Collective Sensing with Self-Organizing Robots

We introduce a way to compose two local rule sets to form an environmental sensor. Each set of local rules results in a different global behavior when interpreted by a set of interacting programmable particles. In the composed system, the particles choose which set of rules to use depending on whether or not a certain condition is true or false about the initial state of the system. The global behavior of the system eventually matches only one set of rules, signaling that the particles have collectively recognized the condition on the initial state. We demonstrate the composition method on our robotic testbed