On 15 – 16 June 2016, the EPFL International Risk Governance Center (IRGC), in collaboration with the EPFL Transport Center (TRACE), organised an expert workshop on Risk and Opportunity Governance of Automated and Autonomous Cars. The workshop, conducted under Chatham House Rule, convened a group of participants from academia, regulatory agencies, insurance and industry to discuss various governance issues surrounding autonomous cars.

The objective of this workshop was to improve the understanding of the risk and opportunities related to automated and autonomous cars, and discuss how to create appropriate governance and regulatory context conditions for the development of autonomous driving. Over the course of two days, participants discussed:

  • The vision for automated and autonomous driving, expected benefits, and current initiatives
  • Risks associated with and proactive solutions for the development of autonomous driving
  • Role of and perspective for insurance and regulation
  • Adaptive pathways, governance and policy recommendations

Key take-aways from the workshop included:

  • It is interesting to note that, although the technology for fully autonomous driving is not ready (at least from the safety perspective), some car manufacturers are selling it, some drivers are using it and some public authorities are strongly encouraging it. The potential and expected benefits, in terms of road safety, air quality and mobility, are seen as quite large.
  • Autonomous driving will not eliminate the risk of accidents. The level of safety that will be required, accepted or desired with fully autonomous cars will depend on technical, regulatory and societal matters. Safety is not purely technical, it also depends on individual and societal choices and preferences, as well as behaviours. In particular the level of accidents involving or caused by autonomous cars that society will accept is not clear, in terms of frequency, severity and type. Accidents caused by human errors are generally more acceptable than accidents caused by machines.
  • To improve the technology for automated and autonomous driving, data and information from testing as well as real road experience and incidents are needed. Issues of data collection, access and use are central to learning from experience, assessing and managing technical risks, and adapting regulation. For example, sharing data from incidents and accidents reporting is needed. Regulators need to make tradeoffs between the need for data sharing and the need to respect the confidentiality and privacy of drivers and passengers.
  • Testing and real road experience is under way in several countries and settings. But more arrangements forĀ  pilot cases are needed to progressively allow certification, standardisation, authorisation and confidence.
  • Effective collaboration between industry, regulators and other actors involved in mobility policies will determine the pace at which technology is being implemented. It appears that the industry and regulators are working closely together and regulation is generally supportive of the technology for automation and autonomy.
  • Connectivity between vehicles and infrastructures will be a key success factor, for optimizing the capability of a vehicle to make sense of its environment (other vehicles and infrastructure), as well as for addressing mobility (congestion) issues. However, connectivity can also be a new source of risk, with which the automotive industry is not familiar. If the vulnerability of vehicles and transportation systems increases as autonomy increases, fail-safe or resilience strategies will be needed.
  • Adaptive pathways whereby regulators, industry and society collaboratively learn how to manage risks and benefits from the technology are desirable to manage fears, risks and incidents, and progressively develop regulations. It might be preferable to make exceptions to current regulations, rather than trying to fix things with new regulations and laws before sufficient experience has been collected through real cases, small steps, failures, learning-by-doing, and collaboration.


Background paper to the workshop (IRGC, June 2016)

Workshop summary report (forthcoming)