Chayatat Ratanasawanya Min He April 6, 2010
Recall previous presentation The goal Progress report ◦ Image processing ◦ depth estimation ◦ Camera placement Obstacles ◦ Combine image processing and control Simulink models Idea for the next step Questions/Comments
4 visual-servoing structures ◦ Dynamic position-based look-and-move ◦ Dynamic image-based look-and-move ◦ Position-based visual servoing (PBVS) ◦ Image-based visual servoing (IBVS) Implemented a simulation of the dynamic position-based look-and-move system. Implemented a Simulink model to locate the centroid of a ping-pong ball in image.
Implement the system using PBVS, IBVS, or both techniques. Tasks to tackle: ◦ Image processing ◦ Depth estimation ◦ Camera placement on the helicopter model ◦ Combine image processing and control Simulink models. ◦ Jacobian matrix derivation
Use the diameter of the ball in image to estimate the depth Depth, D Focal length F=538 pixel Center of projection Actual ball diameter d b =40mm Ball diameter on image, d img
area dm real dis Cal dis
Combining our image processing model and the control model has been a challenging task. Global variable
exex eyey Increment in pitch and yaw angles LQR controller
Questions/comments are welcome