Improving indoor position estimation by combining active TDOA ultrasound and passive thermal infrared localization

Localization techniques for indoor environments are constantly evolving. In this paper, an approach that combines two different techniques is presented. In detail, the passive thermal infrared localization system (ThILo) is used in combination with the ultrasound and RF based localization system Cricket. In contrast to other indoor localization systems, ThILo exploits the natural thermal radiation of humans to determine their position. This key feature allows the localization of several persons in home environments without the need of wearable hardware (tag). Therefore, it benefits in various areas where a tag is undesired or not applicable e.g. assisted living apartments. However, ThILo does not provide any means of target identification which is a prerequisite for user customized location based services. This is why, the Cricket system, which only provides unreliable and also sparse TDOA measurements, is added. Furthermore, a SMC Probability Hypotheses Density (SMC-PHD) filter is used to estimate the location of several persons in a room. The fusion of the ThILo and Cricket measurements is achieved by using a bank of SMC-PHD correctors in parallel which in turn leads to multiple hypotheses. For the combination of these hypotheses into one, various methods are compared. Additionally, the localization results of the combined systems are compared against the single systems.