The first thing you have to do is accept the idea that we will one day all be in driverless cars. But the people who think about such things for a living are seriously convinced this will happen. (Video)
“The technology is pretty much already there,” says Peter Stone, a computer scientist at the University of Texas at Austin. And this was also the jarring promise of Tom Vanderbilt’s recent profile of the autonomous car in Wired. “But the question is when will it be cost-effective? When will the legal industry wrap its head around it, and the insurance industry, and when will people buy into it? I don’t know when it will actually happen. But the potential advantages are so huge that it has to happen eventually.”
Stone is thinking of the advantages for the disabled and elderly who can’t currently drive, for parents who don’t have time to take their kids to soccer (they can take themselves!), and above all for traffic safety and the more efficient movement of people everywhere.
It’s one thing, though, to realize that Google engineers have been zipping through our midst in autonomous concept cars. It’s another to picture what will happen when we’re all in these things – when the eye contact and social rules that currently govern urban driving are replaced by computer systems chatting with each other.
“When they do interact,” Stone says, “it will be at intersections as much as anywhere else on the road.”
He and one of his doctoral students, Kurt Dresner, started thinking about this in 2003 (way back, Stone says, “when people thought autonomous cars themselves were far in the future.”)
They realized intersections will change not just because they’ll need to accommodate driverless cars, but because driverless cars will make intersections much more efficient. Right now, you may wind up sitting at a red light for 45 seconds even though no one is passing through the green light in the opposite direction. But you don’t have to do that in a world where traffic flows according to computer communication instead of the systems that have been built with human behavior in mind.
“There would be an intersection manager,” Stone says, “an autonomous agent directing traffic at a much finer-grain scale than just a red light for one direction and a green light for another direction.”
Because of this, we won’t need traffic lights at all (or stop signs, for that matter). Traffic will constantly flow, and at a rate that would probably unnerve the average human driver. The researchers have modeled just how this would work, as you can see in the animation below. You have to admit the patterns are mesmerizing even if the whole idea still seems far-fetched. The yellow cars pausing at the intersection in this simulation are old-timey human-driven vehicles that haven’t yet caught up with the future (because while we may all be riding in driverless cars one day, that day won’t arrive overnight):
Those human-driven cars would have to wait for a signal that would be optimized based on what everyone else is doing. And the same would be true of pedestrians and bike riders. Stone says the system is designed to have flexibility under the assumption not all decisions would be made by computers alone.
In some ways, their model is an old-fashioned one.
“A lot of people advocate roundabouts,” Stone says, “but that takes a lot more real estate, and there’s just not room for them in many urban areas. This will allow us to use the same roadways we have now, just use them now more efficiently.”
Stone envisions that you’ll be doing this from the back seat of your car while reading the newspaper (you know, but on a tablet), having an experience a lot like the one you’d get on the train. Of course, whether we want to turn the personal car into a private train-like experience is a whole other question. But it’s probably not one for a computer scientist.