HOW Formula 1 Is Using The Power Of AI & Cloud
The world’s most sophisticated sport is Formula 1 (F1) racing. The winning combination is the result of the perfection of both humans and machines. This combination makes F1 racing, or more specifically, the driver talent, so challenging to comprehend. Without the strength of Benetton and then Ferrari and the combined technological brilliance behind those teams, how many races or championships would Michael Schumacher have won? If Lewis Hamilton’s career had taken a different path and he had been limited to back-of-the-grid gear, could we really have witnessed him win seven World Championships?
With drivers reaching speeds of 230 mph (370.15 km/h), making pit stops in less than two seconds, and flying in curves with 5 G forces, FORMULA 1 needs technology as fast as it does. It is a battle between the best pilots in the world but also between some of the most innovative engineers on the planet. Using Cloud Computing technology, F1 uses innovative technologies such as machine learning (ML) models and high-performance computing (HPC) to digitally transform the sport.
Everything Starts With Data
Each F1 car has 300 sensors that generate 1.1 million telemetry data points per second that are transmitted from cars to boxes. This real-time data is combined with more than 70 years of historical racing data stored on an S3 bucket for valuable information to inform, educate and enrich the experience of fans and provide more data on choosing a winning race strategy on the track.
With the collection of historical data and its use to train complex machine learning algorithms of AWS SageMaker, the F1 can predict the results of the race strategy with increasing precision for teams, cars, and riders. These models can predict future scenarios with up-to-date data in real-time as GRAND PRIX races unfold to provide an enriching and engaging experience for fans.
In this graphic model, we can view how the F1 provides and processes key performance data by machine learning through a lot of AWS’s Machine-learning services, including an AWS SageMaker, AWS Lambda, and AWS analytics services:
source: https://aws.amazon.com/sports/f1/
F1 and Cloud Computing Services
A close partnership with AWS, which offers cloud computing power for everything from computer-aided car design to an astounding array of data analytics and visualizations for F1 fans, is at the core of Formula One Group’s technology. Using 1,150 compute cores to run thorough simulations, and more than 550 million data points to model the effect of one car’s aerodynamic contrail on another, a Computational Fluid Dynamics (CFD) project was used to develop the car design for the 2023 season. For example, with EC2 c5n instances, the project ran for six months while providing supercomputer-like performance at a fraction of the price.
Over 70 years' worth of historical race data are combined with real-time data and kept on Amazon S3. A crucial step was moving Formula 1’s extensive archive of 70 years’ worth of photos, audio, and video into the cloud. F1 was able to transfer 150,000 hours of content from a tape archive into the AWS Cloud with the help of the AWS Media2Cloud service, which also allowed the assets to be indexed and tagged to offer historical context for race analysis. An overview, as well as a look at some workstations and IT equipment that power F1’s technical infrastructure, are given in this article.
The following diagram illustrates how the pipeline for Inference Models works in the treatment of the predictions, using a model feeding by the historical data:
source: https://aws.amazon.com/sports/f1/
Running CFD Simulations on AWS
The F1 moved its CFD simulation environment to a high-performance computing (HPC) platform on Amazon Web Services (AWS) to address its challenges. The company uses instances of Amazon Elastic Compute Cloud (Amazon EC2) to run efficient and complex simulations that show the turbulence of the car trail and the impact on the cars that follow them. The F1 chose to use a combination of Amazon EC2 C5n instances and the new C6g instances based on AWS Graviton2 to obtain maximum flexibility in terms of price and performance, depending on the job. Also, the F1 used AWS ParallelCluster to automate the supply of different clusters of HPC for the optimization of the work.
Thanks to Cloud Computing, F1 has reduced CFD simulation time by 80%, from 60 to 12 hours. They can now schedule a simulation to run at night and have the results the next morning. With faster results, They can perform more simulations in general and get to the final car design faster.
The F1 analysts can view any problems most quickly because they can connect via AWS DataSync within the Remote Technical Centre. Refer to the next diagram for a visual context of this flow:
source: https://aws.amazon.com/sports/f1/
Benefits achieved by Formula 1
- Reduced simulation time by 80% from 60 hours to 12 hours.
- Reduced cost of operation by 30% as compared to on-premise using an on-demand service model of the cloud.
- Reduced downforce loss from 50% to 15%, increasing stability and driver safety.
- Digitally transformed the business by innovating car simulation and fan engagement.
The F1 Teams and Cloud
Each of the F1 racing teams closely collaborates with technology partners in addition to F1’s corporate technology relationships on everything from on-track performance to fan experiences (including future plans for virtual reality experiences) and eSports offerings for gamers. The top teams and their technology are shown below.
HPE’s Formula 1 partnership with Mercedes-AMG Petronas
The Mercedes-AMG Petronas Formula 1 car’s complex race day simulations now only require one day to complete, thanks to HPE’s high-performance compute Apollo system. The race engineers and strategy team can now simulate race day on Friday and even Saturday morning before the race on Sunday.
Races are won on the track, but championships are won at the factory. The team generates up to 50 TB of CFD data per week prior to race events by running millions of simulations. To accommodate the team’s workload, high-performance computing systems are needed. This resume is a formula by HPE for the treatment of those simulations.
In the next picture, HPE explains how the Apollo 6500 Gen10 System and the rest of the high-performance compute clusters help the Mercedes team in support of simulation and prediction processes:
source: https://www.hpe.com/
Red Bull Racing Honda and Oracle Partner
In Formula 1 (F1) race preparation and race-day strategy, cloud computing now plays a major role thanks to a partnership between Oracle and Red Bull Racing. Red Bull Racing now runs billions of race simulations every weekend thanks to cloud computing. Future initiatives resulting from the race-day analysis include creating a brand-new F1 powertrain for 2026.
By switching to Oracle’s cloud-based platform, Red Bull Racing was able to double the speed and 25-fold the number of simulations. To improve the way data is used throughout their business, from on-track activities to giving the Team’s international fan base more information, they will make use of Oracle Cloud Infrastructure’s (OCI) machine learning and data analytics capabilities. The team will be able to hone its already imposing competitive edge thanks to OCI-powered capabilities.
In the next infrastructure diagram, Oracle explains how the Red Bull team uses Kubernetes with OCI, Arm cores, and HPC for developing and deploying solutions faster:
Source: https://www.oracle.com/dk/cloud/technical-case-studies/
Summary
The F1 is a data-based sport that generates 3 GB of data per race from 120 sensors in each car at 1500 data points per second. With a strong partnership between the most important cloud providers, Formula 1 can gather and analyze crucial performance information for each car during each turn of the circuits for the various grand prizes by transmitting real-time race data to cloud computing.
Thanks to data models, Formula 1 engineers are training deep learning models with 65 years of historical racing data to extract key race performance statistics, make race predictions, and provide followers information about second-time decisions and strategies adopted by teams and drivers.
F1 can assess a driver’s performance or level of effort by implementing cutting-edge AI machine learning. Fans can learn about the internal operations of their preferred teams and drivers thanks to Formula 1’s use of digital platforms and television broadcasts to share this information.
The F1 keeps coming up with new ideas, together with the Infra Services team and the Top Cloud Providers Solutions Team, to quicken the development of F1 Insights through use case prototyping and the creation of fresh proofs of concept.
The Data Engineering team then assists F1 in putting models into production and integrating them into the F1 infrastructure.
