From giving fans insight into racing strategies to helping them predict the winners accurately—here’s how F1 uses AI and machine learning during the games.
Artificial intelligence (AI) and machine learning have been prominent features in Formula 1 (F1) racing since 2018. F1 has partnered with e-commerce giant Amazon to enhance its racing strategies, data tracking systems and digital broadcasts. With over 300 sensors in each F1 car, the car will generate and transmit 3 GB of telemetry data during the race and 1,500 data points per second. All this data is just numbers on a spreadsheet, which is meaningless to the average viewer unless the data analysis is presented in an organized and easy-to-understand manner.
As former Ferrari and Williams engineer Rob Smedley says, “It is not always clear what you’re watching [on TV]. There are lots of ways data can be used, and you need to use data to engage with people and show people stories you can’t always see.” As such, the collected data needs to be put into bite-sized chunks so that the average viewer can digest it. Let’s talk about how F1 incorporates Amazon’s AI and machine learning to enhance the viewing experience.
F1’s partnership with Amazon
Entering its fifth season with the Amazon partnership, F1’s data scientists have trained deep-learning models from Amazon SageMaker and Amazon Web Service (AWS) to analyze race performance statistics. Amazon SageMaker is a cloud machine-learning platform that builds, trains and deploys machine learning models to make accurate race predictions.
All this started when F1’s Managing Director, Ross Brawn, set out to improve the viewer experience during the race. He aims to provide fans with access to some of the data that is already available to teams in the pit. He also believes that AI and machine learning algorithms can minimize the gap between the average television viewer and the big F1 teams.
Some noteworthy AI features
Information about the technical aspects of racing, like exit speed, predicted pit stop strategy, over-taking difficulties and tire conditions, are shown during the race. These features allow F1 fans to gain more insights into the split-second decisions and strategies that F1 teams and drivers adopt. In the past, viewers could only speculate about the tire wear and conditions of F1 cars. However, with the help of machine learning and AI, information about the race performance of each car can be viewed during the race on the official Formula One website.
Similarly, during official races, drivers will be given a rating based on their qualifying lap time, race pace, tire management and other criteria. These data provide critical information about drivers’ performance on the grid, which can signify how teams can further develop their cars. One of the most notable graphic features shown during the race on screen is the predicted outcome for the qualifying and racing sessions. This feature provides a benchmark for the driver’s performance which it can be measured against. According to the Director of Innovation and Digital Technology at F1, Peter Samara, such machine learning systems “can generate powerful predictions to fans in real-time,” which can improve the viewing experience.
On top of that, F1 had released a separate driver rating score for their 2021 video game which takes into account each driver’s experience, racecraft, awareness, pace and overall rating and rates them from one to a hundred. Such scores are based on the drivers’ abilities during the actual race and can provide another benchmark for the driver’s performance. For instance, seven-time world champion Lewis Hamilton and 2021 world champion Max Verstappen both had a rating of 95.
The use of on-track F1 data for off-track research
The collaboration between F1 and Amazon happens both on and off track. Particularly in the research area, F1 and the Amazon Machine Learning Solutions Lab scientists have employed a data-driven model which can critically determine who the fastest driver might be. Since F1 started, which was around 70 years ago, lots of data has been accumulated. However, it would take a huge amount of time to manually compare every data of every F1 driver; but with the help of machine learning and AI technology, this age-old debate can finally come to an end.
Other uses of AI and machine learning
As a pioneer in implementing machine learning algorithms and AI into sports, F1 has led many sports leagues around the world to incorporate similar strategies to give fans a better viewing experience. For instance, the National Football League (NFL) uses AI and machine learning in AWS to bring advanced statistics to fans and players, aiming to improve players’ physical health and safety initiatives. Another example is the Association of Tennis Professionals (ATP) using machine learning to visually track tennis balls and highlight them, allowing for a better viewing experience.
Not only is AI used within sports sectors, but it is also used in different marketing campaigns. For instance, Red Bull Racing is hoping to include AI in its marketing campaign to better understand its fan base whilst providing the exact type of content an individual fan would want to see.
Future of AI and machine learning in F1
With the 2022 F1 season currently underway, F1 has begun to look into how AWS machine learning services can optimize the design and performance of the cars. The application of AWS is therefore not only limited to data collecting and analysis but also includes the production and design of the cars. This way, F1 can improve efficiency and lower the cost of production. Since F1 introduced a budget cap for the first time in 2021, each F1 team can only spend US$145 million on car design and development this year. Fortunately, with the help of machine learning, the production cost can be greatly lowered to cope with the budget cap, allowing each team to allocate their resources to other developing areas, like powertrain and durability.
- Formula 1 Virtual Grand Prix Series Achieves Record Numbers
- AI Tools for Businesses
- Job Sectors That Will Soon Be Dominated by Robots
Header image courtesy of Unsplash