Scalability of classical terramechanics models for lightweight vehicle applications incorporating stochastic modeling and uncertainty propagation

Paramsothy Jayakumar, Daniel Melanz, Jamie MacLennan, David Gorsich, Carmine Senatore, Karl Iagnemma
Journal of Terramechanics, Volume 54, August 2014, Pages 37-57, ISSN 0022-4898, http://dx.doi.org/10.1016/j.jterra.2014.04.004. http://www.sciencedirect.com/science/article/pii/S002248981400024X
Abstract: This paper investigates the validity of commonly used terramechanics models for lightweight vehicle applications while accounting for experimental variability. This is accomplished by cascading uncertainty up to the point of wheel performance. Vehicle-terrain interaction is extremely complex, and thus models and simulation methods for vehicle mobility prediction are largely based on empirical data. Analytical methods are compared to experimental measurements of key operational parameters such as drawbar force, torque, and sinkage. Models of operational parameters ultimately depend on a small set of empirically determined soil parameters, each with an inherent uncertainty due to test variability. The soil parameters associated with normal loads are determined by fitting the dimensionless form of Bekker’s equation to the pressure-sinkage test data. Similarly, the soil parameters associated with shear loads are determined by fitting Janosi-Hanamoto’s equation to the direct shear test data. An uncertainty model is used to propagate the soil parameter variability through to the wheel performance based on Wong and Reece. This commonly used analytical model is shown to be inaccurate as the envelope of model uncertainty does not lie within the experimental measures, suggesting that model improvements are required to accurately predict the performance of lightweight vehicles on deformable terrain.
Keywords: Terramechanics; Robotic vehicles; Bekker and Wong models; Soil test bed; Uncertainty propagation; Stochastic modeling