智能手机是驾车出行中常用的定位导航工具，车道级的定位精度和可靠性对导航服务至关重要。智能手机原始GNSS芯片解定位精度和可靠性难以满足车道级定位需求，现有研究中采用的改进方案亦在算法结构、优化策略、适用场景等方面具有局限性。本文针对智能手机的车道级定位需求，充分考虑智能手机观测质量低的特性，提出了一套适用于智能手机的GNSS/INS融合定位算法，充分利用现有智能手机的传感器，综合提升其定位精度和可靠性，实现车道级的定位效果。 本文基于传统的DGNSS、RTK技术，采用抗差Kalman滤波、质量控制等抗差策略，提出了一种适用于智能手机的抗差RTK定位算法。为进一步提升智能手机定位精度和可靠性，本文基于扩展Kalman滤波（EKF）和因子图优化（FGO）算法，采用松组合和紧组合融合结构，提出了四种适用于智能手机的RTK/INS融合定位算法。最后，本文基于智能手机路测试验数据，设计了GNSS芯片解、传统RTK、抗差RTK、基于EKF的RTK/INS松组合、基于EKF的RTK/INS紧组合、基于FGO的RTK/INS松组合、基于FGO的RTK/INS紧组合等7种定位算法间的对比试验。结果表明： 在仅使用GNSS数据时，抗差RTK定位算法可有效提升智能手机定位的精度和可靠性，道路横向CEP90精度1.43m，比传统RTK定位精度提升16%，与智能手机GNSS芯片解相比提升52%，定位可靠性为91.3%。 在使用GNSS/INS数据融合定位时，基于EKF的RTK/INS紧组合算法和基于FGO的RTK/INS紧组合算法效果较好，其中前者更适用于实时解算、后者更适用于事后处理，两者道路横向定位CEP90精度分别为1.20m和1.28m，相比于抗差RTK定位精度分别提升15%和10%，定位可靠性分别为93.4%和94.3%。 本文提出的智能手机抗差RTK定位算法相较于传统RTK算法，在抗差策略和算法数学模型上有一定改进，可有效提升智能手机GNSS定位的精度与可靠性；在此基础上对比研究并提出了适用于实时解算的基于EKF紧组合RTK/INS融合定位算法、适用于事后解算的基于FGO紧组合RTK/INS融合定位算法，为中高端智能手机的车道级定位提供了组合导航解决方案。

Smartphones are commonly used as navigation tools when driving. The accuracy and reliability of lane-level positioning are crucial to navigation services. The positioning accuracy and reliability of the smartphone GNSS chip solution cannot meet the requirement of lane-level positioning. The improvement schemes proposed in the existing research also have limitations in terms of algorithm structure, optimization strategies, and applicable scenarios. This paper addresses the need for lane-level positioning and proposes a GNSS/INS fusion positioning algorithm suitable for smartphones. By fully utilizing the sensors available in smartphone, considering the characteristics of low-quality observations, this algorithm aims to comprehensively enhance the positioning accuracy and reliability of smartphones, ultimately achieving lane-level positioning performance. Based on the legacy DGNSS and RTK technology, this paper proposes a robust RTK positioning algorithm suitable for smartphones by using robust Kalman filter, quality control and other robust strategies. The CEP90 horizontal accuracy is 0.034m in the open-sky stationary test for smartphone, with a 90.2% ambiguity fix rate. In order to further improve the positioning accuracy and reliability of smartphones, based on the extended Kalman filter (EKF) and factor graph optimization (FGO) algorithms, either with loose or tight combination (LC/TC) structure, four kinds of RTK/INS fusion positioning algorithm suitable for smartphones are proposed. Finally, based on the smartphone road test data, comparative experiments are designed among 7 algorithms, which are GNSS chip solution, legacy RTK, robust RTK, RTK/INS EKF-LC, RTK/INS EKF-TC, RTK/INS FGO-LC, RTK/INS FGO-TC. The results indicate that: When only GNSS observation is used, the robust RTK positioning algorithm can effectively improve the accuracy and reliability of smartphone positioning. The CEP90 road’s lateral accuracy is 1.43m, which represents a 16% improvement compared to the legecy RTK and a 52% improvement compared to the smartphone GNSS chip solution, and the positioning reliability is 91.3%.When using both GNSS and INS observations for positioning, the algorithms based on EKF-TC and FGO-TC perform well, with CEP90 road’s lateral accuracy of 1.20m and 1.28m respectively, which represents a 15% and 10% improvement compared to the robust RTK positioning algorithm. And the positioning reliability is 93.4% and 94.3% respectively. The robust RTK positioning algorithm for smartphones proposed demonstrates improvements in robustness strategies and mathematical models compared to the legacy RTK algorithm, effectively enhancing the accuracy and reliability of smartphone GNSS positioning. On this basis, the paper presents a comparative study and concludes that EKF-TC RTK/INS algorithm is suitable for real-time calculation while the FGO-TC RTK/INS algorithm is suitable for post-processing. These algorithms provide a fusion navigation solution for lane-level positioning in smartphones of mid-to-high-cost.