摘要

GPS compass equipped with ultra-short baselines (2-5 carrier wavelengths) is a low-cost heading and elevation indicator for many size-limited applications such as small vehicles, robots, miniaturized satellites or aircrafts. Since the single epoch GPS compass can guarantee a total independence from carrier phase slips, recent researches focus on the high success rate ambiguity resolution for this study. However, the accuracy is another important problem, especially for the size-limited application, since a shorter baseline results in a lower accuracy. Although the Kalman filter can be used for accuracy improvement, we might have information about a system that the Kalman filter does not incorporate. Given that the baseline length is known as well for GPS compass, we can integrate the nonlinear constraint into the state estimation of baseline and get better filtering performance than the Kalman filter provides. Based on this scheme, the constrained Moving Horizon Estimation (MHE) approach can be used for the accuracy improvement. Both simulations and actual experiments have been performed to verify the effectiveness of this method. The achieved results are notable to the GNSS compass software developers.