Download or read book Integrating Centralized and Decentralized Approaches for Multi-robot Coordination written by Ke Xu and published by . This book was released on 2010 with total page 107 pages. Available in PDF, EPUB and Kindle. Book excerpt: Autonomous multi-robot systems play important roles in many areas such as industrial applications for repetitive tasks, explorations in hazardous environments, and military missions in extreme conditions. Many existing coordination strategies are developed for two general types of multi-robot systems including strongly centralized systems and completely decentralized systems. For strongly centralized systems, the global information including the environment as well as the locations of all the robots is shared. It is typical for small number of robots in well structural environments and is not robust to dynamic environment or failures in communications and other uncertainties. For completely decentralized systems, each robot is executing its own control schemes completely autonomously. There are no specified leaders throughout the mission, and the team organization does not have a set structure. In many real-world applications, it is beneficial to use so-called weakly centralized systems, in which the leader robot is not specified a priori, but it is selected dynamically during the mission to guide the robot team through dynamic environments or other uncertainties. It is very challenging to develop coordination strategies for this type of systems because of the dynamic nature of the team structures. The strategies should not only allow for on-line leader role selection but also enable formation decomposition and reconfiguration whenever necessary. In this thesis, we describe a general coordination framework for weakly centralized multi-robot systems that integrates the features from both strongly centralized and completely decentralized coordination strategies at the individual robot level. The framework allows the robots to reconfigure the formation dynamically in the presence of obstacles or other uncertainties in the environment, and promotes the main advantages of multi-robot systems such as flexibility and modularity. Since the control schemes can be decentralized and this framework allows for the selection of the motion planner and local controller for a given task, the framework can be naturally applied to multi-robot systems with larger scales. We have implemented this framework on a team of two-wheeled differential driven mobile robots. Significant results from numerical simulations and experiments have been obtained to demonstrate that the coordination schemes are effective and robust, and the framework is viable and can be scaled to relative large scale multi-robot systems.