A Self-Organizing Localization Reference Grid
J. Eckert; F.J. Villanueva; R. German; F. Dressler
Cenference: ACM International Conference on Mobile Computing and Networking
Location: Chicago (EEUU)
Date: 20/09/2010 - 24/09/2010
Pages: 4-6
ISBN: 978-1-4503-0181-7
Volume: 14
Issue: 3
We propose a non-persistent indoor localization system using a self-organizing reference grid of autonomous robot systems. The key idea is to continuously maintain accurate relative positions between the robots using an enhanced mass spring relaxation model. The robots estimate distances between neighboring systems using an ultrasonic system, measuring both the time of flight based distance and the angle between the systems. The algorithm then adapts the local position of the robot in the grid according to its neighbors. We developed a mass spring relaxation model allowing to maintain a completely self-organizing reference grid. In mass spring, newly arriving nodes can introduce oscillations and self-localization might fail or take a long time to converge. Therefore, we first use the available grid to localize the arriving system with reference to the grid before including the robot as a new reference point – this initial self-localization is also used if a node cannot maintain a certain accuracy of its position. Misplaced nodes are detected and corrected by our enhancements. In turn, the grid is able to provide accurate localization services, e.g. for flying robots.