A Cooperative Agent Based Architecture for Environmental Exploration and Knowledge Sharing by Vision

A comprehensive approach to the design and implementation of multi-robots cooperative systems is described. It focuses on a design process that uses the Unified Modeling Language and on a detailed ontology description with the goal of sharing the knowledge on environments that robots can acquire through the use of their vision sub-system. We base the implementation of our robotics vision system on agents inserted in a generic multi-level architectures. The first objective of this work is to provide a framework to perform a rigorous agent-based design process for scenarios where many robots are involved in different operations. Then we introduce a system for describing, upgrading and sharing knowledge about operating environments of a cooperative robot fleet. As a consequence we design a multi-level, agent-based vision architecture that takes advantage of the distribution over several different robots. Details of the methodology, of the ontological approach, of system implementation using FIPA-OS environment, along with real experiments are reported.