An Autonomous Agent System to Simulate a Set of Robots Exploring a Labyrinth

Abstract

This work presents anapplication of the agent ech-nology to the simtdation i the robotic domain. A group of robots cooperate with each other to simulate a large surface which as a labyrinth structure. A robot is represented as a specialist agent which as for task to recognize di~erent paths characterizing thisurface. The application s supported bya hybrid and hierar-chical architecture model called MAGIQUEI. The in-teractions between robots are managed by supervisor agents which ave a global vision of the labyrinth. T e explored parts of the labyrinth are hold at a black-board structure managed by the supervisor. Agents can communicate ither directly by message-passing or indirectly via their supervisor. The paper begins by describing the application and presents different con-straints related to the move of the robots. It discusses how the application is implemented in a distributed programming environment and presents also the prac-tical constraints faced in the instantiation of the model. The application addresses also an issue related to the load balancing between robots. Finally, it concludes by investigating a set of issues to explore. Topic Area: Multiagent systems