Risk based regulation and certification of autonomous transport systems

Abstract

Autonomous transport systems in all modes—road (i.e. autonomous cars), aviation (i.e. drones), shipping and rail are coming. Regulation and testing are on-going in Norway. Risks of autonomous systems are uncertain due to missing data, emerging technology and variation in framework conditions. However, accidents of autonomous cars seem to be 1/3 or 1/2 of current levels. Incidents are different, needing outside interventions sometimes. Based on review of experiences across the modes and regulations, we suggest agile and transparent learning in the whole autonomous ecosystem, between all modes. System certification are needed, and system responsibilities must be clarified. Structures for orchestrating transport (i.e. control of many autonomous vehicles with possible common failures) and marking autonomous transport, should be established. In the interfaces between humans and systems there are differences in autonomy as imagined vs. performed, leading to new incidents and accidents. Emerging safety/security issues must be explored.