Abstract: I will discuss a general mechanism that can lead to complex dynamical systems discovering and "adapting" to patterns in their environment. Specifically, we consider damped systems driven by external forces with varying degree of correlation. Uncorrelated driving environments knock our system around randomly, acting like an additional hot thermal bath. Patterned environments, on the other hand, can make it possible for the system to find configurations that, by matching their dynamics to drive patterns, avoid being randomized and maintain their structure. Besides being merely possible, such "environment-matched" configurations are actually favored in the steady-state by virtue of their stability. I will illustrate our theoretical framework through several toy models, and then show results of its application to a swarm-robotics experiment. In that context, emergence and control of the self-organized dynamics can provide a powerful design paradigm for collaborative swarm behavior.
Those unable to attend can stream the lecture from our YouTube channel.