Multi-Epoch Kriging-Based 3D Mapping Aided GNSS using Factor Graph Optimization

The performance of the Global navigation satellite system (GNSS) is limited in urban canyons because the signals are reflected over building surfaces. Urban localization can be enhanced using the resources associated with 3D building models. Different 3D mapping aided (3DMA) GNSS algorithms have been proposed, in which the 3D building models are used to aid the positioning. Recently, the candidate-based 3DMA GNSS framework was proposed as a particle filter-based approach. The distributed particles are examined, and the particles that best match the observed measurements, that is, the particles with the minimum cost, are identified as the receiver location. Such particle sampling approaches are inexpensive but incur a high computational load. To decrease the computational load, in this study, the cost function of a 3DMA GNSS is modelled using a Kriging-based interpolation method based on the sample data provided by a set of sparsely distributed particles. The regressed model is used to estimate the receiver state through an iterative nonlinear least-squares method. Moreover, the proposed Kriging-based 3DMA GNSS is integrated with the velocity estimated by Doppler measurements through factor graph optimization (FGO). The optimization process can enhance the 3DMA GNSS positioning accuracy and robustness even when smartphones and commercial-level GNSS receivers are used. The proposed method achieves a root mean square error (RMSE) of less than 10 m for most of the designed experiments in Hong Kong urban canyons.