Enhancing Micro Air Vehicle Navigation in Dense Urban Areas using 3D Mapping Aided GNSS

Abstract

The restrictions on operating Micro Air Vehicles (MAVs) with a mass below 200g in an urban environment are greatly reduced compared to heavier Air Vehicles. Operating MAVs directly above streets and close to buildings brings two key challenges compared to larger uninhabited air vehicles (UAVs). Firstly, the weight restriction essentially limits the useable sensors to smartphone technology, such as small cameras, chip-scale inertial measurement units (IMUs) with magnetometers and a barometer, and GNSS receivers using lightweight linearly polarized antennas. The second challenge is that GNSS signals are blocked and reflected by the buildings in dense urban areas, severely degrading the positioning accuracy. By using 3D mapping of the buildings, to predict which signals are visible and which are blocked, GNSS positioning accuracy in can be substantially improved. For land-based applications in dense urban areas, single-epoch positioning is about a factor of five more accurate using 3D-mapping-aided (3DMA) GNSS than conventional GNSS positioning. Here, 3DMA GNSS is extended to airborne use by integrating it with barometric height. Experimental tests at 3 m and 5 m above ground have shown that, where the height above ground is known, the positioning performance and processing load is similar to that of ground-based 3DMA GNSS, with RMS horizontal errors of 4.5 m using an RHCP antenna and 6.9 m using a linearly-polarized antenna. 3DMA GNSS ranging performance improves with increasing height above ground, whereas shadow-matching performance degraded with increasing height at most of the test sites used here. Where the height solution is provided by a calibrated barometric altimeter, the 3DMA GNSS performance depends on the quality of the barometric sensor. Using a smartphone grade barometer, the RMS horizontal position errors are degraded to 8.3 m using an RHCP antenna and 9.9 m using a linearly-polarized antenna. However, using a high-quality barometric altimeter, the 3DMA GNSS position accuracy is only degraded by about 12%, compared to the known-height results. Thus, selection of a good height sensor is critical. Using a two-stage 3DMA GNSS algorithm implementing a 3D search area at the second stage, the RMS positioning accuracy obtained with a low-quality barometer is improved to 5.7 m using an RHCP antenna and to 8.3 m using a linearly-polarized antenna.