The self-driving car could be a decade away. Here's what needs to happen
Remember in “Back to the Future” when Doc Brown almost died because he stole plutonium to fuel the DeLorean? Then in “Back to the Future II” everything was gravy because he’d found a way to fuel the time machine using garbage? That’s kind of the narrative that needs to take place for self-driving cars to come to market. Just, with less Christopher Lloyd.
Stay with me. We don’t need some miracle breakthrough like turning garbage into gasoline. But self-driving cars do need to leave the realm of bombastic experiment and enter the world of financial reasonability. In terms of technology (and not regulatory issues, which is a story for another post), only then will the cars become a commonplace reality instead of the world’s most advanced and expensive parlor trick.
Rob Coneybeer, a partner at Shasta Ventures who has looked into backing the technology, believes that moment will happen in about 10 years. He envisions adoption like so: a customer will be able to walk into a high-end German dealership, like Mercedes Benz, pick out an $80,000 car, and be able to check a box to add on the self-driving option, for $10,000 or $15,000 more. This pattern in the auto industry is nothing new, he says. New technologies – airbags, antilock breaks – have always been introduced in the highest-end luxury cars.
He thinks the technology will be available to make the cars safe and cheap enough for an affluent consumer in about six or seven years. And then it will take about three to five years to emerge in a car design cycle. 10 years after that, we'll have wide adoption, he says.
The cornerstone of the technology is lidar, or light detection and ranging, equipment that looks like a small spinning trashcan on the car’s roof. This is essentially the self-driving car’s flux capacitor. It uses sensors to measure the distance of objects and create 3-D maps in real time, so the car knows how to stop, start and react. It’s also the single technological thing getting in the way of mass production.
One big hurdle is adapting them to weather, says Coneybeer. Like human eyes, blinding sunlight, snow and sleet impair the vision of the lidar equipment. But an even bigger problem is the price tag, which Google has said is about $70,000 a unit. That cost will have to go down significantly.
Last week, the Guardian reported that Oxford professor Paul Newman led a team that developed a piece of machinery that fits onto existing cars and can take over the drive when conditions are good. The most intriguing point is that Newman thinks he can eventually get the cost down to £100, or just above $150. The team has a long way to go. Right now it costs about £5,000. But it’s still a lower starting point than what Google’s engineers are reportedly building, and there haven’t been any indications that it’s a goal of theirs to get the price that low.
The search giant’s version of the wondercar seems geared, even in these early stages, toward something more all-encompassing: a vehicle that will drive for you all the time, in any condition, so you can sleep behind the wheel, non-drive home drunk, and tell your car to find parking while you’re not in it.
There's a huge contrast between the two approaches. There’s the upgraded options version, which would really push the bounds of “fully loaded” when buying from a car dealer. Then there’s Newman’s version, which seems almost DIY-ish, and while more limited, can lead to a more democratized path to adoption. It’s not that these two approaches won’t diverge at some point, and reach the same levels of driverless nirvana, but the point is that they are starting at very different places.
Of course, all of this is speculation for now. Either way, there is a lot of work to be done. And while comparing all of this to “Back to the Future” is fun, it’s not as fantasy-adventure as it seems. “Having run around Silicon Valley closing in on 17 years now, and seeing how these technology waves come, this is a truly inevitable one,” says Coneybeer.