Google's search for self-drive cars


The link between Google’s plans for self-driving cars and secret military experiments is not new

GOOGLE IS lobbying the state of Nevada to allow driverless robot cars to be driven on public roads.

The Nevada state legislature is currently in one of its biennial sittings to consider new laws and Bills. One of those is AB511, a bill that contains a clause to allow robot cars to roam the streets and highways of Nevada. Google is pressing for it to be approved.

Scott Magruder, a spokesman for the Nevada department of transportation explains why. “The bill originally started as one that would allow electric vehicles to use the high-occupancy vehicle [carpool] lanes but it had an amendment added to it that would also allow, not quite the use of autonomous vehicles, but at least allow for them to be registered and licenced.

“We’ve got around 400 miles [644km] of highway running through Nevada, most of it pretty flat and straight, so I guess if you wanted to experiment with cars like that, this’d be a pretty good place to do it.”

Certainly, Nevada has astonishingly wide, open spaces and a sparse population (it’s the seventh largest state of the United States, but has just 2.6 million inhabitants). It’s also home to the US Air Forces’s secret research bases of Groom Lake and Nellis Air Force base, as well as the semi-fictional Area 51; and there’s a firm tradition of cutting-edge research going on in the desert.

There’s a closer link between Google’s self-driving car plans and secret military experiments. The Defence Advanced Research Projects Agency (Darpa) is a US department of defence think-tank that looks as far into the future of weapons and technology as it can. Some time ago the Pentagon, desiring a self-driving robot reconnaissance vehicle, turned to Darpa for just that.

Darpa invited car makers, universities and software designers to come up with cars that could drive themselves without any human intervention. Over the course of three competitions run from 2004 to 2007, the cars evolved from covering simple straight-line runs through the desert to tackling mocked-up urban environments with junctions, pedestrians and other traffic. That’s where Google comes in. Sebastian Thrun is a software engineer who was part of the Stanford University team that won the 2005 challenge with a modified Volkswagen Touareg and came second in the 2007 urban challenge with a robot Passat estate. Not satisfied with this, Thrun became part of the Google team that sought to combine the internet giant’s now-famous mapping systems with robot-driving software.

And here’s where we come to the question: Why Google?

Google itself says company founders Larry Page and Sergey Brin want “to help solve really big problems using technology. And one of the big problems we’re working on today is car safety and efficiency. Our goal is to help prevent traffic accidents, free up people’s time and reduce carbon emissions by fundamentally changing car use.”

Those are the words of Thrun himself, laying out the basics of Google’s robot car project. And it’s already a well-established project, with more than a quarter of a million kilometres covered by its squadron of robot Toyota Prius cars without a single problem, except for when someone else rear-ended one of the cars in traffic.

For Thrun, it’s also a deeply personal project. One of his close friends was killed in a car accident at the age of 18, and with 1.3 million people killed in car crashes worldwide every year, Thrun reckons that millions of lives could be saved by letting the computers take over our driving.

Of course, it’s not an entirely altruistic project. The US department of transportation reckons that the average American spends 52 minutes a day commuting in a car, and as Thrun says: “Imagine being able to spend that time more productively.” Such as surfing the web, perhaps?

But, in the best sci-fi tradition, the robot cars aren’t coming; they’re already here. General Motors experimented with autopilot cars in the 1950s, following wires embedded in the road surface. In 1994, Mercedes drove a fleet of self-controlled 500SELs around the Paris peripherique for 1,000km (621 miles) without incident and the European Union’s Prometheus project, which ran from 1987 to 1995, led directly to such in-car systems as radar-guided cruise control and lane-keeping systems.

While such systems began life in the late 1990s in high-end cars (notably the Mercedes-Benz S-Class and Jaguar XJ), they’ve trickled down to more humble vehicles. So you can have a Ford Focus with radar-guided cruise control, a VW Passat that steers itself if you nod off, a Volvo S60 that brakes itself if it spots someone stepping off the pavement in front of it, or any number of cars that can parallel park.

Next year’s new Mercedes S-Class will have a system that scans the road ahead, spots bumps and potholes, and tells the suspension how to react before it hits them. Meanwhile, the EU’s new Sartre project is experimenting with road-train technology, which allows long convoys of cars, controlled by a lead vehicle, to self-drive along the motorway while the occupants read, work and even nap.

What Google is working on is combining those systems into an artificially intelligent whole, and creating processors fast enough to deal with the torrents of data that cars will need to take in to manage self-guidance along a busy street. So will our first robot cars be Google powered?

However the internet giant fares with its lobbying of Nevada’s legislators, recent history has shown that what Google wants, Google generally gets out there and makes happen. Fast.