Multimodal sensor fusion and machine learning for quality prediction in directed energy deposition

Abstract: Ensuring quality consistency remains a significant challenge in laser-directed energy deposition (LDED) additive manufacturing (AM), with defects such as keyhole pores and cracks leading to build failures. While recent advancements in multi-sensor fusion techniques have demonstrated high accuracy in localized defect detection, they remain reactive rather than predictive. In this talk, I will introduce a multimodal fusion approach for localized quality prediction in the robotic LDED process. The data used in multimodal fusion includes features extracted from a coaxial melt pool vision camera, a microphone, and an off-axis short wavelength infrared thermal camera. Optical microscope (OM) images of printed parts' cross-sections are used to locate defect-free and defective regions (i.e., cracks and keyhole pores), which serve as ground truth labels for training supervised machine learning (ML) models for quality prediction. The trained ML model is then used to generate a virtual quality map that registers quality prediction outcomes within the 3D volume of the printed part, thus eliminating the need of physical inspections by destructive methods. Experiments show that the virtual quality map closely match the actual quality observed by OM. Compared to traditional single-sensor-based quality prediction, the proposed approach has achieved a significantly higher quality prediction accuracy (96%), a higher ROC-AUC score (99%), and a lower false alarm rate (4.4%). As a result, the proposed approach is a more reliable method for defect prediction. The key novelty of this research is a spatiotemporal data fusion method that synchronizes multimodal features with the real-time robot motion data to achieve localized quality prediction. Future research directions and limitations in AM technologies with ML will be discussed

Biography: Dr. Moon is currently an associate professor (tenured) and assistant chair (research) in School of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore. He is a Director of HYUNDAI-NTU-A*STAR Corporate Lab. He received his Ph.D. degree in Industrial Engineering from the Pennsylvania State University, USA, in 2008, his M.S. and B.S. degrees in Industrial Engineering from Hanyang University, South Korea, in 1995 and 1992, respectively. He is interested in the boundary-spanning research that integrates the knowledge of design, engineering, and economics. His current focuses include applying sciences, economic theory, and AI to the design of customized and sustainable products, services and systems, strategic and multidisciplinary design optimization, advanced modeling and simulation, design for additive manufacturing/3D printing, embedded sensor design for 3D Printing, digital twins, smart factory and digital manufacturing for supply chains.

Zoom link: https://purdue-edu.zoom.us/j/92766887904