John Paul MacDuffie discusses the new federal policy on driverless vehicles

A proactive regulatory regime and a cooperative approach from auto makers are the key backdrops of the U.S. government’s policy for automated vehicles that was released last week. Time was when auto makers often clashed with regulators over safety requirements, and regulators often responded with hindsight wisdom. But automated vehicles, such as self-driving cars, could potentially save thousands of lives, especially when 94% of crashes on U.S. roadways are caused by human choice or error, according to a briefing from the U.S. department of transportation.

The central aspect of the new guidelines is a 15-point safety assessment scale to guide the design, development, testing and deployment of automated vehicles. It could also help create consistent nationwide regulations, removing the need to comply with differing regulations as drivers cross state borders. Alongside, auto makers, software developers and fleet operators like Tesla, Google and Uber are sharpening their technologies in preparation for a full-fledged rollout of self-driving cars in 2021.

“Tradeoffs and design choices are being made,” says Wharton management professor John Paul MacDuffie, who is also director of Program on Vehicle and Mobility Innovation at the School’s Mack Institute for Innovation Management. Safety in self-driving vehicles hinges on two critical aspects – “good object recognition and good distance estimation,” he adds. “It may only be that when we have got camera, radar and Lidar (distance estimation using laser illumination) all operating that we may get the accuracy that we need.”

MacDuffie discussed the evolving landscape for automated vehicles on the Knowledge at Wharton show on Wharton Business Radio on SiriusXM channel 111. (Listen to the podcast at the top of this page.)

Here are five key takeaways from MacDuffie’s interview:

Technology Pulls Ahead: Ride-hailing services provider Uber is piloting driverless cars in Pittsburgh, Penn.; Tesla has launched new software for self-driving cars; and auto component suppliers are realigning themselves, among other developments over the past six weeks, says MacDuffie. “It’s a very dramatic time for self-driving vehicles.”

Along with news of fresh technological developments come reports of vehicle accidents and breakups, such as that between Tesla and its supplier Mobileye, he notes. The two parted ways following a fatal accident last May involving a Tesla self-driving vehicle and Mobileye software. Mobileye has since teamed up with component supplier Delphi to develop fully autonomous driving technology.

Ahead of the Curve: The U.S. Department of Transportation has been proactive in developing federal policy for automated vehicles, notes MacDuffie. “They wanted to get out in front of a fast-moving phenomenon…. [They are] trying to give all the players a consistent set of parameters to shoot for, but without constraining with restrictive rules at a time when the technology is very fluid and the strategies for using it are also very fluid.”

Federal policy could also be adopted as the regulatory template by various U.S. states, MacDuffie adds. “If all the states end up with slightly different plans, that makes it confusing for anyone who is trying to operate in this space,” he points out. “Right now when you cross a state line and you’re driving, mostly you just have to make sure the speed limit hasn’t changed, but what if you have completely different laws governing this new technology?”

Cooperative Stance from Automakers: “[It’s an] interesting time [also] in terms of the relationship between the government regulating bodies and the auto industry,” says MacDuffie. “Typically, the industry resists regulations and the government is more reactive to developments, such as enforcing safety guidelines after accidents.” But in the case of automated vehicles, it is “different and potentially more cooperative,” he adds. Almost across the board, auto companies and industry interest groups feel that this time “the government got it right in terms of guidelines that create some direction and some parameters without too many restrictions.”

“Once a human driver takes his or her attention away from driving, to have them come back … to assess the situation and take control of it is too risky, particularly at higher speeds.” –John Paul MacDuffie

Preparing for 2021: In the run-up to 2021, the much-anticipated year when the auto industry expects to have a full fledged launch of self-driving vehicles, resolving known problems is a top priority. Ventures such as Uber’s pilot trials in Pittsburgh and San Francisco will be tests and experiments, some of which will be in collaboration with cities, and often under somewhat restricted conditions, says MacDuffie.

The technology phases for automated vehicles are indicated by four levels. “The big strategic, technical and regulatory issue is about the feasibility or otherwise of Level 3,” says MacDuffie, explaining that this level is the stage where the responsibility for driving is handed back and forth between the artificial intelligence software and the driver. Levels 1 and 2 deal with features like cruise control and alerts when cars stray off lanes. Level 4, where there will be no human intervention in driving at all is much further away, he predicts. “Once a human driver takes his or her attention away from driving, to have them come back — even with bright lights and flashes and vibrations — to assess the situation and take control of it is too risky, particularly at higher speeds.”

Perfecting the Technology: The technology that is evolving for self-driving vehicles “is designed to learn from experience, so all the data from testing goes back to help identify different situations that come up,” says MacDuffie. “People in the traditional auto industry believe the path to this will be when vehicles could communicate with each other, such as with transponders and some agreed-upon standards,” he adds. He lists the main ingredients of the technology as cameras, radar and Lidars. “Cameras are great at identifying objects; radar is great at identifying distance,” he says.

Also in store is a prolonged period of coexistence between the old and the new. “Until we have all the current cars replaced with fully autonomous vehicles, we will have a system that is a mix of human drivers and early adopters of automated vehicles,” says MacDuffie. The regulatory and technology standards will also have to account for the fact that “humans don’t follow the rules when they drive.” Meanwhile, Uber driverless taxis in Pittsburgh always have one or two Uber employees in them to collect data, he adds, noting that Uber plans to post signs suggesting that passengers not speak with these employees so that they can focus on testing the vehicles.