Autonomous technology has been promoted as perhaps the most significant revolution in vehicle safety, however a new study suggests automakers face significant hurdles in validating such claims.
It is well established that human error causes the vast majority of car accidents and road fatalities. Self-driving and semi-autonomous cars address the problem by avoiding common errors altogether or intervening when a driver has made a mistake.
Google's latest self-driving report cites nearly 1.5 million miles driven, with just a single minor accident blamed on the autonomous technology. Such statistics may not reflect how safe the company's cars would be if a highly-trained employee was not hovering over a kill switch, ready to take over.
Google's fleet was operated in manual mode for more than a million miles, representing 40 percent of the total. To be clear, the company has not claimed its cars are ready to put on the market.
"Given that current traffic fatalities and injuries are rare events compared with vehicle miles traveled ... fully autonomous vehicles would have to be driven hundreds of millions of miles and sometimes hundreds of billions of miles to demonstrate their safety in terms of fatalities and injuries," RAND researchers found.
Conservative regulatory proposals call for human drivers to receive special training and remain responsible for operation of the vehicle, always watching the road and ready to take the helm. Google is among the most optimistic, arguing that the safest autonomous car will have no steering wheel, pedals or other manual controls that would allow an error-prone human to interfere.
RAND argues that regulations should be framed to accommodate a broad range of autonomous technology as it progresses over time. Automakers will also have to create "innovative methods" to demonstrate safety without relying on miles-per-accident statistics or virtual simulations.
"It is imperative that autonomous vehicle regulations are adaptive -- designed from the outset to evolve with the technology so that society can better harness the benefits and manage the risks of these rapidly evolving and potentially transformative technologies," the report adds.
The RAND report summary does not provide any specific recommendations for how to validate safety, and its authors caution that it may not be possible to establish certainty.
The National Highway Traffic Safety Administration has promised to hasten federal rule-making for autonomous vehicles. The agency is still listening to stakeholders and the public before it begins pushing forward with new laws and provides clearer guidance to state-level regulators.
Here at Leftlane, we believe the contributions of the community are just as valuable as those of our staff. The Web, at its core, is a medium for information sharing and communication. We strive to provide good information, but without you—the community member—there is no communication. Read More>>