A QoS enabled visual sensor-system applied to the problem of localizing mobile platforms in indoor environments

Abstract

Here we present a sensor-system that integrates two low cost CMOS cameras, LED illumination, 3-axis MEMS gyroscopes and accelerometers, and a data processing node in a compact shell. The sensor-system supports different modes of operation including incremental position tracking, visual self localization and mapping, inertial navigation, pose estimation using artificial landmarks, as well as combinations of the mentioned techniques depending on application scenario and environment conditions. In contrast to existing work we use the output of the inertial measurement unit not only to provide estimates on the systems pose but also to suppress camera output during unfavorable movement thus improving the performance of image processing while reducing utilization of overall system resources, resulting in a dynamically adapting quality of service