Object Detection and Text Recognition for Immersive Augmented Reality Training in Laser Powder Bed Fusion

Abstract

Professional Training for laser powder bed fusion equipment operation typically is time consuming and expensive for end-users in metal Additive Manufacturing (AM). This study proposes a practical solution by leveraging Hyperskill software to develop a immersive training program compatible with Augmented Reality (AR) devices. The program incorporates the YOLOv7 object detection algorithm and the CRAFT(Character Region Awareness for Text detection) Four-stage text recognition algorithm, seamlessly bridging the gap between reality and simulation. Through a human subjects study, we demonstrate the effectiveness and viability of the AR program, showcasing its capacity to deliver real-time, comprehensive training in metal AM processes. This approach offers a balanced and efficient solution for improving professional training in the field of metal AM.