Grid-Based Data-Driven Transient Tire–Soil Contact Model Bridging Computational and Classical Terramechanics Models

Abstract A reliable simulation tool capable of predicting off-road mobility on complex granular deformable terrain is essential for vehicle design and performance evaluation. However, the use of high-fidelity computational models leads to high computational costs, while computationally cheaper classical terramechanics models have a limited capability in modeling transient tire–soil interaction behavior due to the quasi-static modeling assumptions. Therefore, this study proposes a new grid-based transient tire–soil contact model by bridging the high-fidelity computational tire–soil interaction model and the classical terramechanics model through machine learning techniques. To this end, contact data obtained from the high-fidelity computational tire–soil interaction model are mapped onto the contact grid defined on the tire surface by generalizing the motion description in the classical terramechanics model. The grid contact data are then employed to develop a data-driven transient contact model. The proposed grid-based contact model leads to a fast online collision detection process using multiple circular contact lines while capturing transient contact stress responses at active grid points on deformable terrain. Furthermore, the effect of dynamic soil material flow underneath the rolling tire is described by the soil surface velocities within the grid contact patch, and they are learned by neural networks to predict the slip-dependent contact stress characteristics. It is demonstrated by numerical examples that the transient tire–soil interaction behavior on large deformable granular terrain can be predicted accurately in scenarios not considered in the training data while achieving a substantial computational speedup.