In 1957, a group of American power companies unveiled a futuristic-sounding ad featuring a smiling family playing dominoes in the backseat of car propelling itself automatically down the highway.
It’s all due to to electric power, the ad promises, the car’s “speed and steering automatically controlled by electronic devices embedded in the road... No traffic jam... no collisions... no driver fatigue.”
Nearly 60 years later, Google performed a similar experiment, offering its employees the chance to test a self-driving car that would steer itself down a California freeway. As it turned out, the car’s technology offered drivers the chance to be too free.
“What we saw in the car was the people who were told ‘You have to be vigilant,’ because the technology worked so well, they weren’t,” says Chris Urmson, the head of Google’s self-driving car project, addressing a group of government workers and visitors at an event on Tuesday hosted by the Department of Transportation in Cambridge, Mass.
One man, he recalled, discovered his cell phone battery was running low and pulled out his laptop and a series of cables and plugged the phone in, all while moving down the freeway at 60 miles an hour.
“We could spend a lot of time thinking about how to get that driver to pay attention, you know, we could monitor them, and every time [they] look away, we zap them,” he says, drawing laughs from the crowd.
Instead, it provided the company with the opportunity to expand its approach to focus particularly on daily urban driving and heavy traffic faced by many commuters.
“If we go back and think about our mission, improving people’s lives by transforming mobility, a car that drives on a single line on the freeway doesn’t help a blind woman get to work in the morning – it doesn’t really get there, he says. "It turned out it’s a tough problem."
Mr. Urmson had come to the Volpe National Transportation Systems Center, in a bid to pitch the project to a different sort of audience. The room, full of government workers from the building, seemed reminiscent of a college classroom – more curious than reverential – in contrast to an demonstration of Google’s self-driving car in California in October, where the appearance of Google co-founder Sergey Brin reportedly brought the room to a standstill.
“I asked Brin how he planned to humanize the technology, how to take it beyond Google’s privileged and pro-technology bubble into a real world where citizens are more skeptical and less trusting. He looked disappointed with the question,” wrote the Guardian’s Jemima Kiss.
“I’d explain to them the benefits,” Mr. Brin told the Guardian. "Our mothers might be having a hard time driving and still want to get around. It’s important to have that mobility, so I’d just explain it’s an affordable way to get from here to there that is safe."
In Cambridge, the mood seemed looser, with Urmson sometimes employing moments of black humor to underscore what Google says is the goal of its self-driving car: making roads safer by creating an autonomous car that anticipates difficult driving situations.
In the US, he calculated, given an average commute time of 50 minutes per day and 120 million workers, the workforce spends 6 billion minutes a day commuting. Noting that about 33,000 people are killed on the road every year, he said, “we’re not just killing people, we’re killing them slowly. And we think we can do something about that.”
While the company’s self-driving cars – which are currently being tested in California and Austin, Texas – have drawn some headlines for being involved in a series of accidents and mishaps, including being threatened with a ticket for driving 11 miles below the speed limit – Urmson emphasized that the cars performed better than a human driver would.
For example, a Google self-driving car was confronted with a cyclist weaving erratically through the lanes while driving at night. It came to stop at a light, where it was confronted by another cyclist who barreled through the light while wearing black clothing, making him almost invisible in the darkness.
While a human driver might have focused on the first cyclist, Urmson says, causing a potential accident when the cyclist rode forward. Instead, he says, the car detected both cyclists and waited for them to pass before moving forward.
Google has often said the cars are designed to be extra-cautious to avoid accidents often caused by human error. But that approach has had some drawbacks, Urmson says. “Our cars today, we think, do a really good job avoiding accidents. They do by being a little oversensitive to the signals out there. So [the challenge] is: how do they keep that level of paranoia while making it not, kind of, overreact completely.”
Part of this involves “teaching” the car rules of the road, much as a human driver would learn, such as learning to distinguish between objects the car can drive safely around — such as a police car directing traffic – and those the car should not — such as a school bus.
Urmson’s mix of sincerity and humor seemed to win over the government workers, with one person exclaiming “I’ll buy one!” as the meeting broke up.
Arguably, part of this had to do with his own background, having designed a robotic car at Carnegie Mellon that won a 2007 DARPA Urban Challenge award, a contest designed by the Defense Department to create a successful self-driving car.
But his tone also seemed different. In sharing stories about designing an earlier car that had burst into flames and revealing a video showing an encounter between a Google car and a woman chasing a duck, he seemed determined to show a different side to the company than the previous head of the project, Sebastian Thrun, who one commentator once likened to a “Bond villain.”
“Our philosophy has always been, it has to work with the existing infrastructure,” Urmson told the audience at Volpe, responding to a question about the Transportation Department’s possible role in the development of the cars. “Fundamentally for us, if you make the infrastructure better for human drivers, it’s better for us.”