Additive Manufacturing (AM) fabricates three-dimensional objects by continuously adding material layer by layer from a computer-aided design (CAD) model. In recent years, unprecedented level of autonomy in AM has been enabled by Cyber-Physical System (CPS). AM's machines can have constant interactions with each other through the Internet or other computer networks.

However, AM's security concerns are outgrowing due to the sheer volume and pervasiveness of data and increased accessibility in the networks. Particularly—Cyber-Physical Attack that is originated from the cyber domain but can impose physical damage—became a serious threat to the AM operation. Since the infill structure of the 3D model is usually generated during the conversion process of a CAD file to G-code by the third-party program, the process can be vulnerable against such attacks. Also, it is hard to detect the attack on infill structure because interior defects can be occurred without affecting the exterior. Furthermore, any attackers from inside or outside of the system can exploit this attack to perpetrate their malicious action through various networks. Especially, insiders—who are in a security parameter of a system—can inflict more significant threats to the systems than outsiders because insiders have trusted authentication and knowledge of the systems. According to the IBM Security 2019 report, the manufacturing sector experienced a substantial amount of attacks that were committed by insiders. From the 2019 RSA conference survey, the manufacturing industry was found to be the most vulnerable industry against insider threats.

To detect the infill defectives in a timely manner, layer by layer image inspection in real-time can be conducted. Although researchers have developed various defective detection systems by using image classification, such attempts to AM have been few. Also, it is challenging to collect the layered images from the top view in real-time, because 3D printer's extruder always blocks printing objects from being perfectly captured. Using a dummy model to move the extruder out of the object's layer has been proposed, but it is not practical because it will yield redundant printing delay and waste unnecessary printing materials.

To enable infill layered image collection in real-time without redundant delays and materials, this paper proposes a new layered image collection method by using an algorithm identifying a pseudo area in a layered image. The algorithm detects the pseudo area—for example, the area covered by the extruder—by using many image processing techniques, such as an average pooling, max pooling, and rectified linear unit, and accumulates a non-pseudo area until one complete layered image is acquired. To validate and evaluate the proposed method, two CAD models, which have identical exterior shapes but the different interior shapes, were designed to conduct experiments. Printing processes were recorded using a 4K resolution camera to capture the layered images in real-time. Finally, various classification algorithms for infill layered images were reviewed and tested for the result and conclusion.


Layered Image Collection for Real-Time Defective Inspection in Additive Manufacturing

Division / Sponsors: 
Aerospace, Bioengineering, Applied Mechanics, Design Engineering Division, Advanced Energy Systems, Manufacturing Engineering Division, Noise Control and Acoustics, Nondestructive Evaluation, Diagnosis, and Prognosis (NDPD), Nuclear Engineering Division, Process Industries Division, Fluids Engineering, Heat Transfer, Materials, Microelectromechanical Systems, Safety Engineering and Risk Analysis, Computers & Information in Engineering


Paper Proceedings: https://asmedigitalcollection.asme.org/IMECE/IMECE2020/volume/84478

IMECE2020 - Virtual Conference

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.