Chris Gerdes on the Interplay of Technology, Policy, and Ethics in Autonomous Vehicle Safety and Regulation

Chris Gerdes, a Stanford professor and former Chief Innovation Officer at the United States Department of Transportation (USDOT), delves into the intricate relationship between the rapid advancement of autonomous vehicle (AV) technology and the evolving policy frameworks necessary to ensure public safety. He highlights his pioneering work on self-driving race cars, such as "Shelley" (an autonomous Audi TT capable of 120 mph), which leverages physics-based control and learning algorithms to push the limits of vehicle performance by understanding tire friction. This research, initially aimed at speed on the racetrack, directly informs the development of safer AVs for public roads, emphasizing the engineering challenge of maximizing capability while minimizing risk in unknown conditions.

Gerdes critically examines the limitations of current U.S. federal motor vehicle safety standards (FMVSS), which rely on manufacturer self-certification and a slow, multi-year rulemaking process. He explains that existing FMVSS do not inherently impede AV deployment if the vehicles meet current standards, even allowing for the interpretation of "driver" as the AI system. However, truly novel AV designs lacking traditional controls like steering wheels would require specific exemptions. This regulatory gap led to the creation of the Federal Automated Vehicle Policy (FAVP) in September, a system of voluntary guidance that encourages manufacturers to define their "operational design domain" (where the system operates) and "minimal risk or fallback conditions" (how it handles failures), submitting safety assessments rather than adhering to prescriptive rules.

Key aspects of the FAVP's 15-point safety assessment include validation methods and ethical considerations. Gerdes discusses various validation techniques, from test tracks and real-world data collection (like Tesla's fleet data) to extensive simulation, acknowledging that different companies will adopt different blends. He then addresses the often-misunderstood "trolley car problem," arguing that engineers approach such philosophical dilemmas not by making moral judgments on who lives or dies, but by designing robust systems with redundant safety features to prevent such impossible choices. Real-world ethical considerations for AVs, he posits, involve practical decisions like appropriate pedestrian proximity, balancing safety with mobility, and navigating legal ambiguities.

The discussion culminates in the fundamental tension between programming AVs to strictly adhere to traffic laws versus allowing them to mimic human driving behavior, which often involves flexible interpretations of rules for safety or efficiency (e.g., crossing a double yellow line to pass a cyclist safely). Gerdes suggests that rigidly law-abiding AVs might be impractical or undesirable for human co-existence on roads, necessitating either greater flexibility in vehicle programming or adjustments to traffic laws. This underscores the need for ongoing collaboration among engineers, policymakers, and ethicists to develop best practices that foster innovation, ensure public trust, and seamlessly integrate autonomous vehicles into society, moving beyond simplistic moral quandaries to address the complex, everyday ethical dimensions of automated driving.

I've actually been working in automated vehicles since 1992 in the Lincoln Town Cars in the upper corner are part of an automated highway project I worked on as a PhD student at Berkeley.

Shelly can actually go up to about 120 miles an hour or so on that track it's really just limited by the length of the straight it's kind of fun to watch from the outside a little disconcerting occasionally as you see there's nobody in the car although from inside it actually looks all pretty chill.

The interesting thing about this is that we've approached this problem really from one of physics force equals mass times acceleration so the car is really out there calculating what it needs to do to break down into the next corner how much grip that it thinks it has and so forth as it's going around the track it's not actually a learning approach at its core.

Racecar drivers do that to be fast as they say in racing if you want to finish first you have to finish so it's important that they actually be fast but also accident free so we're trying to learn the same things so that on the road when you may have unknown conditions ahead of you the car can make the safest maneuver that's using all the friction in between the tire in the road to avoid ultimately any accident that the car would be physically capable of avoiding that's our goal with that.

Today we have a system of federal motor vehicle safety standards so these are rules they're minimum performance requirements and each of them must have associated with it an objective test so you can tell does the vehicle meet this requirement or does it not meet this requirement.

If you have a vehicle if you start and you automate a vehicle that is currently meeting all the standards because there are no standards that relate specifically to automation you can certify your vehicle as meeting the federal motor vehicle safety standards therefore there's nothing at the federal level that prevents in general an automated vehicle from being put on the road.

The solution was made but by thinking of it as an engineer trying to reduce risk and not by thinking of levels of morality and who deserves to live or die.

I believe if we actually have vehicles that follow the law nobody will want to drive with them and so we need to think about either ways of giving flexibility to the vehicles or to the law in the sense that vehicles can drive like humans do.