Unlocking New In-Situ Defect Detection Capabilities in Additive Manufacturing with Machine Learning and a Recoater-Based Imaging Architecture

Additive manufacturing (AM) has the potential to revolutionize the way products are designed and produced in a wide range of industries. However, ensuring the quality and reliability of AM parts remains a challenge, as defects can occur during the building process. In-situ monitoring is a promising approach for detecting and classifying these defects for in-process part qualification. In this paper, we present a novel approach for in-situ monitoring of laser powder bed fusion (LPBF) processes using a recoater-based imaging sensor and machine learning algorithms. The new sensor architecture is a recoater-mounted contact image sensor (CIS) and allows for high-resolution imaging of the build area during the recoating process, enabling the observation of a wide range of part and process-related defects. We demonstrate the effectiveness of using machine learning for image analysis on a series of experiments on a commercial AM system, showing significant improvements in defect detection accuracy compared to existing methods. Our results demonstrate the potential of the recoater-based sensor architecture for unlocking new capabilities for in-situ monitoring and quality control in powder bed-based AM processes.