Agile Regulation for Auto Safety and Innovation

Download this Commentary (PDF)

Where are the new regulations for self-driving cars?

The futuristic vision of self-driving cars has attracted billions of dollars of investment, generated national press attention, and excited both enthusiasm and concern among consumers. 

Test vehicles operate in many of our neighborhoods with safety drivers, collecting detailed data that are used to develop and improve models of safe decision-making. When self-driving cars are at last able to operate reliably without human fatigue, errors, or impairment, will new rigorous new safety standards be ready?  Should they be?

The promised benefits are exciting: Enhanced mobility promises independence to our friends and neighbors who are unable to drive traditional vehicles, and an expansion of productive time for urban commuters whose attention to driving prevents them from working during long trips to and from the office.  Most importantly, we anticipate that self-driving vehicles won’t become sleepy or distracted, offering the potential to save lives from deadly crashes. Safety stakeholders, communities, and members of Congress have called for swift issuance of regulations to assure the safety of emerging self-driving technology. 

Despite the call for new regulation, we should not expect new safety regulations — at least not in the form of traditional engineering-based safety standards — because development of such standards for an emerging, rapidly changing technology could require decades of work and would prevent the technology from maturing to achieve its greatest social benefit.  But the U.S. has already embraced a system of agile regulation that may be well suited for the rapidly evolving technologies.

Traditional safety standards are unlikely to be appropriate for rapidly changing technology. The reasons for this include the rapid pace of evolution of the nascent technology combined with the obligation to support new safety standards with time-consuming engineering evidence and the important requirements to engage the public and stakeholders in a transparent, well-informed process before issuing regulations that will shape transportation for decades to come.

Traditional Standards Take Time

To illustrate the time required to develop a robust safety standard, consider federal regulations for the safety of electric vehicles, standards that required nearly 20 years to develop and are still undergoing occasional amendment in recent years.  In 1991, the National Highway Traffic Safety Administration (NHTSA) published an Advance Notice of Proposed Rulemaking to determine what safety standards may be needed. Following extensive testing and stakeholder engagement, in 1998 NHTSA proposed to adopt a new safety standard. That standard was finally issued in 2010, requiring that batteries don’t eject, leak or electrify rescuers. 

Even after two decades of research and standards development, the work is not yet complete; decades later, NHTSA continues to work to reduce the risks associated with difficult-to-control battery fires. The defined safety standards do not address every risk, but the technology continues to improve and NHTSA oversight provides a safety backstop.

To design a safety standard under the current time-consuming process requires that the technology be fairly mature and its performance consistent and measurable so that it can be described to the public and subject to laboratory tests. The time-consuming regulatory process is designed both to allow transparency to the public with opportunity to comment, and also to produce engineering-based safety standards and test procedures that manufacturers can follow meticulously to self-certify their compliance, and for enforcement agencies to accurately measure compliance. 

This meticulous but slow process is poorly suited for digital technologies which can be improved and updated quickly — in some cases, within a few days of spotting a safety issue. To establish detailed engineering standards for technologies — some of which are changing as quickly as software can be edited — would delay the improvements that the technologies can offer.

A Two-Prong Approach Supports Safety and Innovation

But that doesn’t mean that unsafe cars are coming to our local roads. 

Congress anticipated that innovation would continue to bring improvements to the transportation system and designed an overarching safety program that supervises safety while allowing innovation to flourish. In addition to specific engineering-based safety standards, Congress established a system of oversight to protect consumers from safety flaws in the design, manufacture or performance of cars and trucks. Those oversight responsibilities don’t require anyone to be hurt to recognize a flaw as a safety concern but acknowledges that certain flaws have the potential to cause serious harm and that manufacturers should recall a product if that product’s flaw is unsafe. 

As a result of this dual system of rigorous standards for established technologies combined with safety oversight for rapidly changing technologies, we have a system of regulation that balances agility and rigor, and which offers a role model for addressing the competing needs of safety today and innovation that supports safety in the future.

The balance of competing objectives attracts critics from both sides: On one side are safety advocates seeking greater safety protections through additional, faster standards-based regulations, and on the other side the advocates of innovation have claimed that a less stringent system is more likely to yield safety and other benefits for consumers. 

What would an improved regulatory regime look like? A brief thought experiment may provide insights: 

Imagine first if we were less reliant on engineering-based safety standards currently designed to improve crash avoidance, crash-worthiness, and post-crash survivability. Individual standards could be made relevant for new technologies through updates — for example, windshield defrosting and defogging system standards could be modified to apply to cameras and sensors — but the broad objectives of avoiding crashes, protecting occupants and surviving a crash continue to be relevant and also quite important regardless of the vehicle type or technology.

Now, imagine greater reliance on engineering-based safety standards and reduced reliance on oversight.  The evolution of some products would likely be slowed — or even stalled — because of the time required for engineering-based testing and standards development.  An example of how existing engineer-based standards can slow the introduction of new innovations is found in a product available in Europe since 2016 but not yet in the U.S.: a camera and screen alternative to the traditional side-view mirror. Side-view mirrors are subject to a safety standard in the U.S. and the new technology must be evaluated using parameters developed for the older technology.  Meanwhile, new safety technologies that do not displace older technologies – for example, blind-spot alerts or lane-keeping assist -- have become popular in new cars.

Self-driving cars, highly automated vehicles, automated driving systems: Regardless of the name, the new technologies that could someday allow a driver to yield control of a car deserve both the excitement and the concern they attract. Many aspects of such vehicles will continue to be subject to some rigorous — and sometimes outdated — engineering-based standards that apply to traditional vehicles: seatbelts and turn-signals for example.  But even the newest innovations are subject to oversight and are required under the current laws and regulations to be safe.

Conclusion

The two-pronged regulatory system designed by Congress decades ago — a combination of specific engineering-based regulations with a broad oversight function — offers an illustration of how Congress and regulators might design future agile governance systems that allow innovation to flourish while also answering a public demand for safety.