A Review of Conceptualizations of Safety and Risk in Current Automated Driving Regulation

TL;DR

The paper addresses how divergent definitions of safety and risk in automated driving regulation can impede market deployment. It compares UN Reg. 157, EU 2022/1426, German AFGBV/StVG, and the UK AV Act to assess their underlying notions of safety, risk, hazard sources, and harms. The authors conclude that UN/EU and the UK adopt predominantly relative, risk-based concepts focused on absence of unreasonable risk, with the UK showing the most consistent approach, while German regulation presents significant ambiguity and potential legal uncertainty. They propose clarifying definitions, embracing broader stakeholder values, and documenting tradeoffs to support timely and responsible market introduction of automated driving systems.

Abstract

"Safety" and "Risk" are key concepts for the design and development of automated vehicles. For the market introduction or large-scale field tests, both concepts are not only relevant for engineers developing the vehicles, but for all stakeholders (e.g., regulators, lawyers, or the general public) who have stakes in the technology. In the communication between stakeholder groups, common notions of these abstract concepts are key for efficient communication and setting mutual expectations. In the European market, automated vehicles require Europe-wide type approval or at least operating permits in the individual states. For this, a central means of communication between regulators and engineers are regulatory documents. Flawed terminology regarding the safety expectations for automated vehicles can unnecessarily complicate relations between regulators and manufacturers, and thus hinder the introduction of the technology. In this paper, we review relevant documents at the UN- and EU-level, for the UK, and Germany regarding their (implied) notions of safety and risk. We contrast the regulatory notions with established and more recently developing notions of safety and risk in the field of automated driving. Based on the analysis, we provide recommendations on how explicit definitions of safety and risk in regulatory documents can support rather than hinder the market introduction of automated vehicles.

Figures (1)

• Figure 1: Graphical summary of a taxonomy of risk related to automated vehicles, extended based on ISO 21448 (iso21448) and favaro2023. Top: Causal chain from hazard sources to actual harm; bottom: summary of the individual elements' contributions to a resulting risk. Graphic translated from nolte2024