When it comes to autonomous vehicles, the federal government has the opportunity — and the resources, now — to act decisively for safety.
NHTSA should establish a comprehensive view of safety that accounts for both cutting-edge real-world and virtual testing. This is important to recognize as we see more commercialization of autonomous vehicles, and Congress continues to explore the best way forward on regulating the technology.
But what does it mean to take a comprehensive approach?
Software is playing an increasingly important role in conventional cars. That is even more true when it comes to autonomous vehicles, where software is the defining element to ensure safe operations. For that reason, autonomous system manufacturers conduct rigorous testing and development that involve real-world (or on-road) testing as well as simulation in a virtual environment. While both test methods are important, the vast majority of the “miles tested” happen in a virtual environment.
Why is that?
In a study by RAND Corp., researchers note that “autonomous vehicles would have to be driven hundreds of millions of miles and sometimes hundreds of billions of miles in the real world to demonstrate their reliability.”
Real-world testing alone is not only insufficient for software, but also not feasible.
The study’s authors discuss the need for virtual testing and simulation, among other methods, to “supplement real-world testing in order to assess autonomous vehicle safety.”
Simulation augments real-world testing and continuously validates the performance and safety of autonomous vehicles. For example, developers can model complex situations in a virtual environment that involve drivers, pedestrians, cyclists and other vulnerable roadway users.
There are also “edge case” scenarios that are unsafe, expensive or impossible to replicate in the real world.
These edge cases include rare-weather events, such as rain on a sunny day; unexpected traffic encounters, such as a sudden pull-over; or any combination of these and countless other situations. Simulation can test for these events with hundreds of variations.
Virtual testing should be a core part of validating the safety of autonomous vehicles before they are deployed on public roads. For both governments and developers, it can help determine safe areas for on-road testing and deployment.
The benefits of simulation can be realized by the public sector, too. Adding simulation capabilities to the government’s test infrastructure would allow NHTSA to answer key research questions that lead to the independent safety validation of autonomous vehicles, and help the agency go beyond its traditional reliance on physical laboratory and test-track testing.
Thankfully, Congress has recognized simulation technologies as a valuable tool to enhance NHTSA’s research and regulatory capacity by providing $3.5 million in FY23 to validate autonomous vehicles through simulation. Moreover, a FY24 House bill includes $4 million demonstrating Congress’ continued and increased support for a software-based approach to safety.
This is an opportunity for the federal government to leverage commercially proven simulation testing as part of its toolbox to ensure safety.
Autonomous vehicles have the potential to revolutionize transportation, bringing significant mobility and economic benefits. The technology holds the promise to not only improve road safety by removing human error, but to also expand access to transportation for so many Americans.
The U.S. must maintain its edge as the global leader in technology and innovation. As such, our nation deserves the highest possible safety standards for autonomous vehicles.
Safety and public trust will be critical in the widespread adoption of this important and captivating technology. The federal government has the opportunity to utilize available technology and tools to strengthen autonomous vehicle safety so we can realize a safe transportation future that serves the public good.
