Under the planar movement assumption, robot motion is described by 3 variables: 2 translations and 1 rotation on the ground plane.

The localization task is performed in two steps:

Original images
Rectified images (from stereo depth information)


Image correlation (by minimizing image difference)





