Find in Library

To improve the indoor positioning accuracy, the methods of position fingerprint and pedestrian dead reckoning (PDR) are used to study the indoor positioning algorithm.For position fingerprint algorithm, the fingerprint database is optimized through the offline data training phase, and the optimization of the nearest neighbor algorithm is carried out by limiting the region weighted value <i>K</i> through the online real-time matching phase.For PDR algorithm, the self-adaptation peak detection algorithm used for step frequency detection is proposed.The improved nonlinear model is used for step size estimation.Moreover, The gyroscope's information are fused to magnetometer's information in the heading estimation.Finally, the unscented Kalman filter is used to fuse the position fingerprint algorithm and PDR method, which improves the positioning accuracy and the practicability of the fusion algorithm is verified by the positioning system.