From my understanding of the alternate solutions available on the market, an Al Model is trained to detect specific types of failures. For instance, if the print dislogged from the bed and is no longer secure or if the print is noticable creating “spaghetti” instead of a sturdy structure. There would be no requirement for Cam Plus to process the Gcode used by the printer to know if the print is legitamately on track, but simply detect an known visual error type that could lead to a saftey concern and need for action.
In Wyze’s recent communications we’ve seen updates about Pet Detection, Facial Recognition and Person Identification notifications. The same concept could essentially be done if given the right data to train an their models for what to look for and then notify the user.