This master thesis encompasses the transformation of the bending process device into a
distributen computer vision system for the bending of thin metal sheets. The device was
completed as part of a different master thesis, however it was only the construction part,
without any governing logic behind it. As part of this work, the device was connected with
the microcontroller Arduino Mega and the microcomputer Raspberry Pi into a flexible metal
sheet bending system. The system offers flexible geometry in the form of an independently
moving punch and supports. The system was also upgraded to a computer vision system,
with the help of a camera, to track the current bending angle of the sheet metal, which allows
in to work as a closed loop system. The bending angle is calculated by using the different
digital image processing alogrithms described in this work, to decompose and process the
images and obtain the bending angle. The user experince is further augmented through the
use of a graphical user interface which enables a intuitive user experince to controle the sheet
bending device. Through preliminary results, a below 1% accuracy of the bending angle
measurements was achieved, which is in line with the application needs. The system also
offers different bending process operations, which remain unresearched.
|
