Surrogate Modeling Approaches for Ground Vehicle Mobility Simulations

Abstract

Modeling and simulation of ground vehicles can be a computationally expensive problem due to the complexity of high-fidelity vehicle models. Several common mobility tests also require a large set of simulations to be run in order to determine a desired mobility metric. This paper explores the use of High Performance Computing resources to sample a set of parameter space points using high-fidelity simulations, in order to create a surrogate model that can be used to predict the value of a wider domain of parameter space points. In particular, three approaches to surrogate model function creation are explored: K-Nearest-Neighbor (KNN), Inverse Distance Weighting (IDW), and Kriging. These approaches are compared for a particular ground vehicle mobility modeling and simulation task through the average error between the surrogate model predicted values and the actual full-fidelity simulation results.