Optimal smoothing techniques in aided inertial navigation and surveying systems

Tactical-grade, low-cost Inertial Navigation Systems (INSs) and Micro-Electro-Mechanical Systems (MEMS) inertial sensors have gained great interests in civilian and commercial fields during the last decade. The Global Positioning System (GPS) is recognized as the ideal complement to INS by offering absolute positioning information and consistent accuracy in open sky to overcome the problem of INS time-dependent error growth. However, GPS suffers from degraded signal acquisition or poor satellite geometry when a vehicle is traveling in urban, dense foliage or canyon areas. In addition, the GPS signals will be totally unavailable in the isolated environments such as tunnels, mines or indoor areas. Hence, alternative aiding instruments or techniques such as odometers, non-holonomic constraints, Zero-velocity Updates (ZUPTs) and Coordinate Updates (CUPTs) become essential to restrict the accumulated time-dependent errors of a stand-alone INS. While Kalman filter is widely employed as the real-time estimation method to fuse the multi-sensor information, optimal smoothing will be utilized as the post-processing methodology to provide better navigation solutions.;In this research, two different fixed-interval smoothing algorithms will be utilized and evaluated. The first algorithm is the Two Filter Smoother (TFS), while the second algorithm is the Rauch-Tung-Streibel Smoother (RTSS). The TFS is performed by combining the results of Forward Kalman Filtering (FKF) and Backward Kalman Filtering (BKF) through minimizing the smoother error covariance. The traditional TFS was not applicable for some INS-based multi-sensor systems because of the high nonlinear characteristics in the INS navigation equations. Thus, the revised TFS algorithm will be derived in details. The performance of Kalman filtering as well as the optimal smoothing methodologies is evaluated in three application conditions: land-vehicle navigation, pipeline surveying, and horizontal/vertical indoor building navigation, surveying and mapping. The integration strategies of INS and the aiding techniques mentioned earlier are proved to be applicable and effective. The results of all investigated applications show that the TFS substantially improve the position estimation accuracy over the corresponding filtered solution. In addition, the estimation efficiency of the TFS is comparable to the commonly used RTSS.