WHEELS : Meet your new driver, the robot

Victorville, November 15:

There was a patriotic mood in the air on race day in the desert north of Los Angeles last Saturday. On a former airbase close to Victorville, a godforsaken town built along California’s Highway 15 to Las Vegas, several cars lined up in front of the grandstand, behind a fence and concrete barriers. Then the starting signal went off. And — nothing happened.The cars remained motionless.

That’s no surprise. These vehicles are entirely computer-controlled and autonomous; with no drivers inside, they take some time to make their minds up.

First a warning siren started flashing on their roofs, then the engines roared. One by one, with a few minutes in between them, the cars — station wagons, Land Rovers and even a lemon-yellow truck — started their journey. Laser scanners, radar sensors and stereo vision cameras served as their eyes, connected by cable harnesses to stacked computers in their trunks.

The task: to master a course of about 90 miles in less than six hours. This was certainly no high-speed chase. The objective was not to race each other, but to survive simulated traffic without breaking any traffic regulations. An unusual urban tour including passing, parking and handling oncoming traffic.

Street car desires: This competition between driverless vehicles was called Urban Challenge -the third race of its kind within three years, organised by Darpa, the research agency of the US Department of Defence.

The motivation: the US Congress has demanded that by 2015, one-third of its armed forces’ ground vehicles must be autonomous.

While the first two races were robot rallies crossing the empty Mojave desert, Darpa had to look for another site to simulate urban driving, and alighted on the defunct military base

near Victorville. Instead of housing soldiers, the abandoned and partly demolished town of Victorville is used by the military to practise urban combat.

A total of 35 teams arrived a week before the race in the desert town, all hoping to make it to the finals. Among them were wealthy university crews such as Stanford Racing, lead by Sebastian Thrun. Its automobile, a blue Volkswagen Passat named Junior, was supported by Google, VW, RedBull and the Silicon Valley venture capital group Mohr Davidow.

Two years ago, Stanford’s all-terrain vehicle, a Volkswagen Tuareg, beat the red autonomous Humvee from Carnegie Mellon University. That team’s leader, robotics expert William ‘Red’ Whittaker, came back seeking revenge. But small teams also took part, such as Axion Racing: its jeep, Spirit, has accumulated nearly 200,000 miles on its odometer.

The only European entrants were four teams from car-crazed Germany. All the teams were lured by the technical challenge, the possibility to promote themselves — and a prize of

$3.5 million. Thrun emphasises the civilian benefit of the technology, “For us, this is mainly about road safety. Thousands of people die yearly on the streets and driver assistant systems can save lives.” Cruise control, distance warning systems and lane departure warning already help today’s drivers. And with self-steering vehicles, Thrun smiles, one day a car could chauffeur its drunken owner home. To achieve such feats, the ‘carbot’ is packed with electronics.

GP Shelps the machine to locate itself within inches, while a digital map aids route planning. Sensors, especially laser scanners, measure the distance to garden fences and pedestrians. Also, measuring the laser’s reflection helps to tell whether the car is heading towards a dust cloud or a brick wall. Many teams also installed a special spinning laser by the US company Velodyne that creates a map of the area.

But sensors are only one side of the story. It’s algorithms, says Thrun, that are going to win the battle, “This is a software race.” The data from all the sensors is send to shoebox-sized computers stacked in a metal frame in the back of the car. They help to create a 3D model of the constantly changing surroundings. Based on this, the programmes make decisions within milliseconds, sending their control commands to the gearbox, brakes and steering rack.

Not all of the 35 teams qualified for the finals; during the testing before the race, teams faced tough tests on different courses. Axion Racing’s Spirit hit a blue Ford Taurus as it tried to make a turn into traffic.

“This is a historical moment,” said team leader Bill Kehaly, “it’s the first collision between a robotic vehicle and a human-driven car.” A German entrant did not fare much better: Team Lux from Hamburg unexpectedly made a turn — and headed straight towards a following vehicle which, tomake matters worse, had Darpa director Ted Tether on board.

In the end six teams remained — and the favourites lived up to their expectations. Junior crossed the finish line first to applause — that came too early. Carnegie Mellon’s more aggressive Chevy Tahoe, named Boss, had started much later than Stanford, but arrived only a few minutes later to achieve the best finishing time without any major traffic violations. And with that, the racing Herbie from the Walt Disney movies had come one step closer to reality.

Mission accomplished: It was impressive, but not exhilarating. When Darpa first announced the Grand Challenge four years ago, the military research organisation wanted to prove that it was technically possible to create a self-steering driving machine. That mission is now accomplished.

But on the way to this race, something got lost. The first events, in 2004 and 2005, were characterised by the feverish obsession of garage tinkerers, savvy engineers and even a horde of unruly high school students.

This time, some teams had the benefit budgets of several million dollars; they arrived with PR managers, their own photographers and industry representatives. The self-driving car has become big business — and with that, small teams cannot keep up any longer.

After the awards ceremony, second-placed Thrun commented, “These challenges may not be the most suitable way to advance this technology.” Tether agreed, “Darpa is in the business of starting things, not finishing them. It might be time for us to move on.”