Artificial intelligence is the great technological frontier of our time, already finding applications in Formula 1 in various areas. However, contrary to what one might think, it is not an entirely new concept, with a primordial project dating back to the last millennium. The protagonist of this story is Pat Symonds, former CTO of Formula 1 and now technical consultant for Andretti Cadillac, who spearheaded a project that was decades ahead of its time.
The Benetton experiment
The story emerged during an episode of The Neil Ashton Podcast, in which Pat Symonds was a guest last June: “Towards the end of the 1990s, when I was technical director at Benetton, I initiated a program to see how we could use neural networks to improve vehicle dynamics. We had built our own data collection system, and I had designed it to act as a database, a sample that could be queried. We started working with the University of Sheffield on a neural network system that could ask what the data said and what the setup was, then correlate everything with the drivers.”
The project’s goal was to train a model with data collected directly on the track. Listening to Pat Symonds’ words, it seems that Benetton wanted to use a neural network system to trace the impact of setup parameters on the car’s behavior, creating a predictive model that would indicate how to improve vehicle dynamics. It was a very bold project, but it didn’t take into account the technological limitations.
The failure
Pat Symonds’ account shows that the theoretical foundations of the early forms of artificial intelligence existed over twenty years ago, but only with the subsequent evolution of processors and computer technology did they begin to find practical application: “For those times, we were years ahead, to the point that the project eventually failed because we didn’t have the computing power we needed. You couldn’t yet purchase a Machine Learning interface to connect to the program, so we had to write the entire protocol ourselves.”
Despite the failure of the program, the 1990s remain a decade of technological revolution for Formula 1, with the spread of the first computerized aerodynamic simulation programs. Twenty years later, the same technology is now being complemented by artificial intelligence, which is set to become the third pillar for aerodynamic analysis alongside CFD and wind tunnel testing. A science that has only recently taken its first concrete steps, but Benetton had already glimpsed its potential in the last millennium.
Leave a Reply