The road to safe delivery of Highly Autonomous Vehicles (HAV) on a global scale is paved with significant technical and societal challenges. Not least among these is how manufactures, policy makers and the public at large can make objective decisions about how these vehicles will perform in complex environments. Current models of communicating safe operation based on 'miles driven' don't effectively scale to the diversity of environments across the world. This challenge only grows when you consider that measures of safety performance for AI may need constant assessment as environments evolve and patches are delivered throughout the vehicle lifecycle. One answer to this problem might be the application of Virtual Validation and Certification (VVC).

This presentation will examine the potential role of VVC in safe delivery of autonomous vehicles to our communities. Covering the direction of current industry and government initiatives. Examining what technology and capabilities might be needed to map between simulations of sensors, control systems, roadside infrastructure, real world scenarios and human factors. Some suggestions on what ‘national infrastructure’ might be needed to aggregate the supporting science, data and technology for VVC.