Complex urban environments limit the reception of global navigation satellite system (GNSS) signals, increasing the difficulty of ambiguity resolution (AR) for GNSS attitude determination based on carrier-phase observations. To constrain the Least-squares Ambiguity Decorrelation Adjustment (LAMBDA) and improve the success rate and accuracy of AR, a fixed length between the antennas is proposed. Heading angle- and length-constrained LAMBDA (HLC-LAMBDA) and converted baseline vector-constrained LAMBDA (VC-LAMBDA) were proposed because baseline length-constrained LAMBDA (BC-LAMBDA) cannot guarantee three-dimensional accuracy. Because BC-LAMBDA, HLC-LAMBDA, and VC-LAMBDA are based on multi-antenna attitude determination, this study introduces heading angle-constrained LAMBDA (HC-LAMBDA), which can also be applied to single-antenna GNSS attitude determination scenarios. To identify the optimal ambiguity candidate in BC-LAMBDA, HLC-LAMBDA, HC-LAMBDA, and VC-LAMBDA, search and shrink strategy were performed based on the boundary function rather than an exhaustive search. Additionally, a simple validation was implemented to ensure the reliability of the ambiguous candidates during the search process. The car experiment was conducted in an urban environment using a low-cost Ublox-F9p. The test of different maximum candidate numbers indicated that the accuracy of the baseline solution did not continue to increase as the maximum candidate number increased in the search space. The constrained LAMBDAs (C-LAMBDAs) with the boundary function had a higher fix rate and smaller bias than those of the exhaustive search; however, the consumed time by the boundary function was larger owing to the Modified Search (M-Search) in the shrink process. The results also show that different constraints have different effects; HC-LAMBDA has a similar performance to HLC-LAMBDA in terms of heading angle, which should be considered when the baseline length is unknown. The baseline vector of VC-LAMBDA performed better in three dimensions if there was an accurate constraint on the baseline vector.

复杂的城市环境限制了全球导航卫星系统（GNSS）信号的接收，增加了基于载波相位观测的GNSS姿态确定模糊度（AR）难度。为了约束最小二乘模糊度解相关调整(LAMBDA)并提高AR的成功率和精度，提出了天线之间的固定长度。由于基线长度约束LAMBDA（BC-LAMBDA）不能保证三维精度，提出了航向角和长度约束LAMBDA（HLC-LAMBDA）和转换基线矢量约束LAMBDA（VC-LAMBDA）。由于BC-LAMBDA、HLC-LAMBDA和VC-LAMBDA都是基于多天线姿态确定，因此本研究引入了航向角约束LAMBDA（HC-LAMBDA），它也可以应用于单天线GNSS姿态确定场景。为了确定 BC-LAMBDA、HLC-LAMBDA、HC-LAMBDA 和 VC-LAMBDA 中的最佳模糊度候选，基于边界函数而不是穷举搜索来执行搜索和收缩策略。此外，还实施了简单的验证，以确保搜索过程中模糊候选者的可靠性。该汽车实验是在城市环境中使用低成本的 Ublox-F9p 进行的。对不同最大候选数的测试表明，基线解的准确率并没有随着搜索空间中最大候选数的增加而继续增加。带有边界函数的约束LAMBDA（C-LAMBDA）比穷举搜索具有更高的修复率和更小的偏差；然而，由于收缩过程中的修改搜索（M-Search），边界函数消耗的时间更大。结果还表明，不同的约束条件具有不同的效果；HC-LAMBDA 在航向角方面与 HLC-LAMBDA 具有相似的性能，当基线长度未知时应考虑这一点。如果对基线向量有精确的约束，VC-LAMBDA 的基线向量在三个维度上表现更好。