Exploring Metrics for Safety & Reliability Assessment in an Automotive Emergency Braking Perception System: A Case Study

Abstract

Assessing the safety and reliability of automotive software systems is challenging, along with knowing which metrics to use to verify intended functional and non-functional requirements. Additionally, the lack of context definitions and ambiguities in metrics, are further issues found in the automotive supply chain. This study presents findings on exploring and comparing common safety and reliability metrics in literature and their applications in an automotive perception system under implementation. It also investigates how targets set in safety and reliability metrics are adapted in the development of perception systems in a practical multi-party development scenario. Facilitating Multi-Party Engineering of Requirements (FAMER), is a collective research group of automotive companies and research institutes in Sweden. A multi-methodological approach was applied in this research study, involving a literature review, followed by a case study on the SMIRK prototype, and a focus group with representatives from a Research Institute, Original Equipment Manufacturer (OEM), Tier 1, and Tier 2 automotive companies in FAMER.