Toward realization of collective intelligence and emergent robotics

Abstract

Describes a survey of studies on collective intelligence of autonomous robotic agents, especially emergent functionality in a swarm. We have many tasks that we want to leave to autonomous robot systems for example, maintenance of nuclear power pants, rescue in disasters, management of mountainous forest, house keeping, etc. However, there are few robot systems which execute those tasks although factories have already many robots for production. The key is adaptability, robustness and autonomy. Motion planning algorithms still have power but require exact models of environments and robots. We have researched collective intelligence and emergence of cooperation by a swarm of autonomous robots. Although collective robot systems have potential power it can not be handled by traditional motion planning. By the analysis of behavior and development of instruments, we are trying to realize methods to handle collective intelligence and to utilize emergent functionality of robotic systems.