Autonomy in Cars Progresses, But Regulators Struggle to Keep Up
The federal government is scrambling to deal with the rapid pace of IT-driven innovation in cars.
Human error is a major factor in the accidents that kill more than 32,000 people on U.S. roads and in the traffic congestion that wastes more than $120 billion each year.
The U.S. Department of Transportation’s first policy statement on the safety aspects of automation in cars reflects the fact that technology advances in vehicles are outpacing the usual regulatory process.
While truly self-driving cars are years away—if they ever arrive—consumers are seeing far more car models bearing sophisticated semi-autonomous features. These include radar assisted cruise-control, which can keep a fixed distance from the car ahead; systems that warn drivers if they veer out of their lanes; and technologies that can prevent oversteering or even apply the brakes when they detect that a crash is imminent (see “Self-Driving Tech Veers into Mid-Range Cars” and “Proceed With Caution Toward the Self-Driving Car”).
Such advancements stem from the falling costs of sensors and processors, and from advances in software that processes information from the sensors. Carmakers and suppliers are producing “new components, new functions, and new applications using cameras and radar and even lidar [laser-based ranging finding] much faster than the traditional regulatory time frame would allow,” says Richard Wallace, director of transportation systems analysis at the Center for Automotive Research, based in Ann Arbor, Michigan.
At the cutting edge of automation, carmakers and others are testing fully autonomous prototypes. Google has built customized versions of sensor-laden, entirely self-driving Priuses and Lexuses that have logged more than 500,000 miles on U.S. roads—albeit with a human driver ready to take control (see “Look, No Hands”).
With three states and the District of Columbia having passed legislation to allow researchers to test such prototypes on real roads, Washington is grappling with how to regulate the cars. John Capp, the director of active safety systems for General Motors, says federal regulators are “trying to understand these things and trying to figure out what role they should have.”
The National Highway Transportation Safety Administration (NHTSA), which sets and enforces car and truck safety standards, issued a preliminary, nonbinding statement on car automation technologies last week; the 14-page document announced the launch of a four-year research project intended to assess how humans interact with automation technologies, how reliable the technologies are (see “Buckle Up for the Vehicular Zombie Apocalypse”), and how vulnerable automation software might be to hacking (see “Hacking Cars to Make Them Safe”).
Unsurprisingly, NHTSA’s statement said that fully autonomous technology isn’t ready for the general public. But the fact that the agency is calling for more study is a reminder of the glacial pace of regulation: in the case of lane-departure warnings and crash-avoidance systems, it’s studying technologies that have already been on the market for several years.
Become an MIT Technology Review Insider for in-depth analysis and unparalleled perspective.Subscribe today