A GPR-Aided Method for Collaborative Relative Navigation for Multiple Unmanned Aerial Vehicles During Ranging Measurement Outages

To maintain reliable relative navigation accuracy of unmanned aerial vehicles (UAVs) during range measurement outages, a GPR-aided ranging method is proposed to assist the extended Kalman filter for measurement updates. Specifically, when the on-board ranging measurement is available, a Gaussian process regression (GPR) algorithm is used to train the mapping model between acceleration, angular velocity of the inertial navigation system (INS), relative position, and the ranging distance. When ranging measurement is unavailable, the trained GPR algorithm is used for distance prediction, replacing the ranging measurement and being used for consensus extended Kalman filter (CEKF) measurement update to reduce the degree of relative navigation error divergence. Numerical simulation was conducted with three UAVs performing cooperative flight, the results show that the proposed algorithm can provide reliable distance prediction during ranging outage and effectively reduce the divergence of relative state estimation error.