Sensmetry’s recent research paper “Self-driving car safety quantification via component-level analysis” (preprint available to download from arXiv) has been peer-reviewed, accepted, and is soon to be published by the SAE in a special safety-focused issue of its Journal on Automated and Connected Vehicles. In the paper Sensmetry presents a rigorous modular statistical methodology for quantitatively arguing safety or its insufficiency of an Automated Driving Systems (ADS). Through an illustrative example, the paper explains the importance of sufficient and necessary conditions at the component level for the overall safety of the vehicle. Establishing and demonstrating compliance with the derived component-level conditions can be used to prove or disprove safety at the vehicle in a cost-efficient manner. The article is cited in and forms a basis of Annex C.6 in the ISO/DIS 21448 SOTIF standard.
Sensmetry as a team is tightly focused on the development of methodologies and products for the numerical safety quantification of ADS, thereby providing clients with clear, established processes and tools by which their ADS products can be brought safely to the market. AV specific safety is addressed by the ISO 21448 SOTIF standard, to which Sensmetry is a key contributor.