Collaborative Autonomy in Uncertain Environments: Exploring Vistas Beyond Consensus

Abstract

Topic Number: PC-02 Technical POC: Matt Bays Networked autonomous, heterogenous agents (robots) (NAHA) performing collaborative tasks are ubiquitous in diverse domains, e.g., marine exploration, disaster management, and reconnaissance. Robust learning, coordination, and distributed control of NAHA demands significant communication (data sharing) and computing resources, inviting threats to data privacy and risks to the security of agents. Moreover, in large-scale swarms, once a cyber-attack or fault is detected, these connected autonomous agents complexity and heterogeneity make it challenging to isolate the compromised ones swiftly and restore operation. While the state-of-the-art event-triggered consensus-based multi-agent control reduces the communication and control overhead, it fails to address the critical challenges of performance guarantee under loss of or limited communications, data privacy, and security (isolation of compromised agents). Therefore, there is a need for breakthrough approaches that goes beyond the consensus-based control algorithms and addresses the above key challenges. The overall objective of this project is to design efficient aperiodic communication protocols, privacypreserving and scalable learning algorithms for inference, filtering, and resource-conscious control strategies for NAHA in a distributed setting. We aim to develop 1) distributed inference for interpretable model learning, 2) resource-efficient distributed estimation and control of NAHA, and 3) a NAHA simulator for validation. We will exploit NAHA s structural and dynamic properties to develop fundamental theories for data-driven inference schemes using perturbation theory and distributed optimization. We will then explore tools from probability theory, interpolation theory, graphical games, and reinforcement learning to handle stochastic uncertainties and limited communication while making filters data-efficient and controllers resource-aware and robust to loss of or limited communication. More importantly, we will focus on involving graduate and undergraduate students who are US Nationals in research. The students will be introduced to research projects working toward technologies that can benefit the national security mission of the Navy, motivating them to join the skilled naval civilian workforce.

Document Details

Document Type
DoD Grant Award
Publication Date
Mar 12, 2025
Source ID
N001742310006

Entities

People

  • Avimanya Sahoo

Organizations

  • United States Navy
  • University of Alabama in Huntsville

Tags

Fields of Study

  • Computer science

Readers

  • Distributed Systems and Data Platform Development

Technology Areas

  • AI & ML
  • AI & ML - Autonomous Systems
  • AI & ML - Machine Learning Algorithms
  • Autonomy
  • Autonomy - Autonomous System Control
  • Cyber