Data-Driven Vehicle Performance Optimization for Formula Student Racing

In the world of motorsports, every driver aims for one thing: to finish with the fastest time. But it's not as simple as hitting the throttle pedal. There are vast numbers of factors involved, from how drivers handle their cars individually to the unpredictable nature of the track and vehicle. Formula Student Racing, where teams are tasked each year to build and race their own cars, is a university level single-seater competition. It's a hands-on learning experience like no other, where students apply classroom knowledge to real-world challenges, pushing the boundaries of automotive technology. Building a fast car is only part of the equation for Squadra Corse PoliTo, the student racing team of Politecnico di Torino. The real challenge lies in integrating and optimizing driver and car performance, which is the main focus of this thesis. Data Science and Machine Learning is implemented to understand how a driver's actions affect the performance of the car and how to perfect it. The goal is to develop a cutting-edge tools for Squadra Corse to offer insights about the hidden relationships between what the drivers put into the car, and how the car responds. By harnessing the power of data and algorithms, the aim is to supply drivers with the competitive edge and deeper understanding they need to outpace themselves and the competition. First, the most critical inputs of driver behaviour on vehicle performance will be identified. Then, with this knowledge, a predictive model that can anticipate how different strategies will impact performance will be built. Final step is to build a framework using both Optimization methods and Imitation Learning techniques to reach the best driver input strategy configuration and create a driver model. With this proposed framework, Squadra Corse can formulate the best strategy for specific configurations and give feedback to their drivers to maximize performance.