Automatic quality inspection using computer vision is an important tool to improve productivity and to ensure 100 % quality checks in manufacturing. For most computer vision methods, ambient light creates challenges because the appearance of the objects changes significantly when the lighting is different. Usually, this is overcome by blocking the ambient light as much as possible.
In our Industry 4.0 lab LISMS, we have a mobile quality inspection system for manufacturing consisting of a gantry to position the camera in 3D and a turntable for the object in front of it. This allows a 360-degree inspection of the quality of relevant features of the object. The computer vision approach to decide whether the quality of each feature, like a stud, is good is done by the method called template matching, which compares reference images with the actual image of the feature. It works, but the technique is sensitive to ambient light changes.
This project investigates how AI and the simulation of a range of ambient light situations can help improve the quality inspection's robustness.
Undergraduate
Expected outcomes:
None
Manufacturing Systems (405.870, Lab)