Development of race car analysis tools and methods for performance optimisation: artificial intelligence based image detection for strategy optimisation

Abstract

W Racing Team is one of the most prominent motorsport teams in the world and is trusted by BMW to operate the BMW M Hybrid V8 in the World Endurance Championship. This championship is currently experiencing a golden era, with eighteen cars competing at the highest level. Winning races has therefore become increasingly difficult, and every detail must be perfect to have a chance of success. One of the most crucial aspects of achieving victory is the ability to design and execute the best possible strategy. This master’s thesis focuses on the development of artificial intelligence tools to enhance strategic decision making. These tools leverage modern technologies such as neural networks and were implemented in Python using simple, modular building blocks to ensure accessibility for future interns who may not have a background in computer science. A key part of the project involved creating a link between the Python environment and WRT’s existing tools, such as HH Timing. The final solution includes both the main code and an executable application designed for ease of use at the racetrack by end users who may not be familiar with coding. The software was tested during real races: one component was operated trackside via the executable by a team member, while additional data analysis was conducted simultaneously at WRT’s headquarters in Bierset. The overall results were highly satisfactory and met expectations. Indeed, thanks to this tools the strategist have half less data to analyse. As a result, the software is now used by the team at every race and serves as a foundation for future projects.