Robotic swarms can pull their weight to cluster or flock

Robotic swarms, ensembles of collaborative robots that work together to achieve tasks, are an appealing solution to tackle complex tasks such as automated exploration, foraging, or transport.

However, to scale well with the number of robots, a swarm cannot rely on an external controller or on complex computation, and requires simple design rules to achieve emergent functions.

In this talk, viewing robots as self-propelled particles, I will show that fine control of collective behaviour can be achieved through simple mechanical ingredients -- using results from analytical, numerical, and experimental work.First, I will show that a large family of robots can be described by Active Brownian-like dynamics with a self-alignment term, whose prefactor is an intrinsic curvature controlled by the size and mass repartition of the robot.

Then, I will show that this signed curvature (“curvity”) is at the origin of an effective attracto-repulsion at the level of pairs of robots.

As a result, curvity seeds the collective behavior of the swarm at the many-body level, offering a direct design rule to control whether the swarm flocks, flows, or clusters -- and how big clusters get.

I will thus demonstrate a computation-free route for decentralized control on collective behavior, paving the way for richer swarm-robotic applications.