Case Study Abstracts

Case Study 1

Stranded: Avoiding the perils of unknown environments through the synthesis of energy, soil, and adversary localization maps

Contributors:  
Faculty: Kira Barton, Dimitra Panagou (University of Michigan)
Ardalan Vahidi (Clemson University)
Lauro Ojeda (University of Michigan)
Student: Michael Quann (University of Michigan)
Government: Denise Rizzo, William Smith (U.S. Army TARDEC)
Industry: Frank Koss, Andrew Dallas (SoarTech)

        This case study brings together three ARC projects to investigate how enhanced situational awareness will improve 3D energy mapping in off-road environments. An important aspect in off-road energy mapping is the synthesis of a priori topography information combined with real-time dynamic updates. To address this need, this case study combines an energy-mapping framework with (1) the utilization of static soil and elevation maps to provide a priori information that will enable prioritization of initial areas of interest within the environment based on uncertainties associated with the energy costs of a given location; and (2) dynamic updates from aerial maps that will provide near real-time information about the terrain (wet, dry) and identify “go / no-go” locations in the environment based on adversarial elements. Thus, the soil, terrain, and aerial maps are integrated into the energy-mapping framework for enhanced decision-making. The addition of these maps is expected to improve the reachability analysis and energy mapping efficiency of multi-robot reconnaissance in an unknown environment through the identification of go / no-go locations and targeted areas of interest. A simulation study is used to evaluate the effectiveness of the combined mapping approach.

Case Study 2

I Want It All: Achieving High Fidelity and Optimal Computational Complexity in Physics-Based Off-Road Mobility Simulations

Contributors:  
Faculty: Hiroyuki Sugiyama (University of Iowa)
Shravan Veerapaneni (University of Michigan)
Hiroki Yamashita (University of Iowa)
Post-doc.: Eduardo Corona (University of Michigan)
Student: Guanchu Chen (University of Iowa)
Government: Paramsothy Jayakumar, Yeefeng Ruan (U.S. Army TARDEC)
Kenneth Leiter (U.S. Army Research Laboratory)
Industry: Mustafa Alsaleh (Caterpillar Inc.)

        A high-fidelity computational vehicle-terrain interaction model is essential for physics-based off-road mobility simulations in achieving accurate mobility performance prediction as well as reliable operational planning. This case study brings together two projects to address accuracy and computational efficiency of physics-based vehicle-terrain interaction simulation capabilities, which can be fully integrated into general multibody dynamics (MBD) simulation algorithms. For this purpose, a hierarchical multiscale terrain dynamics model is developed to eliminate phenomenological assumptions in existing constitutive models and is further extended to tire-soil interaction simulation. The finite-element (FE) model is utilized to predict macroscale soil deformation, while the microscale constitutive behavior is modeled by the representative volume element (RVE) using the discrete-element (DE) approach to describe complex soil failure phenomenon including strain localization. Validation and comparison with single-scale FE and DE approaches are presented. Furthermore, to improve computational efficiency of different components of the coupled MBD-FE-DE vehicle-terrain interaction simulation capability, fast solvers based on hierarchical low-rank factorizations are applied and the ensuing speedups are demonstrated on several test cases.