Multi-agent Flocking Control
Coupled oscillator model for various multi-agent flocking patterns.
Flocking is commonly observed in many multi-agent systems (MAS) in nature. The video on the left showcases a spectacular example of flocking in the flight of starlings. The complex system-level behavior cannot be achieved with a central planner. Instead, a distributed decision process must be adopted in such systems. Inspired by this observation, I started a project exploring the flocking behavior of an artificial collective.
Inspired by the Kuramoto Model and the work by Paley et al., I extended the coupled oscillator model to 3D space, and derived a distributed control law that enabled agents to coordinate with each other through local decision-making such that various global flocking pattern emerges (as shown by videos below).
The proposed controller also allowed the flocking collective to track reference trajectories, which is more suitable for engineering applications.