Design of a Robotic Driver With Adaptive Controllers for Vehicle Speed Tracking Tests

To realize accurately vehicle speed tracking in vehicle tests, robotic drivers with a universal and simple structure are designed, and adaptive PI controllers based on full form dynamic linearization (FFDL) techniques are proposed. The robot manipulates vehicle speed via electric cylinders acting on accelerator and brake pedals. The open-loop plant, from displacements of electric cylinders to vehicle speed, is approximated by a 1st-order nonlinear uncertain system, which is further transformed into an FFDL model. To ensure desirable tracking accuracy and compensate for uncertainties, novel cost functions are considered to establish sampled-data adaptive PI control algorithms. Then, theoretical conditions are provided to ensure closed-loop stability and convergence of set-point tracking errors. Finally, experiments on real electrical cars, which are running on chassis dynamometers, are conducted under Chinese National Standard GB/T18386.1-2021 requirements. To improve the reliability and reduce the period of adaption, several compensation technologies are utilized in the experiments, including rough offline parameter learning, preview target speed, and $\sigma$-modification methods. The experiment results demonstrate the effectiveness and efficiency of our control algorithms and robotic driver hardware.