A Model-Based Parameter Estimation Algorithm for Tire-Soft Soil Contact Model from Off-Road Longitudinal Tests

Virtual simulation of wheeled and tracked vehicles on soft terrain is progressively assuming a central role in the design and validation of vehicles, especially for planetary missions, agriculture, and military framework. Over the years, various approaches have been developed for studying the off-road locomotion with different levels of detail, fundamental assumptions, and computational time. On the other hand, the objective measurement of parameters underlying the interaction between soil and wheel is still a challenging task. The goal of this research work is to develop a model-based estimator capable of identifying tire-soft soil parameter for semi-empirical contact models using vehicle measurements. An experimental campaign is conducted to gather wheel torques and angular velocities while driving straight-forward on a sandy terrain. The experimental campaign for validating the contact model and the parameter estimator is conducted on a flat sandy playground using a B-SUV vehicle in All-Wheel Drive (AWD) locked mode. The experimental test consists of wide-open throttle (WOT) accelerations A 5 Degrees Of Freedom (DOF) virtual vehicle equipped with semi-empirical wheel contact model and fed with the experimental wheel torques is exploited for simulating the real longitudinal maneuvers. An optimization process aiming at minimizing the difference between experimental and numerical wheel angular velocities is defined to tune themain contact parameters.Two distinct experimental data sets are integrated in the optimization loop for pursuing a better estimation. Finally, the algorithm and the contact model are validated through a different experimental dataset.