A Formal Framework for Developing Resilient Teams of Heterogeneous Autonomous Agents
Abstract
Future Naval missions will require heterogeneous teams of autonomous agents capable of understanding large volumes of data in support of long-duration deployments. Throughout these missions, teams will encounter unexpected changes to their environment, and will need to derive intelligence in real time from massive, distributed, and diverse data sources. Teams must be ableto exploit this data, without being dependent on the data or being vulnerable to errors or inconsistencies in the data, in order to adapt to the present situation and persist in accomplishing their missions. The proposed research addresses the need to define and measure the resilience of Naval systems in coping with noisy and incomplete data sources, integrating information from multiplesensing modalities, and coherently propagating and outputting measures of uncertainty, by adapting resilience definitions and approaches from the ecological domain. The methods of resilient ecological systems have been extensively documented, although the approaches to date have oftenbeen derived from individual case studies. Additionally, evaluations of resilience in ecological systems is often performed only after all relevant data has been collected. New methods are needed that permit the system to evaluate its performance in real time and adapt to current and anticipated conditions.The proposed research will develop and formalize an underlying theory of resilience, define and validate associated resilience metrics, and derive resilient system design principles that will enable comparison of the performance of teams in various configurations and under different classes of perturbations from the environment or the data sources. The proposed approach will formalize the definition of resilience that accounts for augmenting agents??? computational potential, physical embodiment, information networks and the system???s capacity to provide services that will accomplish the overall mission, even when simple or complex disturbances are encountered. Metrics will be developed to assess resilience to perturbations (a) in the environment, (b) in robot embodiments,and (c) in communication and information networks. These metrics will be validated through a series of experiments using progressively more complex team configurations and classes of perturbations. The standard for success will be the development of resilient system design principles and the creation of a tool-box to support designing new systems that are resilient-by-design.The proposed research will develop a resilience framework that will support multiple distributed Naval assets, collectively functioning as a multi-agent, swarm, and colony-based system, to reliably achieve collective missions, such as mine counter measures, track and trail, and intelligence, surveillance and reconnaissance (ISR), which specifically addresses Focus Area 4 of Topic3 of the broad area announcement. The ability for these systems to leverage complex information sources in order to improve the overall system???s resilience will increase the U.S. Navy???s ability to maintain dominance, while supporting new missions that may include multiple objectives from multiple distributed organizations, continuous redeployment, and large-scale mission coverage andengagement.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 04, 2018
- Source ID
- N000141812831
Entities
People
- Julie Adams
Organizations
- Office of Naval Research
- Oregon State University
- United States Navy