Opinion Dynamics for Robotic Swarms: Resilient Coordination of Heterogeneous Teams

Multi-robot systems are increasingly being deployed in applications such as surveillance, environmental monitoring, and industrial automation. However, the state of the art primarily focuses on homogeneous systems performing single, simple tasks (e.g., coverage or formation tracking) [1].

The next frontier in robotics lies in Heterogeneous Multi-Robot Systems—teams of agents with diverse sensing and actuation capabilities (e.g., drones, rovers, and manipulators). These systems are required for complex, multi-objective missions, such as precision agriculture (combining foraging, transport, and inspection) or disaster response. The scientific challenge is to seamlessly coordinate diverse agents to perform different tasks simultaneously in expansive, dynamic environments in an autonomous and distributed way.

To address this problem, we look to nature and society. Biological and social networks exhibit robust coordination through decentralized mechanisms like stubbornness (persistence in a goal), clustering (sub-group formation), and inhibitory interactions (competition).

This PhD aims to enhance the coordination of heterogeneous systems by mathematically formalizing these social dynamics into control algorithms. The core innovation lies in adapting models of Opinion Dynamics [2]—traditionally used to analyze static social networks—and transforming them into active control protocols for mobile, physically constrained robots. The research strategy centers on a multi-layered control architecture designed to translate abstract social dynamics into concrete, executable robotic behaviors. At the core of this approach is the development of a distributed "Opinion Layer," where agent interactions are modeled using Nonlinear Agreement Protocols and Signed Graphs [3-5]. In this framework, a robot's "opinion" acts as a dynamic state representing its preference for specific tasks or locations, weighted by its unique capabilities within the heterogeneous team.

Experimental Validation : The theoretical contributions will be validated using the Rainbow Team's experimental facilities, following a roadmap that progresses from fundamental coordination tasks, such as autonomous clustering, to complex long-term missions requiring dynamic team reconfiguration. These scenarios will be executed on a heterogeneous fleet of aerial and ground robots (e.g., Crazyflies, RoboMasters) within fully instrumented multi-robot arenas to demonstrate real-world resilience.