Intelligent fastening assistant system (IFAS): deep learning-driven angle and sequence supervision for manual assembly

Manual screw fastening is essential in assembly and manufacturing processes. Using improper tool angles during fastening can compromise product quality and lower production efficiency. This study introduces the Intelligent Fastening Assistant System (IFAS), an innovative quality control solution for manufacturing. IFAS integrates a supervised machine learning-based object detection model with a novel spatial object angle calculation model. The custom object detection model achieved a high 96.23% mean average precision, establishing a robust algorithmic foundation. System validation involved 12 batches, each comprising ten operations with four screw fastening actions. The overall accuracy rate was 92.0%, with a precision of 94.3%, a recall rate of 94.4%, and an F1 score of 94.0%. These statistical results confirm IFAS’s effective real-time supervision of fastening operations. The proposed IFAS provides real-time visual and auditory alerts to operators, preventing defects by identifying deviations in fastening tool angles or sequences, while also logging non-compliant actions for proactive quality management in smart factory applications.