Our lab studies design and coordination of robot collectives, and the behavior of their biological counterparts. In many scenarios, a collective of simple, locally interacting robots can outperform a single, more complicated robot. Practical applications may, e.g. include search and rescue, chemical spill containment, and collective construction. In other words scenarios where large numbers of robots can work efficiently in parallel, and where failures due to complex real-world interaction will not cause complete system failure.

We take inspiration from social insects in nature, that are able to exhibit remarkably complex behavior, such as foraging, brood care, defense, and nest maintenance, despite the lack of a central controller. The key to their approach lies in swarm intelligence: complex behavioral patterns emerging from local interactions between many simple and noisy individuals, and embodied intelligence: 1) omitting the need for detailed state models, they instead leverage their shared environment directly to share and pass information between individuals, and 2) they optimize the world around them to suit their own (limited) sensory organs.