Visible Light Passive Indoor Tracking Using Background Subtraction

Most of the currently used active visible light positioning approaches require the involvement of the target itself in the positioning and are very susceptible to changes in the background. This paper proposes a passive visible light indoor tracking approach that assumes no direct interaction between the target and the luminous elements, and is shown to be robust against background changes by introducing the concept of background subtraction (BS) to fix the issues of background changes. Using background subtraction, the background is dynamically updated without any prior information about such changes. The position estimate is determined by estimating the object impulse response (OIR) which is the reflections from the target alone after having subtracted the background from the measured impulse responses (IRs) at every time scan. Positioning is accomplished by selecting the position with the maximum likelihood (ML) between the difference signal and the OIR. A converted measurement Kalman filter (CMKF) is used for tracking the target moving under a nearly constant velocity model. The simulation results ensure accurate localization and tracking based on the proposed approach with positioning and speed RMSEs of 3 cm and 1.5 cm/s, respectively, under severe background changes.