Autonomous detection and mapping of meta-events using distributed control agents

Swarm robotics research describes the study of how a group of relatively simple physically embodied agents can, through their interaction collectively accomplish tasks which are far beyond the capabilities of a single agent. From this information they are able to decide their behavior and take the appropriate action. A global behavior can then be witnessed that is derived from the local behaviors of each agent. The presented research introduces the novel method for optimizing the communication and the processing of communicated data for the purpose of detecting large scale meta-object or event, denoted as meta-event, which are unquantifiable through a single robotic agent. The ability of a swarm of robotic agents to cover a relatively large physical environment and their ability to detect changes or anomalies within the environment is especially advantageous for the detection of objects and the recognition of events such as oil spills, hurricanes, and large scale security monitoring. In contrast a single robot, even with much greater capabilities, could not explore or cover multiple areas of the same environment simultaneously. Many previous swarm behaviors have been developed focusing on the rules governing the local agent to agent behaviors of separation, alignment, and cohesion. By effectively optimizing these simple behaviors in coordination, through cooperative and competitive actions based on a chosen local behavior, it is possible to achieve an optimized global emergent behavior of locating a meta-object or event. From the local to global relationship an optimized control algorithm was developed following the basic rules of swarm behavior for the purpose of meta-event detection and recognition. Results of this optimized control algorithm are presented and compared with other work in the field of swarm robotics.