ABSTRACT
The aim of the work is to measure, understand and model the steering behaviour of a human driver at the limit of adhesion. A robust nonlinear model predictive control strategy is implemented. The behaviour of the controller is investigated for nonlinear vehicles travelling at constant speed on circular paths at or near the limit of adhesion. The controller successfully maintains the vehicle as close to the path as the tyre friction allows. A parameter study shows that weighting values in the cost function, the driver’s refractory period, and parameters of the driver’s neuromuscular system have significant influence on the control action. Comparison to data from a driving simulator experiment indicates that the simulated control action bears some similarity to that of the human driver, although there are some discrepancies that require further investigation.
