Intelligent Control Lab. Cenaturo's image aquisition system
The image aquired by Centauro's camera has two distinct areas:
The top part shows the ground just in front of the image (it is reflected on the mirror).
The bottom part is what the camera sees under the mirror.
This part of the image is further divided into the far part of the track (on the top of the bottom half) and the near part of the track (on the bottom of the image).
Only a small number of lines (5-8) are processed in each of the three subimages,
to reduce computational load. The line processor outputs one of four possible
symbols: all_white, all_black, line or error. Lines whose slope is too small are
rejected as noise. Based on the sequential analysis of
the lines in the mirror subimage of the current frame, a state machine determines
the state of the robot: following_a_line, on_a_line_interruption or on_a_´final T´.
The information from the other subimages (whether a straight line or an interruption
is detected there) is used to accelerate/decelerate the vehicle and to make the
decision of going to the external portion of the track.