Racing algorithms

Ten cars roll onto the racetrack and take up their positions at the start. No rumbling engines disturb the silence. The signal turns green. Five cars race clockwise around the track while the other five make their laps in the opposite direction. The electrically powered motors propel the cars to speeds of up to 320 kilometers per hour before they meet at an extremely tight corner halfway round the track. At this point, the two sets of five cars are racing head on toward each other. But the dreaded prospect of a massive pile-up doesn’t materialize. Elegantly dodging each other, the racers avoid collision at breathtaking speeds by mere millimeters. What sounds like utter madness may be a very real possibility for the not-too-distant future. That’s because the race cars are self-driving—no human sits behind the wheel. At least that’s the vision for Roborace in Banbury, a town 120 kilometers northwest of London. Russian businessman Denis Sverdlov established the company with finance from his Kinetik investment fund, which is valued at USD 500 million.

Intended as a supporting series to the FIA Formula E, Roborace aims to be the first international championship for autonomously driven cars. Roborace cars—that is, both the DevBot development vehicle, which can be operated by either a human driver or an artificial intelligence system and the driverless Robocar without a cockpit or steering wheel—have already performed on racetracks in public. So far, the company has three DevBot cars and two Robocars in addition to two non-roadworthy Robocars that were built purely for demonstration purposes. A full fledged racing series would only be conceivable in three to five years’ time at the earliest. But Roborace doesn’t just plan to serve up high-speed driverless action. As Chief Strategy Officer and UK native Bryn Balcombe explains, “Roborace rests on three developmental pillars—electric mobility, networking technologies and autonomous driving. The challenge the series faces is finding the middle ground between sporting entertainment and developing autonomous driving technology that’s truly relevant to everyday life.”

Customarily, motorsport at its highest level paves the way for automotive technology in “conventional” cars. Yet Balcombe, who has pursued a career almost exclusively in Formula 1, strongly believes that this no longer applies today. While electronic driving aids are prohibited in motorsport, it’s precisely this type of technology that needs to be advanced for autonomous driving. “Roborace is a platform on which the technology necessary for autonomous driving—technology that will make our roads far safer in the future—can be tested,” he explains. Perhaps, as a result, car manufacturers will eventually enter their own vehicles in the series. At first, however, Roborace will provide the hardware while the various teams supply their own software—much as is the case in Formula E. Sounds like a championship for programmers. Balcombe disagrees. “The best way to think of Roborace is as a drivers’ championship—even though in this instance, of course, there are no drivers. Artificial intelligence (AI) makes all the difference, since it’s the caliber of the software that determines which team triumphs—especially when all Robocars are otherwise identical.” When it comes to the different software sets for autonomous driving, Balcombe is confident that they will produce very diverse driving styles. “The software system’s performance varies, above all because machine learning plays a role here. Although we are all bound by the same rules, people, too, have different driving styles. It’s the same with artificially intelligent cars. I believe that carmakers are going to develop their brand identities far beyond vehicle shape and design.” You’d think that a world of autonomous driving would send shivers down the spines of people like Lucas di Grassi who earn their living as a racing driver. But the Brazilian, Formula E Audi driver and reigning world champion, has a different take: “While Roborace will never replace motorsport, we still need a racing series to push the technological envelope as regards autonomous driving, networking and electric drives.” Fully convinced of the autonomous racing series’ potential, Lucas di Grassi has accepted the post as Roborace CEO.

Doesn’t that amount to something of a betrayal of motorsport’s die-hard fans? “Not at all. People today still get excited about horse racing even though it’s been a long time since we used horses to get from A to B. I also doubt that anyone has stopped playing chess just because the supercomputer Deep Blue beat Garry Kasparov in New York in 1997. So why should we turn our backs on motorsport just because racing cars piloted by artificial intelligence can go faster than with a human driver? But are we really likely to reach the point where autonomous cars can outperform racing drivers on the circuit? “It’s already happened. At present, the Robocar can easily outpace your average driver. Now, the challenge is whether it can beat a professional racing driver. And that’s just a matter of time.” Di Grassi, too, emphasizes that Roborace will help to make the world a safer place by reducing the number of road accidents. “There are no limits to what we can achieve with Roborace. And best of all, there’s no danger to human life in the process. In this way, the technology can keep on evolving and improving until it’s fully matured. If it works, great. If it doesn’t and there’s a crash, well that’s a financial loss but no one was hurt.” Di Grassi knows what he’s talking about: In the first lap of qualifying for the 2017 GT World Cup in Macau, his Audi R8 was involved in a huge pile-up of twelve cars. The accident happened when the car in the lead got stuck on the edge of the track in a blind corner and the drivers coming up behind couldn’t see the crash, let alone react to avoid it. With the help of networking technology, which connects cars to each other and their environment, such accidents could potentially be prevented in the future.

Yet countless hours of work still lie ahead before humanity can reap the many benefits of autonomous machines. No one knows this better than Roborace vehicle science engineer Teena Gade from the UK. At a very young age, both Teena and her sister Leena, who is a racing engineer and helped Audi Sport to victory at the 24 Hours of Le Mans, developed a passion for motorsport and science. Following stints on the Subaru World Rally team and Williams F1 team, Teena is now responsible for the Roborace simulation program. “There are already cars capable of staying within lane markings on highways without human intervention. But we need to develop tomorrow’s AI cars so that they can also deal with borderline cases,” says Teena Gade. “How does an AI car respond to an unexpected obstacle on the racetrack? Using a simulator, we need to put the Robocar through scenarios in order to train it to respond appropriately to situations.” Gade knows these are the same challenges that autonomous vehicles will also face in day-to-day traffic: “When AI cars are legally approved for use on public roads, most other vehicles will still be driven by people. Other road users will soon notice the AI car but they are not going to take things slower in the morning rush hour just to considerately let an autonomous vehicle into a rotary. We’ll have to put the AI car through its paces to ensure that it moves briskly with the flow of traffic and perhaps even occasionally, metaphorically speaking, can elbow its way in.”