For workshops that want a welding robot, the sticking point is rarely the robot itself; it is programming it. The robot arrives, a program is written for the first part, everything runs well; then a new part comes and it all starts again. In this article we explain how welding robots are classically programmed, what makes it hard, and how marking with chalk changes the picture.
How is a welding robot classically programmed?
- Teach-in: The operator jogs the robot with the teach pendant to the start, intermediate points and end of the seam and records each point. Then travel speed, current, voltage, torch angle and, if needed, weaving are set.
- Offline programming: The path is prepared on a computer from the part’s 3D model, tested in simulation and loaded to the robot. Programs can be written while the robot keeps working, but this needs a 3D model and software skills.
What makes programming hard?
- Every new part means a new program. In series production a program written once runs on thousands of parts; in small batches and one-offs, programming time can exceed welding time.
- It needs skilled people. Someone who speaks the robot’s language and understands welding is hard to find; without that person the robot often stands idle.
- Parts are never identical. Cutting and bending tolerances, fixture differences and heat distortion create a gap between the taught point and the real seam.
How seam tracking forgives the program
Robots get seam tracking to compensate small part-to-part differences. Touch sensing, through-arc sensing and machine vision correct the programmed path to the real seam. This reduces the need for a millimetre-perfect program on every part, but it does not remove the program. See how does weld seam tracking work? for details.
Marking with chalk: leaving the program to the welder
On DKT’s AI welding robot, the welder, not a program, decides where the weld goes. The welder puts the part on the table and draws chalk marks near the spots to be welded. The robot’s camera sees the part; AI recognises the chalk marks and targets the seam nearest to each mark. The robot moves there and welds, while the camera and laser sensors track the seam.
The robot’s initial setup and welding settings are done during commissioning. After that, a new part does not need reprogramming: the welder simply marks it with chalk again.
Which method for which job?
| Production type | Suitable method |
|---|---|
| Thousands of identical parts | Classic or offline program; written once, used many times. |
| Small batches, frequently changing parts | Chalk marking; no program to write for each part. |
| Long straight seams on tanks, silos, pipes | Welding carriage; a rail and seam tracking replace the program. |
Summary
- Classic welding robots are programmed by teach-in or offline; each new part needs a new program.
- In small batches, programming time and the need for specialists can leave the robot underused.
- Seam tracking corrects small differences but does not replace the program.
- With chalk marking the welder decides where to weld and AI recognises the mark; no programming for new parts.
Send a photo of your part on WhatsApp to see whether it suits chalk marking. For cost items see our welding robot price page.