Cooperative Output Regulation of Heterogeneous Multi-agent Systems
Organizers: Selahattin Burak Sarsılmaz (Utah State University), Ahmet Taha Koru (University of Texas at Arlington), and Yamin Yan (Nanyang Technological University)
Topics and goals: In cooperative control of multi-agent systems, one of the fundamental problems is to design a distributed control law such that the output of every agent asymptotically tracks a class of references and asymptotically rejects a class of disturbances while preserving the closed-loop stability. The term ‘cooperative output regulation’ was coined in the 2010s to refer to this problem. It offers a unifying framework that considers heterogeneity in multi-agent systems, paves the way for a capability of tracking and rejecting a large class of signals, and contains typical cooperative control problems such as leader-following and formation as subcases. The main difficulty here lies in the lack of central authority. In other words, each agent can share information with only their neighbors. From a control theory viewpoint, how should distributed controllers (i.e., local interactions between the agents and control protocols) be structured to ensure that the cooperative output regulation is undertaken?
The problem has been mainly treated by three approaches, namely distributed observer, distributed internal model, and the integrated approach combining distributed observer and distributed internal model. All three approaches have reached a level of maturity over the last decade. Yet unlocking their full potential requires advancements in the following technical aspects.
- Solvability: Under which system and graph theoretic conditions is the problem solvable?
- Robustness: Robustness of the approaches against system uncertainties (e.g., resulting from imperfect modeling and changes in dynamics).
- Scalability: How is the number of agents related to the computational complexity of the analysis and design tools?
- Performance Guarantee: Designing distributed controllers with transient performance guarantees for the overall closed-loop system (e.g., minimal decay rate and minimal damping ratio).
- Network Imperfection: Communication constraints (e.g., switching networks and delayed information exchange).
- Continuous-time versus Discrete-time
This workshop aims to showcase the recent advancements considering the technical aspects above and identify the relevant challenges toward civilian and defense applications, including search and rescue, monitoring wildfires, and border patrol. The expected outcomes of this workshop are to provide the audience with state-of-the-art methods in distributed observer and distributed internal model approaches by comparing their capabilities and to discuss the leftover challenges to be addressed to reveal their full potential for applications.
Intended Audience: The workshop is expected to be of great value to (i) theoretical researchers working on distributed control and estimation for multi-agent systems, decentralized control of large-scale interconnected systems, distributed optimization, and networked control systems; (ii) applied researchers and practitioners in networked multi-vehicle systems, sensor networks, distributed Nash equilibrium seeking, and electric power grids; (iii) graduate students interested in grasping the existing foundation and learning about the current challenges in the growing field of multi-agent systems.
Cooperative Output Regulation Workshop at ACC 2024 