Model Parameter Identification via a Hyperparameter Optimization Scheme for Autonomous Racing Systems

In this letter, we propose a model parameter identification method via hyperparameter optimization scheme (MI-HPO). Our adopts an efficient exploreexploit strategy to identify the parameters of dynamic models in data-driven manner. We utilize our for AV-21, full-scaled autonomous race vehicle. then incorporate optimized design modelbased planning and control systems platform. experiments, MI-HPO exhibits more than 13 times faster convergence traditional methods. Furthermore, parametric learned MIHPO demonstrate good fitness given datasets show generalization ability unseen scenarios. further conduct extensive field tests validate model-based system, demonstrating stable obstacle avoidance high-speed driving up 217km/h at Indianapolis Motor Speedway Las Vegas Speedway. The source code work videos are available https://github.com/hynkis/MI-HPO.