Compensating UAV GPS data accuracy through use of relative positioning and GPS data of UGV

We improve the performance of a low-quality GPS data on a UGV through use of multiple low-quality GPS modules on a UGV within line of sight of the UAV. The UGV sends GPS data corrections to the UAV on the basis of the distance from the UAV to the UGV as measured by scaling of a standard image pattern stuck on the UGV. Geolocation of both UGV and UAV are performed through the use of the extended Kalman filter integrating GPS aided INS. The positioning error is reduced by a factor of 2.3 in simulation studies and a factor of 1.6 in experiment when 3 GPS sensors are used on the UGV. This is better than what one can get through pure averaging of the GPS sensors in the presence of noise in measuring the UAV-UGV distance. We show how our exploitation of geometry improves GPS sensor performance as more GPS sensors are used.