DDDAS-AI for Modeling, Prediction and Decisions in Interacting Dynamical Systems
Abstract
We propose a DDDAS-AI approach to modeling, prediction, control and decision making in complex interacting dynamical systems (IDS) consisting of interacting vehicles (e.g., drones, planes), humans-users-warfighters and their surrounding natural and-or engineered environment. The challenging goal is to invent foundational DDDAS-AI methods that will be used by IDS researchers and user groups to revolutionize the understanding, modeling, real-time learning, decision making, control and prediction in complex IDS. Creating DDDAS-AI for IDS will combine advantages through the DDDAS-based system cognizant methods with AI-based methods for modeling complex IDS. This requires novel AI methods capable of capturing complex nonlinear system dynamics, continual-learning, and adaptive control. The complexity of current and future IDS such as a battlefield with humans-users-warfighters and vehicles (e.g., drones, tanks) is enormous. The current IDS are constantly evolving and consist of multi-interdependent and interacting dynamic systems each with its own goals and objectives towards a global objective. The recent revolution in sensor technology is allowing us to obtain measurements at unprecedented volume levels in such environments. These data are real-time, dynamic, nonlinear, multimodal, multiscale, nonstationary. The use of these data for the dynamic modeling, analysis, incorporation of domain knowledge, prediction and planning is not humanly possible. For these types of complex IDS we need to develop novel DDDAS-AI methodologies.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 06, 2024
- Source ID
- FA95502310417
Entities
People
- Dimitris Metaxas
Organizations
- Air Force Office of Scientific Research
- Rutgers University
- United States Air Force