Visual Inspection for soldering & gluing

Aim of the project is to reduce improve production quality and promote LEAN by reducing repetitive and redundant manual work as well as remove the human error aspect of inspection of solder quality in manufacturing. Main benefits are making production more efficient by reducing the amount of time workers need to spend manually inspecting solders and improving the overall quality of solders. A Cognex 902M camera was used to inspect solder quality by using a convolutional neural network. Cognex Insight Suite software was used to build a Neural network in semi supervised mode and algorithm was taught to identify faults. Main benefits are making production more efficient by reducing the amount of time workers need to spend manually inspecting solders. The overall quality of solders was improved with high faulty solder detection with the Cognex system setup.

Links:
Partners: GE Healthcare Finland Oy, Roboco Oy, AP Vision Oy
Keywords: Machine vision, AI
Contact: Markku Hentula /VTT
Robotics Fusion