Image processing with RobotStudio

Alex_Qui

Project Introduction

I'm developing a proof-of-concept for a vision-based quality inspection system for plastic bottles using an ABB GoFa 12 robot arm with a standard webcam (not Cognex or ABB cameras). The system will inspect bottles for contamination, dimensional accuracy (±3mm tolerance), and defects like cracks.

My Question

While ABB Omnicore can detect cameras, I'm struggling with implementing the required image processing pipeline (Gaussian filtering, morphological operations, edge detection, blob analysis) in RobotStudio.

1. Is it possible to implement these complex image processing operations directly in RobotStudio?
2. If not, should I use a separate microcontroller/PC to process images and communicate results to the robot via socket messaging?
3. Has anyone successfully integrated custom vision algorithms with an ABB robot without using proprietary vision systems?

Thanks for your help!