Improving inertial Pedestrian Dead-Reckoning by detecting unmodified switched-on lamps in buildings

• Published in: Sensors (Basel, Switzerland) Switzerland), 2014-01, Vol.14 (1), p.731-769
• Main Authors: Jiménez, Antonio R, Zampella, Francisco, Seco, Fernando
• Format: Article
• Language: English
• Subjects: Accuracy; Algorithms; Buildings; Calibration; Estimates; Geographic Information Systems; Humans; Hypotheses; Illumination; Indoor; indoor localization; Inertial; Lamps; Light; light/illumination; Localization; Matching concept; Orientation; pedestrian dead-reckoning; Position (location); Robots; Sensors; signals of opportunity; smartphone; Smartphones; Usability; Walking
• ISSN: 1424-8220; 1424-8220
• DOI: 10.3390/s140100731

This paper explores how inertial Pedestrian Dead-Reckoning (PDR) location systems can be improved with the use of a light sensor to measure the illumination gradients created when a person walks under ceiling-mounted unmodified indoor lights. The process of updating the inertial PDR estimates with the information provided by light detections is a new concept that we have named Light-matching (LM). The displacement and orientation change of a person obtained by inertial PDR is used by the LM method to accurately propagate the location hypothesis, and vice versa; the LM approach benefits the PDR approach by obtaining an absolute localization and reducing the PDR-alone drift. Even from an initially unknown location and orientation, whenever the person passes below a switched-on light spot, the location likelihood is iteratively updated until it potentially converges to a unimodal probability density function. The time to converge to a unimodal position hypothesis depends on the number of lights detected and the asymmetries/irregularities of the spatial distribution of lights. The proposed LM method does not require any intensity illumination calibration, just the pre-storage of the position and size of all lights in a building, irrespective of their current on/off state. This paper presents a detailed description of the light-matching concept, the implementation details of the LM-assisted PDR fusion scheme using a particle filter, and several simulated and experimental tests, using a light sensor-equipped Galaxy S3 smartphone and an external foot-mounted inertial sensor. The evaluation includes the LM-assisted PDR approach as well as the fusion with other signals of opportunity (WiFi, RFID, Magnetometers or Map-matching) in order to compare their contribution in obtaining high accuracy indoor localization. The integrated solution achieves a localization error lower than 1 m in most of the cases.