It's long The path from pedal bikes to Formula 1. But that's exactly the quantum leap that AI startup Neural Concept and its co-founder and CEO, Pierre Paquet, have made in just six years.
In 2018, the company's fledgling program helped develop the world's most aerodynamic bike. Today, four out of 10 Formula 1 teams use an evolution of that same technology.
Along the way, Baqué has inked contracts with aviation suppliers like Airbus and Safran, securing a $9.1 million Series A raise in 2022. Switzerland-based Neural Concept, which has 50 employees, is currently working on a Series B round While its software helps in the historic F1 championship. Teams like Williams Racing are finding their way to the top of the world's premier form of motorsport.
However, as Formula 1 cars rely on 1,000 hp V6 hybrid engines, Paquet's first practical application of this technology was to use human energy.
Power pedal
In 2018, Baki was studying at the Computer Vision Laboratory of the École Polytechnique Fédérale de Lausanne, where he was working on applying machine learning techniques to 3D problems.
“I got in touch with this guy who was leading this team, and he was designing the sixth or seventh generation of bikes, and their goal was to break the world bike speed record,” Paquet said. That man was Guillaume DeFrance, and the team was IUT Annecy from the University of Savoie Mont Blanc. The cycling team has already gone through six iterations of bike designs.
“After two days, I was back at it looking almost like the current world record holder,” Paquet said. The team was impressed, and asked for more iterations. The result, according to Paquet, was “the most aerodynamic bike in the world at the moment.”
That's a strong statement, but it's backed up by several world records set in 2019. We're not talking about aero-shaped down tubes or dimpled edges to reduce drag. This bike is fully covered, with the rider sweating in a composite cocoon, completely protected from the wind.
The underlying technology is a product called Neural Concept Shape, or NCS. It is a machine learning-based system that provides aerodynamic suggestions and recommendations. It fits into the broad field of computational fluid dynamics (CFD), where highly trained engineers use advanced software packages to run 3D aerodynamic simulations.
CFDs are much faster than carving up physical models and throwing them into wind tunnels. However, it is also highly system-driven and largely dependent on humans making good decisions.
In essence, NCS helps engineers avoid potential aerodynamic hazards while pushing them in directions they may not have considered. In Co-Pilot Mode, the engineer can load an existing 3D shape, providing a starting point, for example.
NCS will then search its neural network to suggest improvements, modifications, and potential paths in a 3D choose-your-own-adventure game. The human engineer then selects the most promising proposals and subjects them to further testing and refinement, repeating his path to aerodynamic glory.
Not just “fool the wind”
NCS is useful not only for racing but also in the automotive and aerospace industries. “The path to widespread adoption in these types of companies is slow,” Paquet said of working in the somewhat conservative airline industry. “This way we are starting to work more with the automotive industry, where the needs are more pressing and will change quickly.”
Neural Concept has secured contracts with several global suppliers, including Bosch and Mahle. Aerodynamics is becoming increasingly key in the automotive world, as manufacturers look for more aerodynamic cars that offer the greatest possible range from a given sized battery pack.
But it's not just about fooling the wind. NCS is also being used in developing things like battery cooling panels that, if made more efficient, could keep the battery at its ideal temperature without draining a lot of power in the process. “There are huge gains to be made,” Paquet said, which means a broader scope.
While the ultimate proving ground for these technologies is always the road, the ultimate laboratory is Formula 1. A global phenomenon in motorsport since 1950, Formula 1 is currently experiencing an unprecedented wave of popularity.
The power of Netflix
The Netflix series “Formula 1: Drive to Survive” brought the excitement of Formula 1 to a whole new audience. While this series focuses on politics and drama between teams, success on the track has more to do with aerodynamics. This is where neurological concepts come into play.
Baqué started watching Formula 1 before Netflix was even a twinkle in Reed Hastings' eye. “I've always watched it, since the time of David Coulthard and Michael Schumacher.”
Today, parts developed with the help of his company's software serve at the pinnacle of global motorsport. “It's a very great feeling of accomplishment,” Paquet said. “When I started the company, I looked at this as a milestone. Not just Formula 1, but just getting parts that were designed with the software on the road. And yeah, every time that happens, it's a great, great feeling.
Formula 1 is also a very secretive sport. Of the four teams Neural Concept works with, only one was willing to be identified as a client, and even it was tight-lipped about the entire process.
Williams Racing is one of the most iconic teams in Formula 1. Founded in 1977 by racing legend Frank Williams, his team was so dominant in the 1990s that it won five World Constructors' Championships, including three consecutive championships from 1992 to 1994.
But as in most sports, success is cyclical for Formula 1 teams, and at the moment, Williams is in a rebuilding phase. The team finished the 2022 season in last place, rising to seventh place only last year.
NCS is one of the tools helping Williams regain its competitive edge. “We are using this technology in different ways, some of which are improving our simulations, and other ways we are working on that will help achieve better results for the first time in CFD,” said Williams’ head of aerodynamic technology, Harry Roberts.
Again, CFD simulations are time-consuming and expensive, a situation exacerbated by Formula 1 regulations that limit a team's ability to test. Actual time in the wind tunnel is severely restricted, while each team also has a limited budget to account for the time they can use to develop their cars.
Any tool that can help a team put together their aerodynamic designs quickly is a potential advantage, and NCS is already very fast. Paquet estimated that a full CFD simulation that would normally take one hour would take up to 20 seconds via NCS.
Since NCS does not perform actual physics-based calculations but rather makes AI-driven guesses based on its network of aerodynamic learning, it is largely exempt from the strict restrictions imposed by Formula 1. “Anything we can do that allows us to Extracting more knowledge and therefore more performance from each CFD and wind tunnel gives us a competitive advantage.”
But teams still have to pay for it. NCS costs vary depending on the size of the team and type of access, but typically range between €100,000 and €1 million per year, Paquet said. Considering that Formula 1 teams also operate within a $135 million annual cost cap, this is a significant commitment.
Roberts Williams were not willing to point to any specific parts or lap time improvements thanks to the NCS program but said it had affected their car's performance: “This technology is used as part of our toolkit to develop the car aerodynamically. Therefore, we cannot attribute a time The cycle is straight to it, but we know that it helps us with the correlation and speed with which we can investigate new aerodynamic conditions.
Beyond aerodynamics
The continuing march of artificial intelligence will not stop there. There is talk of artificial factors on the pit wall calling for shots at race strategy and even car setups.
“It's an exciting time as the growth in the AI/ML industry is so significant,” Roberts said. “However, it is also a real challenge facing anyone working in technology today. What new tools are we taking the time to explore, develop and adopt?
This isn't the kind of suspense that will captivate the average “Drive to Survive” viewer, but for many Formula 1 fans, race after race is the ultimate source of drama.
As for the Neural Concept, the company is continuing to delve deeper into the non-motorsports side of the auto industry, working to develop more efficient electric motors, improve cabin heating and cooling, and even enter crash testing.
Paquet said the company's software can help engineers improve a car's crashworthiness while eliminating unnecessary weight. But for now, the company can only perform crash simulations on individual components, not entire cars. “This is one of the few applications where we have reached the limits of performance,” he said.
Maybe another application for Thriving supercomputing platforms for AI in the EU?
