Optimizing the estimation procedure in INS/GPS integration for kinematic applications.

摘要

Conventional Kalman filtering has been the widely used and accepted procedure for integrating the inertial navigation systems (INS) with the global positioning system (GPS). In this respect, two main application areas are of interest to geomatics, direct georeferencing of imagery from mobile multi-sensor systems and estimating the anomalous gravity field by airborne gravity systems. In both cases, a conventional Kalman filter designed with a fixed estimation algorithm is used to fuse the INS and GPS streams of information. In such applications, the estimation environment is not always fixed. In a changing environment, imperfect a priori information and insufficient estimation time will affect the obtained accuracy of the integrated INS/GPS system if a fixed filter formulation is used. An adaptive filtering formulation, therefore, tackles the problem of imperfect a priori information and provides better tracking of the filter states.; In this research, an adaptive Kalman filtering approach is developed, analyzed, and proposed to replace the fixed (conventional) Kalman filtering approach for the INS/GPS integrated system. The adaptivity of the estimation procedure is carried out through the use of the measurement innovation sequence as piece-wise stationary process inside an estimation window to estimate either or both of the system noise matrix, Q or/and the measurement noise covariance matrix, R. In this dissertation, the performance of each of the two filters in kinematic environment is studied. Besides the flexibility it provides, the proposed adaptive approach has shown that an improvement of 10%–15% (rms) can be achieved to an airborne gravity system, and, in normal flight environments, an improvement of the attitude estimation by 20% (rms) could be achieved.; GPS positioning accuracy directly represents the positioning accuracy of the INS/GPS integrated system. It also indirectly enhances the attitude accuracy through the coupling effect between the filter states. Since the phase observable delivers the best possible GPS positioning information, its initial integer cycle ambiguity must be correctly resolved. It provides robustness and strength to the overall integrated system accuracy and reliability. A new method is developed in this research to resolve the GPS phase ambiguity using the so-called whitening filter. The method is discussed in this dissertation where it proved successful for short baselines and fair satellite coverage.