End-to-End Validation and Verification of Autonomy-Enabled Ground Vehicles-DRAFT

April 23, 2021

Laura Redmond1, Atul Kelkar1, Cameron Turner1, John Wagner1, Yue Wang1 and Matt Castanier2

1Clemson University, 2U.S. Army DEVCOM GVSC

To facilitate rapid virtual prototyping of autonomy-enabled ground vehicles, validation and verification (V&V) processes must be created that ensure the accuracy and suitability of rapidly deployable computer models through digital and physical datasets. In addition, the availability of digital twin tools will empower engineers in the analysis of product design and requirement changes on mission objectives and system performance throughout the lifecycle.

CE Assistant Professor Laura Redmond is heading up a team of faculty and students from Clemson to contribute new V&V methods and tools that can be applied in the planning, prototyping, and active-duty phases of autonomy-enabled ground vehicles as part of Clemson’s new Virtual Prototyping of Autonomy-Enabled Ground Systems (VIPR-GS) center. Ultimately, the proposed methods and tools will have the benefit of improving vehicle safety and performance, as well as increasing the amount of quantitative data available for test planning and risk management.

The technical approach can be divided into several research tasks that span across the life cycle of an autonomy-enabled ground vehicle as shown in Figure 1. Starting in the planning phase, the team has outlined an approach for verification of autonomy software utilizing formal verification methods, of which there is limited research to date. The next task spanning the planning and prototyping phases is surrogate-enabled V&V which aims to leveraging surrogate modeling techniques that will ultimately help inform resource-efficient test planning tools. The third task within the prototyping phase is accessible model validation methods for lean datasets. The proposed validation tool leverages Bayesian methods to provide quantitative data regarding model validation and better inform the decisions made based on vehicle model outputs. The final task within the active-duty phase is creation and application of a digital twin to enable exploration of design and operating issues throughout the product lifecycle.

Figure 1. Division of project scope across the life cycle of an autonomy-enabled ground.  (Click on image to enlarge)




About Dr. Redmond

Laura Redmond is an Assistant Professor in the Glenn Department of Civil Engineering. Her areas of expertise span advanced simulations for civil, mechanical, and aerospace applications, model calibration, model verification, and model validation by test. Examples of current research applications include test-validated finite element (FE) simulations for drive-by-health monitoring of bridges, modeling and design of rigid-flex PCB robotics for extreme impact resistance, Bayesian calibration and validation techniques and design of metamaterials. Research sponsors include NASA, NSF, Army GVSC, and industry partners.

Read more about the research project