Distributed Coordination in Multi-Agent Networked Systems: Algorithms and Fundamental Limits

Abstract

Multi-agent networked systems arise in many application domains that are of extreme importance to the Air Force. Regardless of the specific application domain, the central goal is to achieve desired collective behavior of the networked system through the design of admissible local agent control algorithms. Despite the large number of algorithms that have been developed for various applications, we lack the knowledge and methods required to deliberately and systematically shape the network collective behavior due to various challenges such as locality, limited communication, and non-stationary time varying environments. The overarching goal of this project is to formally advance the systematic design of distributed coordination algorithms to ensure high-performance and high-confidence operation of the multiagent networked systems and to investigate fundamental performance limits placed by the various constraints in multiagent coordination. The project entails three complementary thrusts: 1) developing distributed optimization and control algorithms with rigorous performance guarantees; 2) developing communication efficient schemes; 3) developing online control algorithms to handle both system intrinsic internal uncertainties and external uncertainties.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 04, 2023
Accession Number
AD1230378

Entities

People

  • Li Na

Organizations

  • President and Fellows of Harvard College

Tags

Fields of Study

  • Computer science

Readers

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