2023 ARC Research Seminar - Winter Series

PI: Dr. Cindy Bethel; Co-PI: Dr. Daniel Carruth; GRA & Presenter: Viraj Patel (Mississippi State U.)
project link

The goal of this research is to model operator situation awareness, cognitive load, driving performance, and secondary task performance at various levels of partial autonomy with dynamically allocated secondary tasks. Participants operate a partially autonomous military vehicle through an online simulation while responding to secondary task questions that are dynamically presented at different rates based on whether the autonomous system is currently engaged. Participants will be scored on the factors of driving performance, objective secondary task performance based on latency and accuracy of responses to Situation Awareness Global Assessment Technique (SAGAT) questions, subjective situation awareness from the Situation Awareness Rating Technique (SART), and cognitive load based on the NASA Task Load Index (NASA-TLX)and SOS Scale.

PI: Dr. Evgueni Filipov (U. of Michigan, Civil & Environmental Engineering, and Mechanical Engineering)
project link

This talk will discuss our progress on exploring deployable thin sheet origami-inspired structures for applications in multifunctional cloaking and protection of ground vehicle systems. Following a broad overview of possible applications, the talk will focus on three connected efforts including lightweight load-bearing origami at the meter scale, deployable acoustic cloaks, and interpretable design for multifunctional origami. We first discuss fundamental principles of origami-inspired systems, and show how the geometric mechanics can be tailored to create thin sheet reconfigurable structures with unusually high stiffness-to-weigh ratios. A physical two-meter-long prototype is demonstrated for its rapid assembly, versatile modular design, and its ability to support structural loads. Next, we present a compactly stow-able system that when deployed creates a structure of thin sheets spaced at prescribed distances. Multi-physical simulations of this structure show that it has anisotropic density and bulk modulus properties and thus has a potential for use in acoustic cloaking. Finally, the talk presents a machine learning approach for inverse design of origami systems. This approach considers categorical and continuous origami features to design for both shape and function of different origami-inspired engineering systems.

PI: Dr. Daniel Carruth (Mississippi State U.)
project link, project link

In off-road operations, sensor performance will be degraded by environmental effects such as raindrops, water spray, dirt, mud, and snow that will distort or occlude sensor imagery. Current efforts to reconstruct distorted imagery have focused primarily on the effects of rainfall and water spray. Current published datasets are limited in quantity and scope of data and do not include ground truth data for features occluded by the dirt and mud. This talk will discuss our progress on a multi-pronged effort to address the detection, diagnosis, and cleaning of occluded imagery. Lab and field data collections are expanding the scope of available datasets to include occluded imagery from different types of material, varying levels of water content, and application methods. The project is leveraging public and expanded data to train computer vision models to detect and segment occlusions. Multiple methods for assessing occluded sensor capacity are being explored including a simple percentage occluded assessment, 2D and 3D projected effective field of view, and assessment of overlap of occluded regions with functional regions of interest for assessment of loss functional capacity.

PIs: Ezequiel Martin, Casey Harwood, and Hiroyuki Sugiyama (University of Iowa)
project link

Accurate prediction of ground vehicle mobility in shallow water, such as in river crossings, is critically important for reliable operational planning and autonomous navigation in highly complex environments. However, only limited study has been conducted in the vehicle-water interaction due to the complexity in integrating computational fluid dynamics (CFD) and multibody dynamics (MBD) mobility solvers. Furthermore, there is little or no experimental data available describing the effects of shallow water on vehicle maneuverability, such as changes in traction and tire slips due to the hydrodynamic loads. Therefore, in this study, we develop a fast simulation tool for predicting vehicle mobility in shallow water using a data-driven hydrodynamics model and propose model-scale experiments for validation. To this end, a high-fidelity coupled CFD-MBD model, accounting for the vehicle-water interaction, is developed and used to generate training data for the data-driven hydrodynamics model. The predictive ability and computational time reduction achieved by the proposed model are presented. Furthermore, model-scale experiments are conducted for validation, and test results are presented to characterize the hydrodynamic loads exerted on the vehicle in shallow water.

