Mobile client localization using data fusion of wireless LAN signals and three-dimensional camera image sequences

This paper proposes a method for determining the location of a mobile client using a data fusion of localization results based on wireless LAN signals and 3-D camera-image sequences in large structures, such as factory buildings, plants, and public spaces. Data pertaining to the physical location of personal electronic devices or mobile robots are important for information services and robotic applications. In indoor structures, visual localization using a one-shot measurement from a camera image does not work well because of the symmetry of the structures; there are multiple candidates for the estimated location in symmetric buildings. There is also some uncertainty in location results estimated using wireless LAN signals, due to uncertainties in the signal strength caused by the reflection and multipath fading of the signals from the wireless LAN access points. We propose a sensor data fusion scheme whereby the estimated location of the client is used to search a database of natural feature points obtained by the camera-image sequences. Experimental results show the feasibility of the proposed method.