Modelling Safety of Connected and Autonomous Vehicles (CAVs) under Cyber-Attacks on Perception and Safety Monitors

CAVs have recently attracted interest both for researchers and automotive industry. Among the various issues with the design of Autonomous vehicles (AV) including CAV, safety and security assurance as well as dealing effectively with the trade-off among these have been recognized as very important. The debate regarding what level of safety and security of (C)AV is socially acceptable is very active at the moment [1], [2]. In this paper, we present a probabilistic modelling approach to dealing with the problem of safety assessment of CAV under cyber-attacks, and demonstrate its plausibility and usefulness in ranking various modes of vulnerability of the essential components of (C) AV such as the AV perception system and safety monitors.