An Improved PDR/UWB Integrated System for Indoor Navigation Applications
Shuli Guo, Yitong Zhang, Xinzhe Gui, Lina Han
Abstract
The challenges of the inertial navigation system based pedestrian dead reckoning (PDR) are mainly stochastic errors and serious accumulated errors caused by sensor variance, while the ultra-wideband (UWB) based positioning approaches are vulnerable to the external environment and produce many outliers under non-line-of-sight (NLOS) conditions. To overcome these shortcomings, this paper proposes a three-level improved PDR/UWB integrated system, in which the gait detection is first performed by a dual-frequency Butterworth filter, and the step length is accurately estimated based on a linear combination model. Then the position of the target is calculated by combining the step length and the heading information but calibrated periodically through the drift-free output of the UWB system. Finally, the noise distribution is dynamically adjusted through the NLOS assessment function, and the positioning accuracy is improved at information fusion level using the proposed variable noise variance Kalman filter. The positioning data is collected by our integrated small-scale sensors in both LOS and NLOS environments, and experiment results have demonstrated that the proposed PDR/UWB integrated system can significantly improve the accuracy of positioning information and can apply in indoor navigation applications.