A monocular vision-based attitude estimation approach for small Unmanned Aerial Vehicles and its experimental verification

To improve the sensor information redundancy of small Unmanned Aerial Vehicles(UAVs), an image-based attitude estimation approach is proposed. Based on aerial image segmentation and between-class variance analysis, a horizon extraction strategy is put forward, and then an attitude estimation model is established referring to the knowledge of camera projective geometry. To verify the effectiveness of the proposed algorithm, flight tests are carried out on a small fixed-wing UAV Cessna172-140 with Go - ProHero1 camera on board. It is shown that the proposed algorithm can give an accurate estimation of the attitude, which will provide Inertial Navigation System (INS) of small Unmanned Aerial Vehicles (UAV) with valid redundant information.