| Popis: |
In dense urban areas, GNSS signals are blocked and reflected by the buildings, severely degrading the positioning accuracy. By using 3D mapping of the buildings to predict which signals are visible and which are blocked, the accuracy can be substantially improved. Many different techniques have been demonstrated since the start of the decade, showing that errors can be reduced from tens of meters to a few meters. However, a practical implementation of 3D mapping aided (3DMA) GNSS must also be computationally efficient enough to run in real time on a mobile device and, at the same time, be able to handle initialisation errors of the order of 100m. Here, we present the real-time implementation of UCL’s 3DMA GNSS algorithms on Android devices, exploiting the “raw” measurements supplied by the latest GNSS chipsets. We have developed demonstrator apps for Central London and Downtown San Francisco and Downtown Miami. In London and San Francisco, we achieve a root mean square (RMS) horizontal position accuracy of about 6m, compared with about 40m using a conventional GNSS positioning algorithm. The paper will briefly summarize the 3DMA GNSS algorithms, describe the implementation of the Android app, presenting some positioning results using both single-epoch and multi-epoch versions of the 3DMA GNSS approach, and show how accuracy can be traded off against speed. The steps needed to take the 3DMA GNSS app from a research demonstrator to a full service will then be discussed. |
