Exploring Tekkotsu Programming on Mobile Robots:

Low-Level Vision

Prev: World state
Up: Contents
Next: Color segmentation

For a variety of reasons, robots don't see very well. Their visual abilities are far inferior to cats or dogs, and probably even rats. Nonetheless, a robot with a built-in video camera is far more interesting than one without.

Tekkotsu provides multiple levels of vision facilities, beginning with access to raw camera images, color segmented images, and color segmented connected components ("blobs"). At the more abstract levels, Tekkotsu's "dual coding representation" provides primitives for manipulating both iconic and symbolic image formats, and interfaces with other components, such as the MapBuilder, which can construct a map of the AIBO's environment.

The following topics are covered in the Low-Level Vision chapter of this tutorial:

  1. Color image segmentation

  2. The vision pipeline

  3. Ball detection events

More advanced topics are covered in the Visual Routines chapter.

Prev: World state
Up: Contents
Next: Color segmentation


Last modified: Mon Nov 28 00:38:07 EST 2005