Fuzzy coordination graphs and their application in multi-robot coordination under uncertainty

In distributed systems, simultaneous access to shared resources is the main cause of conflicts among rational agents, and therefore coordinating the agents´ actions is an essential task for multi-agent systems towards achieving their goals. The coordination problem gets more challenging when multi robot systems interact in real world and have uncertainty and incompleteness in their perceptions of the world and other agents. In order to construct a more exact model of world uncertainty, enhance the coordination task of the robots, and attain a more rational joint behavior of robots acting under uncertainty conditions, we have introduced the new approach of Fuzzy Coordination Graphs, which models the uncertainties of coordination problems more realistically based on Fuzzy Games. Also, a new fuzzy elimination algorithm for solving fuzzy coordination graphs and extracting the optimal joint action is introduced in this paper. An application of this approach in a multi robot coordination context is described and its efficiency in modeling the world uncertainty is discussed. The results show that the new approach helps the robots to be more responding and to act more rationally than deterministic coordination graphs.