Real-Time Seam Tracking for Robotic Welding Via Registration-Based Deformation Estimation
Surag Balajepalli, Amrish Baskaran, Naik Shounak, Christopher Eubel, Abolfazl Meyarian, Andong Dai, Pradeep Rajendran, Katsu Yamane, Trazkovich Alex
AI summary
Problem
Thermal deformation during arc welding dynamically displaces the seam path, causing robots to miss joints on long seams. Standard laser-line detection also fails under intense arc glow, while perception delays complicate real-time correction.
Approach
The system registers live laser scans against a pre-weld point cloud using constrained ICP with spatial propagation, retrained a deep learning network to detect laser lines under arc-on conditions, and employs an asynchronous multi-threaded architecture to deliver smooth, high-frequency robot commands despite perception delays.
Key results
- Laser-line detection F1 improved from 0.59 to 0.84 on arc-on images
- Weld bead stayed in joint for full 97 cm length with <2 sec cycle-time overhead
- Field deployment increased weld quality acceptance rate from 81% to 95–98%
- Seam placement error reduced to ≤1.0 mm in field tests
Why it matters
Enables reliable, high-quality robotic welding on long seams without costly rework, directly improving industrial manufacturing efficiency and automation reliability.
Abstract
Arc welding induces thermal deformation that continuously displaces the seam path during execution, causing the robot to miss the joint on long seams. We present a real-time seam tracking system with three principal contributions: (1) a constrained ICP registration of live leading-laser scans against a prescan point-cloud prior combined with exponentially decaying spatial propagation; (2) a laser-line detection network retrained on 1,000 arc-on images, raising F1 from 0.59 (prescan baseline) to 0.84 on a held-out arc-on test set; and (3) an asynchronous execution architecture for ensuring that smooth joint commands are sent at the robot’s control cycle (40 Hz) even with perception delays or interruptions. Internal testing confirms the system remains in the joint on 97 cm-long seams with less than 2 sec cycle-time overhead. Field deployment improved weld quality acceptance rate from 81% to 95–98%.