PIs: Dr. Corina Sandu, Dr. Alba Yerro-Colom; GRAs & Presenters: Varsha Swamy and Rashna Pandit (Virginia Tech)
Quad members: Denise M. Rizzo, Katherine M Sebeck (GVSC), Vinita Kumari (John Deere), David Gorsich (GVSC)
project link

The performance of ground vehicles depends on the operating environment. This research is dedicated to the modeling of one of such challenging operating environments, saturated cohesive soils. For the first time, a fully hydromechanically coupled model for the soil undergoing large deformations for mobility purposes is attempted. The prediction of the traction performance of a tire on this saturated soil for short-term and long-term effects is of primary concern for this project. In this seminar, we will discuss the progress in this project regarding the material parameterization techniques and the modeling of short-term effects with a single pass. To open the talk, we will first discuss a few basic terminologies, including effective stresses vs. total stresses, over-consolidated vs. normally consolidated soil, and drained vs. undrained behavior. The realistic behavior of clay is touched upon next as they serve as guidelines when modeling (material model) the clay. To follow different soil testing techniques for saturated clay implemented in our Geotechnical lab for material calibration and validation are discussed. Parameterization of material models based on total stress (Soil and Foam Model) and effective stress (Geological cap) are presented. The talk will be concluded with the preliminary results from pressure-sinkage simulations and tire traction tests for single pass using different numerical frameworks for the soil. The pore water pressure and soil effective stresses evolution due to the tire placement and rolling are presented for the case of FEM soil. Finally, we see how solutions using FEM are stiff, and methods like ALE and SPH prove to be better moving on for these types of high displacement problems.

PI: Dr. Rada Mihalcea (U. of Michigan)
project link

Current autonomous vehicles are able to explore large and unchartered spaces in a short amount of time, however they are not able to “report back” the information they collected in a manner that is easily accessible to human users and does not produce information overload. This limits the situational awareness for both humans and machines, as they interact with complex and dynamic environments; increasing this awareness has the potential to allow for rapid reactions to events that take place in the surrounding environment, and enables more informed and thus better decision making. We are addressing the task of in-the-wild multimodal question answering with visual evidence, in which given a video recording of an in-the-wild territory (as obtained by e.g., a vehicle charting a new territory), we can answer questions about the entities, things, and events it has observed (e.g., “What objects do you see?”, “What is blocking the road?”), while also providing visual support for the answers. We will provide an update on our research work to date, and describe the methods we are developing to help us achieve a better understanding of in-the-wild visual scenes.

PI: Dr. Joyce Chai (U. of Michigan)
project link

In the real world, autonomous driving agents navigate in highly dynamic environments full of unexpected situations where pre-trained models are unreliable. In these situations, what is immediately available to vehicles is often only human operators. Empowering autonomous driving agents with the ability to navigate in a continuous and dynamic environment and to communicate with humans through sensorimotor-grounded dialogue becomes critical. In this talk, we first introduce Dialogue On the ROad To Handle Irregular Events (DOROTHIE), a novel interactive simulation platform that enables the creation of unexpected situations on the fly to support empirical studies on situated communication with autonomous driving agents. We then talk about the Situated Dialogue Navigation (SDN), a navigation benchmark that is developed to evaluate the agent’s ability to predict dialogue moves from humans as well as generate its own dialogue moves and physical navigation actions. We further describe a transformer-based baseline model for these SDN tasks and discuss our empirical results.

PI: Dr. Jeffrey D. Naber (Michigan Technological University)
project link

This project examines the constrained optimal control framework for integrated power, energy, and thermal systems with varying time constants operating in highly uncertain and unstructured environments in application to vehicle autonomy in unstructured off-road environments. Integrated systems often have dynamics varying over different timescales, such systems arise in chemical processes, micro-grids, power generation systems, etc. The project objective is to develop methods that ensure energy sufficiency and reduced vehicle thermal signature in high-risk zones utilizing reduced order, multi-time scale dynamic models. This work builds upon uncertainty information for the terrain and mobility from existing ARC projects.