Robust Multi-Agent Networks

The objective of this research is to develop decentralized methods for achieving robustness in multi-agent networks through self-organization. Multi-agent networks typically consist of numerous components that interact with each other to achieve some collaborative tasks. In many applications, the network may face functional or structural challenges such as failures, noise, or malicious attacks, to name a few. Under such perturbations, a desirable property is to avoid significant performance degradation.

In this research, multi-agent networks are modeled via their interaction graphs, where the nodes represent the agents and the edges denote direct interactions between the corresponding agents. Using this representation, the goal is to design local reconfiguration schemes for decentralized formation of interaction graphs that are robust to structural (e.g., node or edge failures) and functional (e.g., noisy interactions) perturbations. More specifically, we investigate how to obtain a well-connected interaction graph from any connected graph without a significant change in the number of edges.

 

Form_robust_graphs

Both interaction graphs have the same number of nodes and edges. However, the system on the left can be significantly influenced by perturbing a few nodes (e.g., those inside the red rectangle), whereas the system on the right is more robust to such perturbations.

Investigators:

A. Yasin Yazıcıoğlu

Magnus Egerstedt

Jeff S. Shamma

 

Related Publications:

  • A. Y. Yazıcıoğlu, M. Egerstedt, and J. S. Shamma, Decentralized Degree Regularization for Multi-Agent Networks, IEEE Conference on Decision and Control, Florence, Italy, Dec. 2013.

Comments are closed.