Research Of Embedded ARM-LINUX GPS Positioning System Based On Data Fusion

With the promotion of the Navigation technology and control theory , integrated navigation system already has many of the navigation sensor module, multi-sensor data fusion become a research hotspot. However, due to the existence of each sensor can not overcome its own shortcomings, for example, by blocking factors GPS, inertial navigation zero drift and so on, so robust real-time positioning of the target becomes very difficult. This paper focuses on realization of robustness and real-time multi-sensor information fusion algorithm in the Embedded System. This major research work and results as follows:1. To build the embedded environment, this topic chooses ARM9 platform as hardware platform, embedded Linux as the operating system, then the kernel and file system have been transferred to GEC2440 development board. Embedded Linux serial port is used to communicate the GPS module and the INS module for data extraction, at last adopts the program of information fusion algorithm burned in the Flash in the development board to process positioning data.2. In this paper, information fusion algorithms are studied. Firstly Kalman filter model of the sensor module is established, in view of the heavy calculation burden of multi-sensor information fusion, a distributed federated Kalman filter algorithm is introduced. The accuracy of filter model is improved through the redistribution of information. In accordance with filtering instability of integrated navigation systems, the information fusion model based on wavelet transform and neural network is chosen to enhance robustness. Simulation results show that the algorithm in the same sample set has a higher stability and real-time than the centralized Kalman filter algorithm.3. According to the method of data fusion processing of multi-sensor in order to improve the reliability and precision of data processing, data multi-scale distributed fusion estimation algorithm is put forward for estimating the data of multi-sensor that is based on Kalman filter. The algorithm firstly establishes the system dynamic equation and observation equation, then the data is processed on different scales by wavelet transform, and realization steps of the algorithm are concluded. Finally, by the simulation in GPS/INS, the result testifies the algorithm can improve the precision of multi-sensor's data processing. At last, this paper's algorithm is applied to mobile robot platform,and to actively locate the target.