How Driverless Car Racing Works

Obviously, driverless cars weren't invented just so we could race them around a track. Sure, that's pretty awesome, but the real reason was more practical. If cars drove themselves, people who rely on others for transportation because of age or disability could have much more independence, not to mention your average commuter, who could have more time to relax. As an added bonus, driverless cars could ease congestion, reduce accidents, lower fuel consumption and even alleviate the demand for parking.

The idea for a driverless car can be traced all the way back to Leonardo da Vinci, who sketched a pre-programmable cart in 1478 (seriously, what didn't that guy think of?). But it wasn't until General Motors' "Futurama" exhibit premiered at the 1939 World's Fair that driverless cars really gained public attention. The exhibit was a scale model of how the world might look in 1960, as envisioned by GM, and prominent on its tiny freeways were cars that drove themselves. Early attempts to actually develop this technology in the 1950s focused on so-called smart highways, which were roads embedded with steel cables that cars could detect and follow [sources: Vanderbilt and Weber].

Today's driverless cars, which navigate roads using sensors and computers to scan and interpret the environment, came out of the 1970s. That's when engineers from Japan's Tsukuba Mechanical Engineering Laboratory successfully programmed a car outfitted with two cameras and an analog computer to follow white road markers at the blazing speed of 18.6 miles per hour (30 kilometers per hour), though a steel rail helped it along [source: Weber]. Over the next couple of decades, however, universities and companies across the world took advantage of improvements in computer technology to create driverless cars that actually performed quite well in real-world road tests.

In 2003 the United States military, under the Defense Advanced Research Projects Agency (DARPA), decided these cars were ready to race. More specifically, they hoped that challenging driverless car researchers to compete against one another would spur innovation in a technology that could potentially have important military applications. Thus the DARPA Grand Challenge was born. With a $1 million prize on the line, teams gathered in the California desert in 2004 to unleash their ultimate autonomous speed machines, and the results were, well, interesting [source: Shipley]. But before we get into that, let's talk about the technology it takes to build a driverless race car.

It's probably not surprising that the technology in driverless race cars is similar to that in your run-of-the-mill autonomous (or even semi-autonomous) car. That's because navigating on a racetrack and cruising down a highway are alike in many fundamental ways: Both require you to keep away from other cars, stay on a designated course and avoid unexpected obstacles. Even more advanced racing skills, like taking a corner as fast as possible without spinning out, can be likened to taking a turn on an icy freeway. Still, driverless race cars face unique challenges, and engineers have come up with some pretty genius ways to deal with them.

A typical driverless car uses a variety of sensors to observe its surroundings. Radar sensors placed around the car can detect the vehicle's position in relationship to the cars around it. Video cameras help keep track of nearby traffic, but they also read traffic lights and road signs, and help the car avoid obstacles like pedestrians. Lidar — or light radar — bounces pulses of light off the car's surroundings to create a three-dimensional view that's useful for identifying lane markings and the edge of the road. A computer pulls all this data together and manipulates the steering, acceleration and braking accordingly [source: Armstrong].

Racing demands even more from a driverless car. Given the higher speeds, acceleration and braking are much harder, and there's far less reaction time. Techniques like determining the best line through a turn, fighting for position and compensating for load shifting during steering are also necessary. Many of these skills are instinctual for race car drivers, but driverless cars can get it right only if they're programmed with the proper algorithm [source: Adams].

So what's a racing-challenged engineer to do?

They take a look into the heads of race car drivers — literally. One research team at Stanford actually hooked drivers up to electrodes that measured brain activity as they raced against other cars. Information gleaned from this experiment helped the engineers fine-tune the computer systems of the driverless race cars. For example, by examining the human drivers' instinctive response when the vehicle spins out on a turn, engineers were able to improve the timing of the software that helps stabilize the car [source: Knapton].

Has all this talk of driverless car racing got your inner geek ready for the green flag? Well, don't get too excited, because there are still some kinks to work out before the sport is ready for prime time.

Sure, engineers have been racing these vehicles for more than a decade, but it hasn't always been action-packed. Take the first official driverless car race, an off-road contest known as the 2004 DARPA Grand Challenge (we mentioned it earlier but left you hanging about the outcome). Of the 15 teams that started, only six actually made it out of the starting chute without going bonkers. Those that made it out were able to travel between 1.2 and 7.4 miles (1.9 to 11.9 kilometers) before getting stuck on a rock, careening into a fence or meeting some other inglorious demise. Those results are even more cringe-worthy when you consider the course was 142 miles long [source: Hooper].

To be fair, the DARPA Grand Challenge is about encouraging technological advancement, and by that measure the 2005 race did not disappoint. Twenty-three vehicles entered this time, and four managed to make it down the 132-mile course within the 10-hour time limit. "Stanley," the car from Stanford University, finished first with a time of six hours, 53 minutes, and 58 seconds. It was an amazing technological accomplishment, but at an average speed of 19.1 miles per hour (30.7 kilometers per hour), it was hardly the heart-pounding action race fans are used to [source: Hanlon].

The final DARPA driverless car race was the Urban Challenge held in 2007. This time Carnegie Mellon's team beat out 10 other competitors to take home the $2 million first prize, completing the 60-mile course 20 minutes ahead of the next finisher. Again, speeds were slow: only 14 miles per hour (22.5 kilometers per hour) on average. Perhaps the biggest bummer for race fans, however, was that the cars were judged on how well they followed traffic rules. What kind of race is that? [source: Sofge]

Race enthusiasts will be glad to know that engineers are now working on driverless race cars that can go much, much faster. Stanford University's Audi TTS, nicknamed "Shelley," has gotten a lot of attention lately — and for good reason. In February 2015 it became the first driverless race car to beat a human race car driver, beating out amateur touring class champion David Vodden at California's Thunderhill Raceway Park by 0.4 seconds [source: Knapton].

After hearing about the current state of driverless car racing, motorsport fans are probably thinking, "That's pretty cool, but when are we going to see some REAL racing?" Great question. And you'll be glad to hear that the answer is soon.

In November 2015, Formula E, a racing class for electric cars, announced Roborace, an international racing circuit for driverless electric vehicles. The plan is to hold these races before each Formula E event, beginning with the 2016-2017 season. Roborace will feature 10 teams using identical cars, so the competition will actually come down to the different sensors and computer algorithms team engineers create to make the car autonomous. And the best news for racing fans? One of the event's founders promises speeds as high as 186 miles per hour (299.3 kilometers per hour) [source: Davies].

But will anyone watch it? That's the big question raised by many of Roborace's critics. These doubters feel that if you take the human element out of racing, it becomes less interesting and perhaps doesn't even qualify as a sport. It's the strategic miscalculations and the crashes that people tune in to see, so what will happen if computers are fine-tuned to avoid these mistakes? What will racing be without the drivers' emotional celebrations in the winner's circle? There's even some concern that driverless cars will become so good at racing that human drivers will become obsolete.

Not everyone agrees, however. Supporters argue that this type of competition will actually make the sport much more appealing by attracting a new type of fan that didn't really care much for racing before. And as for the human element, Roborace advocates don't see that going away, either. They hope the engineers will take on that role, finally enjoying some of the spotlight after being holed away in a laboratory for much of their careers [source: Fagnan].

In the end, the head of Formula E doesn't see his Roborace as the end of motorsports. Rather, he wants fans and competitors to get excited about innovation — and maybe have a little fun in the process [source: Khorounzhiy].