Automatic inspection of finished goods or parts
One interesting use case in the discrete manufacturing space is the automated inspection of finished goods or parts. Finished goods are inspected by analyzing video feeds from digital cameras; by deploying the required machine learning (visual inspection) models on the Device Gateway (DG), these models output a binary value indicating whether a part is defective or not.
Important note
The term sensing device is not limited to video feeds from digital cameras; alternate sensing technologies (such as infrared cameras and x-ray cameras) may very well be used to support specific operating conditions and use case requirements.
If a defective part is detected, a robotic arm will take that part from the conveyor belt and put it in a waste bin. These models are trained at a central server and refined models are pushed to a gateway from the central server. If the model is not able to identify the part with the required level of confidence...
