Researchers at the University of North Dakota offer a solution to this problem. In a February, 2015 article in the journal Machines entitled “Initial Work on the Characterisation of Additive Manufacturing (3D Printing) Using Software Image Analysis,” UND researcher Jeremy Straub demonstrates how model-based assessment can be used to detect issues with 3D printed parts.
While this work was initially performed to detect defects due to material issues, equipment malfunction and happenstance, it is also applicable to preventing malicious attacks, as well.
“An independent detection system, using a model of the expected output as a baseline, would be able to identify defects created by maloperation as well as maliciously introduced ones,” commented Straub.
“The level of separation that is practically required will depend on the severity of the impact of a defect, the likelihood of attack and what other countermeasures are in place to prevent or mitigate such an attack.”
The system proposed would still be reliant on the resolution of the sensors, a key concern raised by the NYU researchers. However, the technology can be used with any relevant and position-correlated pixel-based sensing technology, meaning that even microscope-detail-level imagery could be used to assess an object, if dictated for a given application.
Object position issues and other larger changes to the expected printing results could also be easily detected. These would, of course, require a user to select the desired printing orientation and would not attempt to detect an orientation selection mistake made due to human error.
A UND-based team is working on commercialising this technology, at present, with support from a North Dakota Department of Commerce Venture Grant. This work is part of ongoing 3D printing research at the university.
