Driving Closer to the Limit: Improved Virtual Racecar Drivers with Data-Driven Control

Accurate simulation of racing cars is crucial in motorsport to quickly identify effective setups before track testing. Typically, professional drivers provide feedback in simulation, but this process is costly and time-consuming. A capable virtual driver, combined with precise car simulations, could significantly speed up setup development. This paper proposes a data-driven predictive control approach, Data-enabled Predictive, for trajectory tracking in racing simulations. We compare our approach against an industry-standard Proportional-Integral-Derivative controller and a state-of-the-art Model Predictive Control controller, demonstrating that our method is feasible and yields substantial performance improvements, particularly when trajectories approach the car’s physical limits.