Another reason Ducati is all-powerful – it’s solving MotoGP’s biggest riddle
MotoGP
How Ducati uses artificial intelligence and machine learning to solve MotoGP’s greatest riddle: the exact interaction of man and machine on the racetrack
The Ducati seat camera films the rider’s every movement – torso, arms and legs. The footage is used to calculate the rider’s centre of gravity and so on. This is Danilo Petrucci head towards Turn 12 at Sepang
Whenever I write a MotoGP tech story I’m fully aware that I’m barely scraping the surface, because pitlane engineers keep their biggest secrets increasingly close to their chests. This is probably just as well, especially regarding the computing aspects of MotoGP, because to make sense of this stuff you really need a PhD in electronics engineering.
Recently I wrote a blog about Aprilia using motion-caption technology to precisely record every movement riders make aboard its RS-GP machines, because this is bike racing’s most unexplored area, its greatest unknown.
The Aprilia system employs the technology used in the movie industry: actors wear suits full of sensors, their movements are filmed, then an alternative person or creature – Gollum in Lord of the Rings, for example – is uploaded to inhabit the actor’s movements.
“Rider movements cannot be recorded with conventional technology”
Rider position and inputs are motorcycle racing’s last remaining riddle and this is a vital area, because how riders move around their motorcycles and what kind of loads they put into the handlebars and footpegs are pretty much the most important factors in bike racing, especially at MotoGP level, where the riders are searching for thousandths of a second at every corner by minutely adjusting body position.
Only once engineers have this information can they fully understand the dynamics of motorcycle racing, so this is a huge leap forward for the sport.
Ducati has been doing its own work in this area for at least five years, which is one reason why the factory is so dominant, for example, filling the first eight places with its Desmosedicis in Saturday’s Thai grand prix sprint race. Ducati engineers never talk about his work, but its parent company Volkswagen has been more open
Christoph Witte’s interview with Volkswagen Data Lab scientists Marc Hilbert and Yury Dzerin for Sigs Datacom tells us exactly how VW is helping Ducati solve the riddle of man-machine interaction, to make the Desmosedici and its riders even faster than they already are.
Ducati’s all-seeing eye – looks like a standard Dorna onboard camera, but inside it’s recording ultra-high-resolution footage for Ducati engineers
Oxley
This isn’t the only way that Ducati’s MotoGP project benefits from being part of the VW family, which also includes Bentley, Porsche, Lamborghini and Audi, which is directly responsible for Ducati.
Two of the Desmosedici’s most important technical developments – the tuned mass-damper (which damps out chatter) and ride-height device (which increase acceleration and top speed) – were developed by physicist Robin Tuluie, who was seconded to Ducati from Bentley, where he was vehicle technology director and responsible for the VW group’s simulation strategies and digital-twin technologies – making virtual replicas of cars, motorcycles and anything else.
The Volkswagen Data Lab is based in Munich and is staffed by around 80 experts working in artificial intelligence (AI), machine learning (ML) and data analytics. AI and ML are hugely important in motorcycle and car racing, because they use the massive power of big data and modern computing to help engineers make better decisions, quicker.
Three of Ducati’s game-changing MotoGP technologies were created by a MotoGP engineering genius you’ve probably never heard of…
By
Mat Oxley
“Ducati approached us with a task that they could not solve with conventional means,” Hilbert tells Witte. “The aim is to make the rider’s movement on the motorcycle measurable and to use it to simulate the race – and to optimally adjust the motorcycle to the track and rider behaviour on the basis of the simulation.
“The rider’s movements are not linear, they are very individual, driver A moves differently than driver B. The movements cannot be recorded with conventional measurement technology. That’s why we were asked whether the problem could be solved with artificial intelligence.”
The data for this analysis comes from the forward-facing seat cameras fitted to all the fastest Desmosedicis. Dorna fits these cameras to most MotoGP bikes, but the Ducati cameras are mega-high definition, so they can film the rider’s every movement with excellent clarity.
Each of Ducati’s seat cams are recording every movement made by Jorge Martin, Marc Marquez and Pecco Bagnaia at Le Mans
Michelin
These videos are fed frame by frame into high-performance computers where they are processed and analysed by a deep-learning algorithm.
This is all part of MotoGP’s high-tech revolution of artificial intelligence and its subsets: machine learning, deep learning and geometric deep learning.
What do these strange things do? And why now? They crunch the numbers and think better, faster and more originally than humans. And it’s only really happening now thanks to developments which allow the vast amount of data required to be captured, analysed and acted upon incredibly quickly.
The rider accounts for around a third of the combined mass of a MotoGP bike. Engineers have tons of bike data but they need to know where the rider is sat and what he’s doing
By
Mat Oxley
Ducati, no surprise, is racing ahead in all these technologies, using VW’s massive resources and also working closely with factory-team sponsor Lenovo computers. These technologies are vital, especially because MotoGP is getting so technically complicated, while there’s less time allowed for testing and the sprint format has reduced practice time, so finding ways to reach solutions to problems in the few hours before the next practice session or race is more important than ever.
Deep learning – used to get the most out of Ducati’s seat-cam footage – is an AI method that teaches computers to process data in a way that’s inspired by the human brain. It uses so-called neural networks of digital neurons and synapses that learn from experience and find answers from complex and seemingly unrelated sets of data. In other words, it can be many, many times cleverer than a human brain.
Ducati’s seat-cam algorithm quickly works out the rider’s position – torso, arms and legs – at every point of the race track, and from that info it works out the rider’s centre of gravity, which allows engineers to make better decisions about machine balance, electronics, tyre choice and so on. It also allows Ducati riders to exactly analyse their position on the bike compared to their fellow Desmosedici riders, so they can make improvements to their riding technique.
Will any of this benefit the everyday road rider? VW and Ducati say it could, because it will further refine the range of electronic settings that are increasingly available to road riders.
