The Global Positioning System (GPS) can be used to determine the attitude of a vehicle. To do this, multiple receivers should be used with the antennas in known fixed positions in the vehicle body frame. Since the estimation of the attitude angles needs high precision, the carrier phase measurements (with millimeter precision) are used in combination with the pseudorange measurements. Nevertheless, the phase measurements have the disadvantage of having an unknown integer number of cycles, named phase ambiguities. To find these phase ambiguities, the search method LAMBDA is used, however, this method is only efficient with dual-frequency measurements (L1 and L2 bands). When low-cost receivers with single-frequency measurements (L1 band) are used, another method, known as Ambiguity Filter, must be applied to stabilize the solution given by the LAMBDA method. This filter makes use of the prior knowledge of the distance between the antennas. As soon as the phase ambiguities are found, the baseline vector, that points from one antenna to the other, can be estimated and used to determine the attitude angles. When using carrier phase measurements, it is possible to achieve results for the baseline vector with millimeter precision, however, this only happens when the solution for the ambiguities is correctly determined. If the Ambiguity Filter fixes an incorrect set of ambiguities, the solution for the baseline (and attitude angles) starts to degrade, which may not be immediately visible.
This paper proposes a new implementation within the Ambiguity Filter in order to improve the detection of false locks for the solution of the phase ambiguities. This implementation is made by creating a new verification step in the already existing Ambiguity Filter. With this verification step, it was possible to detect false locks in incorrect phase ambiguities and to decrease the Ambiguity Filter sensitivity to the value of the baseline length that is given as input to this method. The results presented in this paper were based on field tests performed in static conditions with a single-baseline setup in the university campus, allowing the estimation of heading and pitch.