A Heterogeneous Secure Testbed for Learning and Adaptation Research of Complex Networked Dynamical Systems
Abstract
(publically releasable)The potential of leveraging networked dynamic systems such as robotics and autonomous systems (RAS) with human-in-the loop using machine learning/artificial intelligence (AI) to achieve the functions such as preserving critical capabilities, assets and activities (CCAA), and any active protection processes that characterize the threat and provide countermeasures to preserve CCAA is critical for Department of Defense operations. Even in situations where there exists an adequate means of accomplishinga task/function via a manned solution, the potential of performing the function remotely via teleoperation, RAS, or AI is preferred. Protection related decision support tools leveraging machine learning/AI from multimodal data using RAS are needed for autonomously emplace bridges, breach explosive, counter UAVs, emplace kinetic and construct non-kinetic obstacles.At Missouri S&T, investigators from Electrical and Computer Engineering, and Computer Science Departmentshave significant experience with reinforcement learning schemes for optimal adaptive control and predictive analytics. The investigators have funded research grants from Office of Naval Research and Army Research Office on topics related to machine learning-based control scheme research with application to autonomous vehicles and their countermeasures. This proposal aims to set-up a test-bed infrastructure comprising of heterogeneous devices for supporting research, education and training in the domain of complex networked dynamic system control and intelligence for securely countering threats. The test-bed will comprise of off-the-shelf sensors, IoT, laptops, high performance computing for decision making tools and autonomous ground and aerial vehicles, localization information, all integrated to form one secure holistic network dynamic system. In particular, the research activities to be implemented and experimented on the test-bed include learning and optimal adaptive control, human-in the-loop studies; predictive analytics using data from multimodal sensors including images; and mitigation of attacks/counter RAS on autonomous systems, all of which are critical issues in CCAA. While our current grants support basic research on these issues, the proposed project, will help us take the next step in building, experimentation and integration of a prototype system, to better validate our research and to train future scientists and engineers for DoD missions. The test-bed and results from the proposed project will be used extensively to train personnel from Fort Leonard Wood Army Center on heterogeneous networked dynamic systems, use cases, and performance assessment.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 06, 2023
- Source ID
- N000142312195
Entities
People
- S. Jagannathan
Organizations
- Office of Naval Research
- United States Navy
- University of Missouri System