A Theory for Allocating Autonomous Systems
Abstract
Technological advances are paving the way for the design and realization of autonomous systems capable of undertaking a diverse set of tasks. Examples of such tasks range from informational gathering and surveillance, management of supply lines, establishment of communication and sensing networks, among many others. At the heart of any autonomous system is a collection of decision-making entities (or agents) that are preprogrammed to make independent decisions in response to locally available information. The success of a given control design hinges of whether or not these preprogrammed decision-making rules facilitate the emergence of a collective behavior that eciently accomplishes the task at hand. Developing competency in autonomous system designs for a diverse set of tasks could in theory provide a library of capabilities that a central commander could choose to employ as part of an overall mission plan.The goal of this proposal is to develop a foundational theory for how to successfully integrate autonomous systems into an overarching mission plan. Developing this theory requires a formal understanding of both (i) how to allocate a limited amount of autonomous assets to a series of dierent tasks and (ii) how to locally coordinate the behavior of the autonomous assets assigned to each task. The rst set of research thrusts in this proposal center on the framework of cooperative system design in non-adversarial environments. In particular, the proposed research questions will seek to characterize the decomposition strategies and admissible coordination algorithms that optimize the quality of the emergent collective behavior. The second part of this proposal focuses on cooperative system design in adversarial environments. Here, informational awareness about an adversarys intentions and capabilities can potentially be exploited to derive admissible control algorithms with improved performance guarantees. The research questions posed in this section will seek to characterize the interplay between knowledge, exploitation, and performance guarantees in adversarial environments.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 17, 2020
- Source ID
- N000142012359
Entities
People
- Jason R. Marden
Organizations
- Office of Naval Research
- United States Navy
- University of California, Santa Barbara