Flight Tests of Error-Bounded Heading and Pitch Determination with Two GPS Receivers

• Published in: IEEE transactions on aerospace and electronic systems 2012-01, Vol.48 (1), p.388-404
• Main Authors: Lau, L., Cross, P., Steen, M.
• Format: Article
• Language: English
• Subjects: Algorithm design and analysis; Algorithms; Antenna measurements; Antennas; Dynamical systems; Dynamics; Errors; Geographic information systems; Global Positioning System; Integrated approach; Position measurement; Receivers; Satellite navigation systems; Satellites; Studies
• ISSN: 0018-9251; 1557-9603
• DOI: 10.1109/TAES.2012.6129643

This paper describes a heading and pitch determination algorithm using single-epoch measurements from two GPS receivers. The algorithm aims to provide accurate heading and pitch solutions with controlled quality. Since the separation of the two antennas (baseline length) is known, this length has been used in the algorithm to constrain the heading and pitch solutions in the final stage of the estimation process. Moreover, error bounds (EB) of heading and pitch solutions are calculated at each epoch in order to provide statistical bounds for their errors. The proposed algorithm has been tested with high (frequent turns with quick change in pitch angle to ±50° and bank angle to ±80°) and low dynamic flight test data collected in Germany and the solutions are compared with high-accuracy heading and pitch provided by a high-grade GPS/INS system. Test results of achieved accuracies and success rate of bounds on heading and pitch errors are described. This attitude determination work is part of the ANASTASIA (Airborne New Advanced Satellite techniques & Technologies in A System Integrated Approach) project, which was a 6th Framework programme of European Commission Project that was led by Thales Avionics, France. It has been found that about 96% of accepted solutions in the low dynamic data sets have heading and pitch errors less than the ANASTASIA required accuracy of 0.4°. Moreover, the rms error and error bounding analyses of the results show that the performance of the heading and pitch algorithm on the low dynamic data sets is better than that of the high dynamic data sets. Extensive analysis of the relatively poor performance on the high dynamic data sets is described and the problems/error sources are identified and discussed.