3D tracking and control of UAV Using planar faces and monocular camera
Chapter in Scopus
- Additional Document Info
- View All
A method for tracking the 3D pose and controlling an unmanned aerial vehicle (UAV) is presented. Planar faces of target vehicle are tracked using the Efficient Second Order Minimization algorithm, one at a time. Homography decomposition is used to recover the 3D pose of the textured planar face that is being tracked. Then, a cuboid model is used to estimate the homographies of the remaining faces. This allows switching faces as the object moves and rotates. Cascade and single PID controllers are used to control the vehicle pose. Results confirm that this approach is effective for real-time aerial vehicle control using only one camera. This is a step towards an automatic 3D pose tracking system. © 2013 Springer-Verlag Berlin Heidelberg.