カンファレンス (国際) Camera-Based Position Estimation using Frequency-Multiplexed Luminance Gradient

Shota Shimada, Hiroaki Murakami (the University of Tokyo), Kota Tsubouchi, Takuya Sasatani (the University of Tokyo), Yoshihiro Kawahara (the University of Tokyo)

The 22nd International Conference on Pervasive Computing and Communications (PerCom 2024)


We present a new indoor positioning method to overcome the challenges faced by Visible Light Positioning (VLP), specifically its dependency on multiple light sources and constraints in environments with high ceilings. Our method employs omni-directional LEDs, each modulated at a unique frequency, mounted on the ceiling. The system calculates the position by analyzing the luminance and gradient of diffused light reflected on the ceiling, captured by the user’s smartphone front camera. This allows for accurate distance and angle estimation between the point of interest and LEDs with just two light sources. The inherent characteristics of ceiling surfaces, being largely unobstructed, enhance the diffusion of modulated light, enabling the capture of signal even from partial and indirect views. Experimental validation in a setting with a 3.6m baseline between lights resulted in a 0.462m average positioning error and a 21.7-degree average orientation angle error, showcasing its efficiency and potential for broad indoor application.

Paper : Camera-Based Position Estimation using Frequency-Multiplexed Luminance Gradient新しいタブまたはウィンドウで開く (外部サイト)