LedMapper: Toward Efficient and Accurate LED Mapping for Visible Light Positioning at Scale
Liang Qing, Yuxiang Sun, Chengju Liu, Ming Liu, Lujia Wang
Abstract
Indoor localization of high accuracy has been widely interested. Among competitive solutions, visible light positioning (VLP) is promising due to its ability to deliver high-accuracy 3-D position and orientation with low-cost sensors by sharing the LED lighting infrastructure widespread in buildings. Most VLP systems require a prior LED location map for which manual surveys are costly in practical deployment at scale. In this article, to address this difficulty, we propose a novel system for efficient and accurate offline mapping of LEDs for VLP. With input from visual–inertial sensors and existing or surveyed priors, it builds the map by posing a full simultaneous localization and mapping (SLAM) problem within a factor graph formulation. Compared to manual surveys, it greatly saves human labor and time while yielding an accurate and workspace-aligned LED map. With real-world experiments in a room-scale testbed and a <inline-formula> <tex-math notation="LaTeX">$15\times $ </tex-math></inline-formula> larger lab office, we extensively evaluate the LED mapping system to verify its efficacy and performance gains.