Reduction of ultrasonic indoor localization infrastructure based on the use of graph information

This paper presents a constrained navigation on a Metric Description Graph (MDG) based on the use of a H-Infinity Filter (H-∞) including the restriction on the graph as a fictitious observation. The main goal is to reduce the number of the required ultrasonic beacons for covering an extensive indoor area. This reduction of the localization infrastructure involves an increment of the error in the estimation of the position, which can be acceptable depending on the application (i. e. people navigation). This approach is especially useful when the target is navigating along corridors or narrow passing areas since it is not necessary an accurate position estimation. The proposal has been simulated in a localization area describing a real environment. It has been applied for different noise conditions of the relative information (odometry) comparing the performance between a traditional beacon infrastructure and the reduced one including the constraints of the MDG.