Alexa F. Siu. click for home.

Image-Based Visual Servoing (Spring 2015)

Alexa F. Siu1, Michael Sobrepera1

Github-MarkGithub Files

Image-based visual servoing (IBVS) was implemented to navigate a Scribbler robot to a desired pose. The target was to dock the robot facing forward and one foot away from a square landmark. The controller used a RANSAC algorithm to identify the four corners of the target landmark. When all four points were successfully identified, the robot's pose was iteratively adjusted based on three different parameters:

  1. The center of the square.
  2. The ratio of the square’s leftmost line over the rightmost line (skew ratio).
  3. The square width.

Experimental setup
Experimental setup.
Scribbler docking
Scribbler at target position.
RANSAC example for IBVS
Scribbler path docking from 45 degree angle
Scribbler path docking from 5 feet away