Control, Estimation, and Learning in Multi-Team Dynamic Network Games with Asymmetric Information

Abstract

U.S. Air Force systems and operations have seen dramatically increased complexity, requiring coordination of large, complex networks of decision-making agents, including autonomous and semi-autonomous systems. Many decisions in battle engagement are made implicitly, or explicitly, in the context of a dynamic game, where (at least) two sides compete in a dynamic and uncertain environment to achieve their objectives while denying the other’s. The vast majority of research in dynamic games assumes that all players have full information about environment models and network states. However, in reality players have only limited and disparate information and must use their noisy data to estimate or learn relevant quantities for models and network states to make optimal decisions. This information asymmetry introduces challenging belief representation and theory of mind issues, where agents must impute belief states and estimates of other agents to inform their own strategy. This incentivizes cooperation and signaling amongst teammates, conflict, stealth, and deception in the presence of adversaries, and assessing trust in information. In this project, we will build a novel framework for modeling and solving multi-team dynamic games in complex networks with asymmetric information, where players’ uncertainties about the network and environment are integral parts of the approach. We will create next-generation architectures and algorithms for approximating and learning equilibrium strategies by integrating tools from dynamic game theory, feedback control, robust-stochastic optimization, data science, and machine learning. Our project will deliver robust and perception-aware team autonomy stacks that can provide radically enhanced capabilities for coordinating future Air Force systems. Our approach will be based on rigorous theory while also addressing concrete and realistic application scenarios, such as multi-agent pursuit-evasion, capture-the-flag, signaling games, etc.

Document Details

Document Type
DoD Grant Award
Publication Date
Mar 06, 2024
Source ID
FA95502310424

Entities

People

  • Tyler Summers

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force
  • University of Texas at Dallas

Tags

Fields of Study

  • Computer science

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Systems Analysis and Design
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.

Technology Areas

  • AI & ML
  • AI & ML - DoD AI Strategy
  • Autonomy
  • Autonomy - Autonomous System Control
  • Autonomy - Human-Robot Interaction
  • Autonomy - UAVs