The U.S. National Highway Traffic Safety Administration (NHTSA) has begun developing federal-level performance standards to assess the safety of self-driving cars, judging that common benchmarks are needed to evaluate how safely autonomous-driving technology operates on real roads. On Sept. 20, TechCrunch reported that NHTSA and SAE International launched a consortium called ASCEND to accelerate development of data-based performance standards for autonomous-driving systems. There is currently no common safety evaluation framework for autonomous-driving systems comparable to a human driver's licensing test, and companies have mainly relied on demonstrating safety using real-world driving data. NHTSA believes existing Federal Motor Vehicle Safety Standards (FMVSS) alone make it difficult to sufficiently address new forms of self-driving vehicles, since current standards assume a conventional passenger car design and are hard to apply as-is to vehicles such as Tesla's Cybercab or Zoox's robo-taxi that lack a steering wheel, pedals, side mirrors and rearview mirrors. Jonathan Morrison, NHTSA's chief counsel, called evaluation of autonomous-driving system capabilities ASCEND's top priority at the AI Horizons Summit in Pittsburgh, saying standards for autonomous-driving performance could later be reflected in federal motor vehicle safety standards. Morrison said NHTSA aims for a regulatory system that keeps safety at the center of policy without overly restricting technological innovation, and indicated it would avoid adding rules after the fact each time a specific accident or problem occurs, or regulating based on a predetermined view of the direction of technological