TY - GEN
T1 - TinyML-Powered Tack Weld Detection for Robotic Welding
AU - Waseem, Hizza
AU - Wu, Di
AU - Coatanéa, Eric
AU - David, Joe
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.
PY - 2025
Y1 - 2025
N2 - The integration of robotics and ML is driving innovation in digital manufacturing. Robotic welding benefits from precise defect detection and monitoring. Tack welds, crucial in pre-welding assembly, can affect final weld quality if improperly formed and unaccounted for while welding over them, making its rapid detection essential. Leveraging TinyML for on-device inference enables immediate tack weld detection using a welding camera, eliminating reliance on high-latency cloud processing. This study integrates TinyML-based tack weld detection on a Renesas EK-RA6M5 platform running micro-ROS. By leveraging convolutional neural networks and the Edge Impulse platform, a model to classify tack weld online is developed. During development, the model reached an estimated F1-score of 98.7%, 10 ms inference time, and minimal resource use (75.2 KB RAM, 78.3 KB Flash). In real-world deployment, an inference time of 80 ms was achieved, with RAM and Flash usage at 330 KB and 230 KB, respectively. Despite higher computational demands, the system maintained respectable accuracy and responsiveness, confirming its viability on-board for edge analytics. This on-device solution reduces latency, enhances autonomy, and provides a foundation for future advancements in autonomous robotic welding.
AB - The integration of robotics and ML is driving innovation in digital manufacturing. Robotic welding benefits from precise defect detection and monitoring. Tack welds, crucial in pre-welding assembly, can affect final weld quality if improperly formed and unaccounted for while welding over them, making its rapid detection essential. Leveraging TinyML for on-device inference enables immediate tack weld detection using a welding camera, eliminating reliance on high-latency cloud processing. This study integrates TinyML-based tack weld detection on a Renesas EK-RA6M5 platform running micro-ROS. By leveraging convolutional neural networks and the Edge Impulse platform, a model to classify tack weld online is developed. During development, the model reached an estimated F1-score of 98.7%, 10 ms inference time, and minimal resource use (75.2 KB RAM, 78.3 KB Flash). In real-world deployment, an inference time of 80 ms was achieved, with RAM and Flash usage at 330 KB and 230 KB, respectively. Despite higher computational demands, the system maintained respectable accuracy and responsiveness, confirming its viability on-board for edge analytics. This on-device solution reduces latency, enhances autonomy, and provides a foundation for future advancements in autonomous robotic welding.
KW - Micro-ROS
KW - Robotic Welding
KW - Tack-Weld Detection
KW - TinyML
KW - Vision-Based Edge Analytics
UR - https://www.scopus.com/pages/publications/105020571065
U2 - 10.1007/978-3-032-05610-8_54
DO - 10.1007/978-3-032-05610-8_54
M3 - Conference contribution
AN - SCOPUS:105020571065
SN - 9783032056092
T3 - Lecture Notes in Mechanical Engineering
SP - 549
EP - 556
BT - Flexible Automation and Intelligent Manufacturing
A2 - Srihari, Krishnaswami
A2 - Khasawneh, Mohammad T.
A2 - Yoon, Sangwon
A2 - Won, Daehan
PB - Springer
T2 - International Conference on Flexible Automation and Intelligent Manufacturing
Y2 - 21 June 2025 through 21 June 2025
ER -