Improvement of strapdown inertial navigation using PDAF

A new application of PDAF (probabilistic data association filter) for improving the accuracy of autonomous strapdown inertial navigation systems (SINS) is presented. The proposed method is a terrain-aided navigation (TAN) algorithm based on landmark detection combined with a classical SINS. It is shown via a set of simulations that the method can improve significantly the precision of autonomous navigation if the landmark spatial density and quality of landmark detectors are good enough. This new concept of navigation called PDANF (probabilistic data association navigation filter) can be integrated with a relatively low cost in existing operational TAN systems     

Joint probabilistic data association for autonomous navigation

A new autonomous navigation scheme based on the joint probabilistic data association (JPDA) approach that processes landmark detections in the field of view (FOV) of an on-board sensor is developed. These detections-some true, some false-are associated to a set of stored landmarks and used to update the state of the vehicle. The results obtained from Monte Carlo simulations prove the ability of this navigation filter to perform in very high false alarm environments. In the different environmental conditions tested in the simulations, the performance of the JPDA navigation filter (JPDANF) is very close to that of the filter based on perfect data association. The very efficient cluster decomposition algorithm presented for the purpose of the navigation problem can also be used in many multitarget tracking applications