This thesis presents a new approach of indirectly tracking an static AprilTag with a monocular camera. The method developed, yields a metric pose estimation of the marker even when it leaves the cameras field of view. This was accomplished by using a keyframebased direct odometry that was enriched by a spherical motion stereo algorithm in order to perform a 3D-to-2D tracking. The proper scaling of the transformation is achieved by a fusion of the depth maps from the marker and the motion stereo. Furthermore an error model was developed, which yields an uncertainty estimation for every pose, to allow the merging of pose estimations from other sources. The work was done with consideration of an application to support a landing approach of a micro aerial vehicle in a real scenario. The results that were obtained during this work are promising to further develop this approach.
Pose Estimation by Detection and Tracking of Artificial Markers for Planetory Exploration
2020-06-30
Miscellaneous
Electronic Resource
English
ArUco markers pose estimation in UAV landing aid system
IEEE | 2019
|Multiple people tracking and pose estimation with occlusion estimation
British Library Online Contents | 2012
|Model-free non-rigid head pose tracking by joint shape and pose estimation
British Library Online Contents | 2016
|Vehicle pose estimation for vehicle detection and tracking based on road direction
BASE | 2017
|