A robotaxi artificial intelligence safety failure

Full self-driving capability in autonomous vehicles within all operational domains is one smart mobility vision companies compete to achieve. They view self-driving as aiding transportation efficiency and enabling the commercialisation of ”robotaxi” services and other use cases. However, the testing of robotaxis within urban areas continues to see safety-related incidents. We provide a case study of a 2023 San Francisco event to illustrate why the behaviour of Automated Driving Systems is under scrutiny. A serious injury to a pedestrian by a robotaxi occurred and the subsequent mishandling of the regulatory reporting requirements saw further investigations. This resulted in suspended commercial operations, job losses, and a vehicle recall. We construct a timeline of the incident and the follow-up events. Subsequent third-party analysis of the incident provides further insights and raises a typical dichotomy of safety versus commercial considerations. We use the insights to argue for governments and regulators to consider a different approach to developing and testing self-driving capability within public spaces, namely, the collective engineering of one universally verified self-driving artificial intelligence usable by all vehicle manufacturers.