Adaptive Active-Passive Networked Multiagent Systems

TL;DR

This paper introduces an adaptive control method for active-passive multiagent systems that estimates and mitigates information exchange uncertainties, ensuring robust performance in distributed tasks like human-robot collaboration.

Contribution

It presents a novel adaptive control algorithm that estimates uncertainties and recovers system performance in a distributed manner for active-passive multiagent systems.

Findings

Agents converge to a neighborhood of the average input

The control algorithm effectively suppresses uncertainties

Validated in a human-robot coverage task

Abstract

Active-passive multiagent systems consist of agents subject to inputs (active agents) and agents with no inputs (passive agents), where active and passive agent roles are considered to be interchangeable in order to capture a wide array of applications. A challenge in the control of active-passive multiagent systems is the presence of information exchange uncertainties that can yield to undesirable closed-loop system performance. Motivated by this standpoint, this paper proposes an adaptive control algorithm for this class of multiagent systems to suppress the negative effects of information exchange uncertainties. Specifically, by estimating these uncertainties, the proposed adaptive control architecture has the ability to recover the active-passive multiagent system performance in a distributed manner. As a result, the agents converge to a user-adjustable neighborhood of the average of the applied inputs to the active agents. The efficacy of the proposed control architecture is also validated from a human-robot collaboration perspective, where a human is visiting several task locations, and the multiagent system identifies these locations and moves toward them as a coverage control problem.