Title: Self-organizing Robot Swarms
Abstract: Robot swarms that operate in a fully self-organized manner, without any central coordinating unit, have been widely demonstrated. These systems rely on decentralized architectures where collective behavior emerges from local interactions. This design provides key advantages such as scalability, fault tolerance, and the absence of single points of failure. However, it also introduces challenges in terms of system-level control and manageability.
In contrast, centralized systems are easier to design and control but lack scalability and are more vulnerable to failures at critical nodes.
In this talk, I will present a novel swarm architecture that enables self-organized hierarchy, combining the resilience of decentralized systems with the controllability of centralized ones. Using a heterogeneous team of ground and aerial robots, I will demonstrate how the swarm can self-organize into a dynamic hierarchical control structure through local asymmetric communication. I will present the results of experiments that illustrate capabilities such as autonomous sub-swarm splitting and merging, dynamic replacement of failed robots, and real-time adaptation of collective behavior, while preserving the key benefits of self-organization, including scalability and interchangeability of individual robots.