Learning Model for Organizational Learning in Coexistent Sub-Groups of Swarm Robots

This paper proposes a learning model in which individual robots generate and acquire adaptive behaviors, that is, their own roles in an organized group without outside control. We focus upon emergent processes in which swarm robots dynamically form an organized group in coexistent subgroups. A simulation of a task to construct a truss in space using swarm robots, shows that individuals can adapt themselves to a group with their own local functions, and that an organized structure can emerge from coexistent subgroups to solve the task more eeectively and eeciently than those with a global function.