The development of Automated Driving Systems is challenging international regulators to make change in order to determine how such systems may be legally sold and driven on our roads. Road safety is expected to be one of the main benefits of automation but it also offers the potential for significant new risks, particularly in the early stages. The insurance industry has a core interest in both the technological and associated regulatory changes because of their impact on liability, cost, and underwriting and has sponsored significant research activity to understand the issues.
Driver assistance systems that act in the brief moments before a collision, or to support and not replace driver inputs, are being progressively proven effective and full automation (SAE levels 4&5) is expected to be highly beneficial. However, where the driver is not needed for the driving task but must be capable of resuming control at any moment (SAE 3) considerable risks of public confusion and pricing difficulty have been identified. Modelling of crash risks suggests that the net benefits of such systems will be positive but that significant risks remain. This would support a binary definition of Automated Vehicles at SAE level 4 and above. Anything at level 3 or below would be considered Assisted Driving where the driver remains responsible for safe operation. It is considered that to avoid becoming a barrier to positive changes, developing the requirements for automated vehicles should commence now. In the short term, the development of requirements related to assisted highway driving at SAE level 2/3 should continue with the aim of maximising safety benefits as well as minimising risks. The latter may require full driver monitoring systems rather than simple measurement of whether driver’s hands are on the wheel. Consideration should be given to creating a separate regulation integrating requirements for assisted and autonomous driving as well as to the implementation of different regulatory approaches, potentially including elements of self-certification. The aim should be to maximise the speed and flexibility of the regulatory process while providing clear, coherent and robust requirements.