Romer Talks

Self-organizing Hierarchical Networks for Robot Swarms

Abstract

From its inception, swarm robotics has relied on the self-organization paradigm, where swarm control emerges from simple stochastic rules that govern interactions among robots and with the environment, without relying on centralized control or global knowledge. While fully decentralized approaches provide robustness, fault tolerance, and scalability, they often struggle to meet the demands of deployment in highly dynamic real-world environments.

This seminar explores self-organizing hierarchical networks for robot swarms—a framework that enables swarms to dynamically balance distributed autonomy with temporary hierarchical coordination. These networks allow robots to aggregate local information, adapt communication structures, and reorganize roles in real time, resulting in greater responsiveness and manageability in complex scenarios. We present mathematical methods for systematically building, merging, splitting, and restructuring hierarchical networks, supported by experiments with mixed ground and aerial robots that demonstrate the scalability and effectiveness of this approach for swarm coordination.

Speaker Bio

Sinan Oğuz earned his B.Sc. in Electrical and Electronics Engineering from Ankara University in 2017. Following graduation, he worked as a design engineer in the defense industry (2017–2019), where he contributed to projects on autonomous UAVs, swarm intelligence, and navigation in GPS-denied environments.

Since October 2019, he has been pursuing a Ph.D. at Université Libre de Bruxelles, Belgium, where he is jointly affiliated with the Artificial Intelligence Laboratory and the Control Design and Systems Analysis Laboratory. His research focuses on swarm robotics, networked control systems, GPS-independent navigation, and unmanned autonomous systems.