Game-Theoretic Task Allocation and Adversarial Action

Abstract

The project proposes developing formal theory and algorithms for an adversarial game of service agents performing search operationsand an interdictor agent to stop them by increasing the time required for search tasks. Here, the project assumes there exists a set of assets with varying capabilities currently performing search tasks within an area. The assets perform the tasks under a pre-defined but estimated schedule based on initial knowledge of the environment and desired outcome. However, there also exists a resource-constrained interdictor agent that may arbitrarily increase the duration of a subset of search tasks. The goal would then be to develop formal methods to assess the optimal strategy for both the search agents and interdiction agent. The project will develop the adversarial game involving unique theoretical conditions commonly found in logistics operations such as mine countermeasure (MCM) operations with cross-schedule constraints. Cross-schedule constraints are constraints on different agents such as precedence that havea mutual impact their schedule. An example within the MCM domain is that of an unmanned underwater vehicle being transported by a USV which needs to be deployed at a given location before it starts a search operation. The project will solve this adversarial game by combining game-theoretic methods and optimization techniques. The research team believes the game can be reformulated and solved using mixed-integer linear programming (MILP) techniques. The result will be improved search performance, robust field operations, and/or reduction in completion time for operations by recognizing and exploiting potential adversarial actions in the game. The developed methods could have significant impact on Navy search tasks such as those found in mine countermeasure (MCM) operations but are extensible to other logistics operations where there exists cross-schedule constraints and potential adversarial actions.

Document Details

Document Type
DoD Grant Award
Publication Date
May 05, 2021
Source ID
N000142112432

Entities

People

  • Jorge A. Sefair

Organizations

  • Arizona State University
  • Office of Naval Research
  • United States Navy

Tags

Fields of Study

  • Computer science

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Game Theory.
  • Operations Research

Technology Areas

  • Autonomy
  • Autonomy - Autonomous System Control