Abstract | ||
---|---|---|
Most passive location systems (e.g. pressure sensitive floor mats) inherently do not come with an identification mechanism, i. e. are ambiguous concerning the number of persons who cause readings: Two persons close beside each other could seem like one single person. In contrast, most active location systems operate with user-bound tags, and thus directly give the number of used tags. Possibly, both sensor types may be used in conjunction, but the number of persons and their respective locations need to be simultaneously estimated. This paper presents an approach that allows to fuse observations from both sensor types. We introduce a probabilistic model for representing the joint density of number and location of persons as well as the available sensor modalities. We show how this model can be used for estimating the posterior placement density given the sensor observations, applying a Gibbs sampling approach. |
Year | DOI | Venue |
---|---|---|
2012 | 10.1109/UPINLBS.2012.6409760 | Ubiquitous Positioning, Indoor Navigation, and Location Based Service |
Keywords | Field | DocType |
probability,sensors,Gibbs sampling approach,multimodal localization sensor setup,passive location systems,probabilistic model,sensor modalities,user number estimation | Modalities,Computer vision,Statistical model,Artificial intelligence,Fuse (electrical),Location systems,Mathematics,Modal,Gibbs sampling,Machine learning | Conference |
ISBN | Citations | PageRank |
978-1-4673-1908-9 | 1 | 0.42 |
References | Authors | |
5 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Gernot Ruscher | 1 | 4 | 0.89 |
Thomas Kirste | 2 | 76 | 9.35 |