Exploiting physical constraints: heap formation through behavioral error in a group of robots

In this paper we describe a collective heap building process by a group of robots. Instead of predefining "cognitive" capacities, we exploit the physical structure of the robots and the self-organizing properties of group processes. The robots' control program effectively contains just one behavioral rule to avoid detected obstacles. Due to the constrained sensory input, the robots collide with objects that are exactly in front of them. In this way, objects are pushed and clusters are formed. To study the dynamics of the cluster process we conducted experiments in which the number of objects and robots were varied. We found that a limited amount of mutual interference is crucial for the fusion of clusters; hence large, single heaps did never emerge in trials with just one robot. However, with more than 4 robots the heap building process slows down due to increased mutual avoidance movements.