Precise navigation is crucial for the long range unmanned aerial vehicles (UAVs). But today’s UAVs still rely on global navigation satellite system (GNSS) for accurate navigation; this brings in the safety issues as the weak satellite signals can be interfered. To achieve accurate autonomous navigation, we propose to utilize publicly available remote sensing maps to compensate the drifts from the inertial navigation systems. To deal with severe visual variations between the camera images and the map image, we propose a hierarchical outlier filtering strategy to filter out the feature correspondence with a voting algorithm at the first stage, filter out the wrong matched position with the predictions from the inertial navigation system at the second stage. Then an Extended Kalman Filter (EKF) is constructed to combine the matched position with the Inertial Navigation System (INS) to achieve accurate and drift free localization performance. To test the proposed algorithm, we have conducted long range real flight experiments with a CH-4 UAV and the overall localization error is less than 10 m.
Map Aided Visual-Inertial Integrated Navigation for Long Range UAVs
Lect. Notes Electrical Eng.
International Conference on Guidance, Navigation and Control ; 2022 ; Harbin, China August 05, 2022 - August 07, 2022
2023-01-31
10 pages
Article/Chapter (Book)
Electronic Resource
English
Homography-based Vision-aided Inertial Navigation of UAVs in Unknown Environments
British Library Conference Proceedings | 2012
|Fail-safe visual-inertial navigation for UAVs (Invited Paper) [9836-65]
British Library Conference Proceedings | 2016