Precise Indoor Localization Method Using Dual-Facing Cameras on a Smart Device via Visible Light Communication


IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E100-A    No.11    pp.2295-2303
Publication Date: 2017/11/01
Online ISSN: 1745-1337
DOI: 10.1587/transfun.E100.A.2295
Type of Manuscript: Special Section PAPER (Special Section on Smart Multimedia & Communication Systems)
Category: Vision
visible light communication,  indoor localization,  visually impaired,  pedestrian navigation,  smart device,  

Full Text: PDF(3.4MB)>>
Buy this Article

In this paper, we propose a precise indoor localization method using visible light communication (VLC) with dual-facing cameras on a smart device (mobile phone, smartphone, or tablet device). This approach can assist the visually impaired with navigation, or provide mobile-robot control. The proposed method is different from conventional techniques in that dual-facing cameras are used to expand the localization area. The smart device is used as the receiver, and light-emitting diodes on the ceiling are used as localization landmarks. These are identified by VLC using a rolling shutter effect of complementary metal-oxide semiconductor image sensors. The front-facing camera captures the direct incident light of the landmarks, while the rear-facing camera captures mirror images of landmarks reflected from the floor face. We formulated the relationship between the poses (position and attitude) of the two cameras and the arrangement of landmarks using tilt detection by the smart device accelerometer. The equations can be analytically solved with a constant processing time, unlike conventional numerical methods, such as least-squares. We conducted a simulation and confirmed that the localization area was 75.6% using the dual-facing cameras, which was 3.8 times larger than that using only the front-facing camera. As a result of the experiment using two landmarks and a tablet device, the localization error in the horizontal direction was less than 98 mm at 90% of the measurement points. Moreover, the error estimation index can be used for appropriate route selection for pedestrians.