Smart Defect Analysis and Remediation for Product Quality Assurance (Smart-DARPQuA)

Many enterprises — and particularly small and mid-size manufacturers (SMMs) — rely on inspection methods (such as cameras and other sensing technology) that produce digital data but still require manual evaluations to classify defects and determine remediation steps. While manual inspections can be useful to catch defects and errors in manufacturing processes, they have limitations that can negatively affect the quality, efficiency, throughput, and cost-effectiveness of production.

The team will develop Smart-DARPQuA to improve the classification and remediation of defects using artificial intelligence (AI) and digital inspection techniques in manufacturing processes. It will consist of a multi-camera image acquisition subsystem, an image processing subsystem, and a recommender subsystem with feedback. Novel image-processing algorithms will perform defect detection, classification, localization, and quantization on different planar surfaces (e.g., cutting saw tooth image and display surface image). A knowledge-based remediation module will translate the know-how of skilled inspectors and the relationship among different pieces of the setup as well as identify defect examples and options to fix them. This will enable proactive or reactive remediation actions — catching observed defects and avoiding them in the future. Finally, digital thread integration will tie all inspection data to the target object geometry through a digital twin that will provide insights into the inspection process through a user interface. Smart-DARPQuA’s integration of various technical elements (defect analysis, remediation recommendation, digital thread integration, edge deployment) will produce a robust Technology Readiness Level 7 solution.

While Smart-DARPQuA will be deployed at the MK Morse facility to enhance the quality inspection process for high-quality saw blade welding, its core technology will be extended. It will be used to develop another proof-of-concept for the eWriter display surface quality inspection at Kent Displays, thereby showcasing how Smart-DARPQuA can be used in multiple industries. Effectiveness of SMART-DARPQuA will be evaluated through extensive ROI statistics collection at both facilities.

The defense industrial base consists largely of SMMs that are less digitally mature than larger manufacturers. Their adoption of digital manufacturing technologies such as SMART-DARPQuA is critical to the Department of Defense supply chain digital transformation.

Currently there is no standard available for the use of AI-based inspection. Existing manual inspection standards and interpretation of inspection are available. But these standards do not include guidelines or specifications for optical or image-based AI-based inspection. New guidelines could be developed as recommendations for adoption of the proposed technology.