AV) safety standards and identifies gaps in their scope and depth. It is the first to present a thorough review and critique of such standards. It suggests that they lack the necessary guidance to ensure the safety of AVs, causing a bottleneck in transitioning from AV development to public adoption. Primary challenges of standards are presented, explicitly covering gaps and weaknesses pertaining to AVs’ user-system interactions, operational design domains (ODDs), safety testing methods, safety metrics, acceptance criteria, and machine learning pit- falls across the lifecycle of AVs. A way forward to address these challenges is also proposed, emphasising the need to decomplexify the ecosystem of AV standards and promote stake-holder interdependence. The paper concludes by recognising the criticality of legislative and regulatory frameworks for AVs and advocates for cohering safety standards, certification, and regulation.

Publication
In IEEE 26th International Conference on Intelligent Transportation Systems