Localization is one of the fundamental services for many applications in wireless sensor networks. However, the real-time localization systems (RTLS) based on wireless sensor networks developed so far are able to provide information about the positions of a moving target or mobile sensor node only with a limited precision, where the error between the real and the estimated position is in the range of few meters. In this work, we present the solution that significantly improves the results of the existing real-time localization systems. By fusing the rough position information from real-time localization systems together with the image information taken by only one camera, we can refine the estimated location of a target and pin down the estimated position error to few centimeters.

Keywords

Wireless sensor networks, localization, sensor fusion, test-bed experiments.