How is artificial intelligence changing Formula 1?
Before delving into how F1 has benefited from artificial intelligence, it is worth understanding just how the AI industry is interacting with the sport – and why.
Alexandre Bonnet, the Lead ML Solutions Engineer for Encord told Autosport that the scope of what F1 offers as a platform for AI and machine learning is appealing to those who want to be involved.
“The reason F1 is so attractive is there’s such a wide variety of things you can do with all the different types of technology, all the way from the manufacturing stage, like parts, like quality analysis, like industrial process optimisation, like robotic manufacturing all the way through to the actual driving pieces,” he said.
“Then the optimisation of the actual vehicle designs, things like the biomimicry stuff that’s come out over the last few years, all through to the actual televisation, that is, again, a whole other round of computer vision and AI work that’s going on.”
Bonnet also feels AI firms can benefit from their association with a sport that has always been pioneering in its use of technology.
“AI companies want to be involved because it’s a complex sport, you are on the cutting edge where it’s the fastest cars, it’s the fastest drivers, all that sort of stuff, which isn’t really true of any other well-known sport,” he added.
“F1 teams are specialised in car manufacturing and of course, they do have in-house AI talent but they do also need to draw on that external expertise of the most up to date and innovative machine learning systems, I think it is a symbiotic relationship in that sense.”
Lewis Hamilton, Mercedes W14
Photo by: Erik Junius
Formula 1 Teams
Formula 1 has forever been a sport driven by data; lap times, deltas, top speeds, and tyre temperatures.
The sheer reams of information available to a team looking to crunch the numbers and make the most of the results had previously made the task a difficult, and time-consuming effort.
Now, though, AI can sort through such statistics in seconds as Andrew McHutchon – Head of Data Science at McLaren – explained in a recent interview on the team website.
“Previously, we’ve collected data that we didn’t know what to do with, but now, with AI and by working with Dell Technologies’ AI Factory, we can process the data in a much richer way to extract meaningful learnings from it,” he said.
“If it’s a decision related to pitting, you may only have a third of a lap before the car passes the pit lane, and after that, you’ve lost your opportunity, so you need to be fast. You could have terabytes and terabytes to analyse, which could take half a day or more to answer just one question without AI.”
Formula 1 is, at its heart, all about being the fastest and that speed of AI assists off the track as well as on it.
“Even when it comes to the team back at the factory working on the car’s development, speed matters,” added McHutchon.
Charles Leclerc, Scuderia Ferrari, 1st position, lifts the winners trophy
Photo by: Sam Bloxham / Motorsport Images
“You may have five questions, and if it takes you half a day to answer each, that slows everything down. AI speeds all of that up, and the faster we can answer these questions, the faster we can develop the car and the more likely we are to win championships.”
Formula 1 fans
So, AI can certainly play a role in the racing of the future – but what about those spectating?
AWS has been an innovation technology partner of Formula 1 since 2018 and a year later launched F1 Insights.
Using just a slither of the one million data points churned out by every car, every second there are now 23 F1 Insights available to producers of television feeds, including predicted pit strategy, striking distance and undercuts.
These are being utilised to aid the viewers’ experience, as Neil Ralph, Principal Sports Partnership Manager of AWS explains.
“Drive To Survive has driven a huge number of new fans into Formula 1 and they are trying to figure out how an F1 race works. We use the data in the sport to unpack the complexities to that spectrum of fandom,” he told Autosport.
“It is a lot harder to work out how an F1 race works than if you compare it to, say, a football match where a lot of the attention is on one area of the pitch whereas F1 has a 5km track with 20 cars spread across it so you can only cover so much using video.
Lando Norris, McLaren MCL38, Max Verstappen, Red Bull Racing RB20, Lewis Hamilton, Mercedes F1 W15, George Russell, Mercedes F1 W15, the rest of the field at the start
Photo by: Zak Mauger / Motorsport Images
“So the addition of using data to tell stories helps and increases that fan engagement for new fans but also adding to the experience of long-standing fans.
“What AI allowed us to do was unpack race strategies, pit windows and undercuts – by using those data-driven on-screen graphics we can help every fan understand what is happening and allowing commentators around the world to keep the levels of excitement going.”
David Croft, Sky Sports F1’s lead commentator since 2012, is one such broadcaster now given the chance to utilise the insights offered up by AWS.
“To me, AI is a massive help to the commentators and the audience to help understand how a story is developing, what I don’t ever want to see is it telling people how the story will end,” he told Autosport.
“Live sport in unpredictable and joyous and heartbreaking in equal measure depending on who you follow, so as long as it never spoils the outcome then what we are getting is fantastic.
“It doesn’t necessarily impact massively on my job, I’m commentating, I’m in the moment, absorbing data and relaying that to tell the story.
“There is only so much I can look at! Where I think it is really helpful is in situations like, in the past, Martin (Brundle) and I would make a guestimate to when we thought a driver would be right behind the other.
Charles Leclerc, Ferrari SF-24, Lewis Hamilton, Mercedes F1 W15, Sergio Perez, Red Bull Racing RB20, Oscar Piastri, McLaren MCL38, the rest of the field at the start
Photo by: Michael Potts / Motorsport Images
“Now that information is coming through so everyone knows that, and they know it from real-time data and AI simulations and more often than not it is very accurate. Also, when you say ‘striking distance is in four laps’ that doesn’t mean the story is over. It means, get on the edge of your seat and start getting excited and we can start to ramp up the narrative.”
Croft may be worried about how far AI can go in predicting Formula 1, but just how much further can it take the viewing experience? Ralph has his own ideas on that.
“We don’t see all 23 of those race insights every weekend but as we talk about personalisation and giving fans the choice what they see, perhaps in the fullness of time it isn’t the F1 technical production team that decides which graphics are available – the insights are available to anyone and you decide what you want to see,” he added.
“As we look at the future, personalisation of broadcast is where people are looking, allowing viewers to choose how much of this data is part of their broadcast, maybe via second-screen experiences, giving the choice of just watching the cars or having a huge amount of data available.”
Formula 1 drivers
With audiences satiated and car design, performance and strategy taken care off, it seems the only thing in Formula 1 that will remain sacrosanct in an increasingly AI age is the role of the driver, or is it?
When Autosport puts this theory to Kevin Magnussen of Haas, he replies: “Are we?
“If you’re just looking for efficiency eventually AI is going to drive the car better than us, there is no question about that. But then it is not really entertainment any more, I would think that would be really boring, to watch a Formula 1 race driven by computers, I wouldn’t care about that.
Kevin Magnussen, Haas F1 Team
Photo by: Simon Galloway / Motorsport Images
“You can relate to the driver in the car and what he is going through, the skill a human has to drive a Formula 1 car is fascinating because you can relate.
“When I watch footballers I think they are incredible, I know how good I am at kicking a ball and the level they are at is fascinating and I think that relationship has to be there otherwise there is no interest.”
With the help of AI, teams are developing newer and faster ways of analysing data that could prove pivotal to a race weekend result.
All of those statistics can be fed through as much software as possible but all of the results that are produced end up with the driver behind the wheel.
A far cry from racing of yesteryear, the new way of approaching marginal gains is to employ AI to aid strategy decisions and pit calls, something which Kevin Magnussen of Haas believes is only going to develop over time.
“Those things are already in effect in some way, but it is done manually, it is running simulations on millions of different scenarios and coming up with a suggested strategy,” he told Autosport.
“In the future, you could see your strategist be a computer, further into the future you could see AI setting up your car and developing your aerodynamics but take out that human element and people watching would not find it interesting.”
