Decentralized coordination for multirobot exploration

In order for multiple robots to explore an unknown environment, they need a strategy for cooperation. We propose a strategy where robots share percept ual information but maintain separate maps. These robots make independent d ecisions about where to explore using frontier-based exploration. This allo ws each robot to use the information from other robots while retaining the capability to explore independently. As a result, the multirobot team is ro bust to the loss of communications or the loss of individual robots. We hav e implemented this system on real robots and have demonstrated that they ca n explore and map indoor environments effectively. (C) 1999 Elsevier Scienc e B.V. All rights reserved.