Xiong Zhi

Research on pedestrian navigation system aided by indoor geomagnetic matching

Indoor pedestrian navigation technology is playing a more and more important role in our life. For low cost strap-down navigation system, error is large and spreads rapidly. Hence, this paper proposes a zero velocity update (ZUPT) detection algorithm based on acceleration magnitude square variance detector (AMSVD), angular rate magnitude square variance detector (ARMSVD), acceleration magnitude square detector (AMSD) and angular rate magnitude square detector (ARMSD) to distinguish pedestrian kinetic characteristics and restrain the divergence of errors. Meanwhile, this paper studies the pseudo measurement information in ZUPT and builds a Kalman filter for optimal estimation of errors. Simultaneously, we study the indoor magnetic anomaly and put forward a kind of indoor geomagnetic matching algorithm based on k-nearest neighbor algorithm (KNN) and a multi-layer constraint which is suitable for engineering application. Finally, we construct an integrated navigation system to realize optimal fusion of MEMS pedestrian navigation system and geomagnetic matching model to improve accuracy and adaptation and provide continuous position information, besides, it weakens the effect of MEMS sensor over range. Tests show that the accuracy of the integrated navigation system is increased by more than 55% and the final position error is only 0.4%.

An improved method of SINS/CNS integrated navigation algorithm based on dual quaternions with gyro error corrected online

In consideration of the high dimension problem in the existing SINS/CNS integrated navigation system based on dual quaternions, the observable degree analysis method of system states based on singular value decomposition is introduced. According to the analysis result, the state components with good observability are estimated and the bad ones are removed, thus the computation time of Kalman filter is significantly decreased and the real-time performance of SINS/CNS is improved. To ensure that inertial navigation can also maintain high accuracy over a period of time when celestial auxiliary information is lost, an improved method of SINS/CNS integrated navigation algorithm based on dual quaternions with gyro error corrected online is proposed. The Kalman filter estimates constant gyro drift with the aid of celestial information and utilizes the calibration value to compensate gyro output online. The simulation results demonstrate that the proposed method can realize the effective estimation of constant gyro drift. In addition, the accuracy of inertial navigation is improved effectively.

Research on information intermittent fusion of ASPN system in the long-endurance UAVs

In the long-endurance UAVs, the conventional INS/GNSS integrated navigation system is widespread used. But with the increasing man-made interference and electromagnetic environment worsening, the GNSS system will be easily failing in location. In that case, using other aided navigation systems based on the characteristics of objects & geophysical field is necessary. However, the aided navigation systems such as celestial navigation, scene matching navigation and terrain aided navigation cannot work continuously during the long time flight. This will cause wrong detection by the traditional method. So a scheme of seamless navigation based on the characteristics of objects & geophysical field and an improved algorithm of residual chi-square fault detection are proposed in this paper. The simulation results show that the designed method can effectively solve the problem of wrong detection, and ensure the seamless navigation system works properly during the flight mission. The Navigation Research Center (NRC) of Nanjing University of Aeronautics and Astronautics (http://www.nuaanrc.com) is engaged in the research of navigation technology, and has done a series of research on all source information fusion. Up to now, a great deal of achievement has been made in the following fields: Strap-down Inertial Navigation System, Inertial Integrated Navigation System, Global Satellite Positioning System, Terrain Aided Navigation System, Fault-Tolerant Technique of Multi Inertial Navigation Subsystem, Modern Optimization Filtering Theory, Precision Measurement System and so on. This paper is part of the NRC research achievement.

Study on unequal-interval federated filter algorithm based on dynamic information distribution

Federated filtering algorithm cannot fuse the local information at every cycle due to the different data update frequency of various navigation devices in integrated navigation system, hence may loss information and induce factitious errors. The traditional unequal-interval algorithms mainly deal with the frequency inconsistency of navigation devices, but did not take the dynamic allocation of information into account. This paper proposes an Unequal-interval Federated Filter Algorithm based on dynamic information distribution. This method can dynamically adjust the information distribution factor according to the number of effective local filters, system estimation accuracy and the observability of the local filters, reset the local filters, optimally fuse different measurement information, and further improve the performance of federated filtering algorithm.

Research on improved SIFT feature matching arithmetic in the SAR vision matching aided navigation

The performance of the image matching arithmetic will affect the precision and dependability of the SAR vision matching aided navigation system. When the real-time SAR image is influenced by the noise badly, the image matching result will be affected greatly. In order to improve the performance of the image matching arithmetic, improved SIFT feature matching arithmetic based on the wavelet transform is researched in this paper. The improved arithmetic adopted wavelet transform to restrain the noise of the real-time SAR image. At the same time, the paper utilized the multi-resolution analysis of wavelet transform to minish the realtime images size. Consequently, it enhanced the precision and the rapidity of the matching arithmetic a lot. The simulation result and contrastive analysis conclusion made it clear that, compared to the basic SIFT matching arithmetic; the arithmetic proposed in this paper shows the better predominance in the matching positioning precision and speed.