Firefly-Inspired Synchronization in Swarms of Mobile Agents

Synchronization can be a necessary prerequisite to perform coordinated actions or reach consensus in decentralized multi-agent systems, such as robotic swarms and sensor networks. One of the simplest distributed synchronization algorithms is firefly synchronization, also known as pulse-coupled oscillator synchronization. In this framework, each agent possesses an internal oscillator and the completion of oscillation cycles is signaled by means of short pulses, which can be detected by other neighboring agents. This thesis focuses on a realistic mode of interaction for practical implementations, in which agents have a restricted field of view used to detect pulses emitted by other agents. The effect of agent speed on the time required to achieve synchronization is studied. Simulations reveal that synchronization can be fostered or inhibited by tuning the agent (robot) speed, leading to distinct dynamical
regimes. These findings are further validated in physical robotic experiments. In addition, an analysis is presented on the effect that the involved system parameters have on the time it takes for the ensemble to synchronize. To assess the effect of noise, the propagation of perturbations over the system is analyzed. The reported findings reveal the conditions for the control of clock or activity synchronization in swarms of mobile agents.