Plan-based multi-robot cooperation

Abstract

Several issues arise if one wants to operate a team of autonomous robots to achieve complex missions. The problems range from mission planning taking into account the different robots capabilities to conflict free execution. This paper presents a general architecture for multi-robot cooperation whose interest stems from its ability to provide a framework for cooperative decisional processes at different levels: mission decomposition and high level plan synthesis, task allocation and task achievement. This architecture serves as a framework for two cooperative schemes that we have developed: M+NTA for Negotiation for Task Allocation, and M+CTA for Cooperative Task Achievement. The overall system has been completely implemented and run on various realistic examples. It showed effective ability to endow the robots with adaptive auto-organization at different levels. A number of performance measures have been performed on simulation runs to quantify the relevance of the different cooperative skills that have been proposed.