Integrating human swarm interaction in a distributed robotic control system

Abstract

A multi-agent control architecture for swarm robotics applications which includes an innovative human-swarm interface is proposed. The architecture is designed to allow an operator to monitor and guide a robotic swarm to accomplish its missions. The system is composed of three types of agents, a graphical user interface agent, and a pair of a local and a social agent for each robot in the swarm. The local agent implements low level robot-specific functionalities like movement, obstacle avoidance and localization. The control algorithm is implemented in the social agent and is based on an adapted distributed version of the Particle Swarm Optimization technique. An original method, Gravity Points Method, for representing goals which are used by the human-swarm interface is also proposed. Experimental results using simulated e-puck robots are presented and directions for further developments are given.