Flexible and Resilient Autonomous Systems
Abstract
The objective of this proposed effort is to carry out fundamental scientific research in the area of multi-agent planning and network aware coordination in a dynamic, partially observable adversarial environment with the goal to provide theoretical foundations for building Flexible and Resilient Autonomous Systems (FRAS). The proposed research project relies on the available state-of-the-art, but mainly leverages previous research results of the two co-PIs in the respective field. The proposed project will apply (i) the latest technology multi-agent planning and distributed coordination for autonomous unmanned aerial assets together with (ii) the latest research in game-theoretic models of automated reasoning in adversarial environment and (iii) capability to use machine learning for generalizing adversary behavior patterns and learning heuristics that drive directions in game-tree search and thus allow for scalability of the methods developed as in (i) and (ii). The research results will provide US Air Force systems with the ability to (a) successfully identify, (b) adapt to, and (c) recover from adverse events such as a loss of communications, asset loss, changing enemy tactics, and/or changing mission objectives. Additionally, the research will develop learning algorithms so as to learn, actively adapt strategies and tactics based on the observed enemy response from past engagements, so as to ensure resiliency and mission success.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 09, 2018
- Source ID
- FA95501810097
Entities
People
- Katia Sycara
Organizations
- Air Force Office of Scientific Research
- Massachusetts Institute of Technology
- United States Air Force