Real-time Infrastructureless Indoor Tracking for Pedestrian Using a Smartphone

Demands for accurate pedestrian indoor tracking on mobile platforms have been increasing rapidly these years, however, conventional tracking system based on dead reckoning suffers from inherent sensor drift, which leads to bad performance in long-term tracking. In this paper, an infrastructureless pedestrian dead reckoning system called iPDR is proposed and implemented on a ready-to-use smartphone, offering real-time indoor tracking for pedestrian without previous knowledge of the target area. The iPDR fuses the sensor data using a novel filter approach coined as hybrid orientation filter, which deploys a combination of the information filter and the complementary filter to achieve accurate heading estimation. A fast loop closure detection method called rapid loop detection is also presented in the iPDR to calibrate the tracking trajectory significantly whenever encountering a loop. Experiment results show that the iPDR system achieves typically sub-meter accuracy in localization and has an average orientation error of 2.89 degrees over a total turning of 1800 degrees.