/robot-vision-pick-and-place

RoboDK simulation with robot and vision (2d) pick and place

Primary LanguageJupyter NotebookGNU General Public License v2.0GPL-2.0

robot-vision-pick-and-place

RoboDK simulation with robot and vision (2d) pick and place for AAU (Aalborg Univserity) Robot Vision Course. https://moduler.aau.dk/course/2020-2021/ESNVGISK2K2?lang=en-GB

Vision To-Do

  • Color Segementation
  • Detection
  • Jupyter Visu
  • 3D Calibration
  • Connect with RoboDK Camera
  • Move Robot
  • Build Simpsons
  • Plug Everything together

Image Processing

Camera Raw Image

Logo

Scale Down Input

Logo

Add Gaussian Blur

Logo

Convert to Grayscale

Logo

Inverted Binary Thresholding

Logo

Trace Contours, Add Discrete Colors to each Contour Object, Estimate Polygon, Use Polygon Corners to Calculate Orientation, Use Homography Transform to get World Coordinates from Pixel Coordinates

Logo

Full Visu

Logo

Initial Calibration is being done by using 4 additional red bricks with knows world locations.

Logo

Logo

Program in Action

https://user-images.githubusercontent.com/42101377/117690134-cd388980-b1ba-11eb-881e-b1d171341d41.mp4 https://www.youtube.com/watch?v=betZsiYEAE0