Read time: 10 minutes

AI – Artificial Intelligence – has enjoyed a rapid rise to prominence in recent years, exploding across the wider world and developing even faster than people’s understanding of how to use it. Unlike many inventions, it comes with no ‘one-fits-all’ user manual and so many possibilities, the potential for which remains largely untapped.

Formula 1 is a perfect fit for AI. Not only is the sport an enthusiastic early adopter of new technology but it is also heavily data-driven, and AI thrives on data: the more, the better.

McLaren has always led F1’s push to develop new sensor technology and cutting-edge telemetry systems – but the volume of data coming off the modern F1 car (or swirling around the McLaren Technology Centre’s design, production, build and testing facilities) is far too much for a human to assimilate, and with time always a factor, there’s a natural tendency to focus on a very limited subset of data.

AI doesn’t do that: it rapidly takes everything in and sees the whole picture, discounting nothing. We’re still a sport dominated by human performance – but through our use of Dell Technologies’ AI Factory, we ensure those humans are better informed and better prepared than ever before. From the drawing office, to the production hall to the track, AI is working in the background to make us a better racing team.

How do we use Dell Technologies’ AI Factory to spot anomalies and mistakes?

In Formula 1, development isn't solely focused on making our cars faster, but on maximising every little detail and every possible differential imaginable. This includes ensuring the machinery is as safe, efficient and reliable as it possibly can be. Today, we’re discussing the latter.

F1 cars are prototypes: designed afresh for each season, subject to a constant onslaught of upgrades, such that they never race in the same configuration twice, always chasing that ill-defined edge of the performance envelope.

It’s in the nature of such machines to have a high rate of failure and, for decades, it was accepted F1 cars would occasionally – perhaps regularly – grind to a halt with mechanical problems. In 2005, McLaren won 10 of the 19 races, but also had five mechanical DNFs [Did Not Finish]. That’s more than in the most recent five seasons combined, despite the longer seasons and greater complexity of the cars. There will be problems - nothing is bulletproof - but the incidence of failures is dramatically lower, and a lot of that is down to our close partnership with Dell Technologies and their solutions.

For instance, their AI Factory helped the team in Monaco earlier this year. Having been quick through practice, Lando was mysteriously slow in the Q1 Qualifying session, facing the distinct possibility of an early elimination and any chance of scoring points on a circuit where overtaking is virtually impossible.

With time running out, telemetry coming off the car and analysed by Dell Technologies’ AI Factory indicated a loss of downforce but nothing physically wrong with the MCL38. The problem was… external. Lando was called into the pit, where an errant section of advertising banner was found lodged under the floor. That was removed and he had just enough time to get out, put in a clean lap, and qualify for Q2. A hugely important find.

“We often see things like this,” says Andrew McHutchon, head of Data Science. “Brake cooling is a good example. We’ll see brake temperatures becoming unstable and the team is trying to figure out what's happening. AI can help understand that there’s nothing wrong with the car – and therefore, the problem is likely to be debris in the duct.

“Knowing this means it’s something you can fix: the car will pit, the crew will scoop out the foreign object – visor tear-off, leaves, debris etc – and the car can continue. If you don’t diagnose the problem, the brakes can easily overheat and that’s race over.”

The telemetry being studied by the Dell Technologies AI Factory is the same as that being viewed by our engineers and systems technicians. Where AI comes into its own is being able to take a holistic view of that data, instantly making connections between disparate channels that would require detailed detective work on the part of a human who really doesn’t have the time to dig into it.

This isn’t AI-as-watchdog, programmed to bark once a certain set of parameters are breached. The nature of the sport demands something a little more intuitive.

“The whole idea with Formula 1 is nothing ever breaks twice,” says McHutchon. “After every single race, we look at any issues we've had, figure out what caused them, and do everything in our power to prevent them from ever happening again. So typically, issues and anomalies with the car are things we've never seen before.

“What we need from AI is something able to be intelligent: to know what looks normal, and then react when something happens outside the that range. Dell Technologies’ AI Factory looks at our data gathered over many seasons, it learns what normal F1 data looks like. This is the same process our engineers go through: they’ve worked with these cars for many years and understand what the data should look like. They become very good at playing ‘spot the difference’. AI is doing the same thing: looking at telemetry and finding patterns that don’t match.”

How do you we use Dell Technologies’ AI Factory to make predictions on things that might go wrong?

When something has gone wrong, a swift diagnosis is very useful – but prevention is always better than cure. Humans are instinctively good at seeing the signs of impending failure – noting when something feels off or sounds different. Motorsport has long-since taken that further, analysing oil samples between sessions for tell-tale contaminants, or assigning warning values to sensor data – but AI has moved that to another level: spotting worrying trends earlier – making connections between seemingly unrelated sensors that just wouldn’t be apparent to engineers looking at individual data channels in isolation.

Jude Hunt is a Specialist Data Scientist working at the MTC, using AI to deftly nip problems in the bud. “We need to know what's going to happen before it happens – and that's impossible without Dell Technologies’ AI Factory and hardware.

“We're working with so much data, it's coming in at such speed that no human is going to be able to look at that and predict everything that's going to happen. We're not fortune-tellers – but AI is the closest thing we have. We're working with hundreds of thousands of different channels during the race. AI can look across all these different channels, pointing out patterns that might be problematic.

“We’re talking about spotting anomalies that can be miniscule: perhaps a tiny spike somewhere, or a combination of a thousand different factors, or just a change in a pattern. A person might not be able to pick-up on that – but AI can.”

Does the Dell Technologies AI Factory make the operation of an F1 team easier, and less fraught? In a word: no… but F1 likes to be on the edge of the envelope and AI helps us move that edge a little further out every day.

Thanks to the Dell Technologies’ AI Factory, we can give everyone access to better, more insightful information, can extract more from the car and get closer to the theoretical maximum. It doesn’t diminish the amount of sweat on the brow, or reduce the joy in the high-fives after a job well done. For all of its innovation and technology, F1 remains a sport in which the people make the biggest difference.

As Official Technology Partner of the McLaren Group and McLaren Formula 1 Team, Dell Technologies brings its expertise as an innovation catalyst empowering McLaren to accelerate AI outcomes. Find out more here.