1374367 - Systematic Investigation procedure for identifying cause and consequence on accident involving level 4 autonous vehicle

Like most industries, the transportation sector is undergoing a huge transformation driven by artificial intelligence (AI) and information and communication technology (ICT), a move towards mobility. Among them, autonomous driving is the biggest topic in the global automobile market. All manufacturers are developing technologies for autonomous driving, and the government is also making efforts to commercialize autonomous driving by selecting cities for autonomous driving pilot projects. However, National Highway Traffic Safety Administration (NHTSA) released data on 10 months of crashes (a total of 392 crashes involving Level 2 ADAS vehicles). Since traffic accidents are directly related to human life, it is crucial to investigate the causes and risk factors and prepare future countermeasures. However, unlike human drivers, autonomous driving systems can be manipulated by cyber attacks. For instance, an autonomous driving system may perform unintended control or deceleration due to a cyber attack which may lead to an accident. Furthermore, traffic accidents caused by cyber-attacks cannot be identified using the existing police accident investigation system. Therefore, this study proposes an accident investigation process to identify the cause of an accident that has not occurred before (e.g., cyber-attacks). The accident investigation process defines the process of autonomous driving traffic accidents based on five mechanisms; physical condition, security, communication, functional events, and driving. The investigation objectives can be divided into (1) pre-crash, (2) crash, and (3) post-crash stages, and each stage will have a given role. Roles are divided into on-site police, autonomous vehicle investigation team, digital forensics team, and autonomous vehicle traffic accident analysis center. The pre-crash stage will include causes of failure (ODD, human factor, and digital infrastructure) and vehicle failure (physical, functional, communication, driving, and security). The crash stage includes elements that should be performed by the on-site police, who are the first to arrive at the scene of the accident. In the post-crash stage, future accident prevention and management will be included. Also, this paper proposes a bow-tie diagram that can identify the cause and consequence analysis with multiple case studies.