From manual programming to automated tracking
In traditional laser welding, robots only do exactly what they are told, requiring tedious manual programming for every new part geometry. My role was to give the robot "eyes."
The Solution
Using state-of-the-art computer vision models, I developed a system that processes live camera feed data to recognize complex parts. The system autonomously calculates the optimal welding trajectory along seams, allowing the arm to adapt to the part right in front of it. This shift to autonomous seam detection bridges the gap between rigid automation and the flexibility required for modern manufacturing.
