For conventional robotic welding, repetition is an advantage.

Once a program has been created and verified, the robot can repeat the same welding path efficiently across a large production batch.

But many fabrication environments do not operate that way.

In high-mix, low-volume production, workpiece types and welding requirements may change frequently between production batches.

In this type of high-mix, low-volume production, the challenge is not simply whether welding can be automated.

The more practical question is:

How much programming and setup work is required every time the workpiece changes?

This is where teach-free robotic welding offers another approach.

Instead of relying primarily on manual point-by-point teaching, the welding program can be generated from a 3D model or reconstructed workpiece data, while laser seam tracking helps compensate for deviations between the programmed path and the actual workpiece.

For manufacturers dealing with frequent product changes, this changes the role of robotic welding from a fixed repetitive process into a more adaptable production tool.

Why Traditional Robot Teaching Becomes a Bottleneck

Traditional robotic welding often requires an operator to define welding points, positions and trajectories for a specific workpiece.

For stable, repetitive production, this can work effectively.

The difficulty becomes more obvious when production involves:

·       small batches;     

·       multiple workpiece types;

·       frequently changing weld paths;

·       large or structurally complex parts.

In the project material provided by Xunchi, the target application specifically involved small-batch, multi-variety components, where conventional manual programming was considered relatively inefficient. The proposed solution therefore combined offline trajectory generation with laser seam tracking and robotic position compensation.

The objective is not to eliminate engineering preparation altogether.

Instead, the idea is to move a significant part of trajectory creation away from manual robot teaching and into a digital programming workflow.

1. Start with the Workpiece, Not the Robot

A teach-free welding workflow begins with information about the workpiece.

Xunchi's current system supports two workflow modes:

Model-based programming and model-free programming.

In the first workflow, an existing 3D model of the workpiece is imported into the offline programming environment.

The software uses geometry from the model to generate welding trajectory points, robot poses and a robot program that can subsequently be executed by the robot.

The technical documentation specifically describes selecting geometric information such as points, lines and surfaces from the imported 3D model to generate the welding trajectory and robot posture. It also generates information used for subsequent positioning, including point-cloud model and user-coordinate-system data.

This approach is especially relevant when the manufacturer already has usable CAD data for the component.

But CAD data is not always available.

For that reason, the documented workflow also includes a second route based on scanning the actual workpiece and generating a model through reverse modeling before weld seams and robot trajectories are generated.

That gives the system two possible starting points:

Digital model → welding path or Physical workpiece → reconstructed model → welding path

This gives manufacturers a practical workflow for projects both with and without an existing 3D model.

2. Offline Programming Moves Path Creation Away from the Robot

Once the workpiece geometry is available, the offline programming system is used to generate the required welding trajectory.

According to the current Xunchi workstation configuration, the system uses CypWeld offline simulation software to generate workpiece welding trajectory points, robot poses and robot trajectory programs.

The main practical advantage is straightforward:

This reduces reliance on manual point-by-point robot teaching when creating welding paths.

The workstation documentation specifically states that the offline simulation system is intended to reduce program editing time and reduce the level of specialized operator skill required for programming.

There is another important detail for repeat production.

Once a program has already been generated for the same part, the documented workflow states that the software generation process does not need to be repeated before returning to the welding step.

For previously programmed parts, the documented workflow allows the system to return directly to the welding step without regenerating the program.

3. A Digital Path Is Not the Same as the Real Workpiece

Generating a robot trajectory from a digital model solves only one part of the problem.

Real fabricated parts are rarely identical to an ideal CAD model.

In actual production, the workpiece position can deviate from the programmed reference. The documented solution therefore scans workpiece deviations and guides the robot to compensate for positional differences.

This is why the teach-free solution combines offline programming with a laser seam tracking system.

The workstation configuration includes a BCW600P laser seam tracking system, which is used for laser positioning, real-time tracking and robot trajectory compensation.

In the documented solution, the tracking system scans deviations in the actual workpiece and guides the robot to compensate for positional differences and during the welding process.

In other words:

Offline programming defines where the weld should be.

Seam tracking helps the robot respond to where the weld actually is.

That distinction is central to making model-based automation practical in real fabrication environments.

4. Why the Workstation Uses Eight Axes

Programming flexibility alone does not solve another common limitation: robot reach.

Large structural parts may contain welds distributed across long or wide working areas.

For this reason, Xunchi's intelligent cantilever welding workstation combines a conventional 6-axis robot with two additional external motion axes in the X and Y directions, creating a maximum of eight working axes.

The purpose of these external axes is not simply to make the system more complex.

They extend the robot's reachable area so that the workstation can accommodate a wider range of component sizes and configurations.

The documented workstation includes a cantilever travelling structure and a heavy-duty floor rail. The rail is specified at 14 m total length with 12 m effective travel, while the configured QJ2000H robot has a 2010 mm reach and ±0.05 mm repeat positioning accuracy.

This architecture is aimed specifically at industrial steel structures, shell-type structural components and other complex fabricated assemblies.

5. One Workstation, Different Workpiece Types

High-mix manufacturing only benefits from automation when the system can accommodate more than one narrowly defined component.

The current workstation documentation groups suitable workpieces into three general categories.

Type A: Strip-shaped parts

Typical joints include T-joints and fillet welds.

Type B: Ribbed parts

These can involve T-joints, flat fillet welds and circumferential fillet welds.

Type C: Combination parts

These include more complex assembled structural components.

This distinction is useful because it illustrates what flexibility means in engineering terms.

It is not simply that the robot can move in many directions.

The system needs to coordinate:

workpiece geometry + trajectory generation + robot reach + seam detection + welding process parameters.

Only when those elements work together does robotic welding become useful for changing production tasks.

6. Teach-Free Does Not Mean Process-Free

There is an important distinction here.

“Teach-free” should not be interpreted as “no engineering required.”

The welding process still requires appropriate welding parameters, workpiece preparation, system configuration and process verification.

The workstation therefore includes more than a robot and welding power source.

Its documented configuration includes:

·       the robot and robot controller;

·       X- and Y-direction external axes;

·        welding system;

·        water-cooled robotic welding torch;

·        BCW600P laser seam tracking system;

·        CypWeld offline simulation software;

·        welding process database software;

·        travelling cantilever;

·        heavy-duty ground rail;

·        torch cleaning and wire-cutting system;

·        electrical and control components. 

Xunchi's technical material also describes a welding process database intended to provide welding parameters for different materials and welding orientations, with the broader goal of reducing dependence on operator experience and supporting more standardized welding operation.

So the value of teach-free welding does not come from one individual component.

It comes from connecting digital programming, sensing, robot motion and welding process control into one workflow.

From Repetition to Adaptability

Traditional robotic welding is strongest when the product stays the same.

Flexible robotic welding becomes more valuable when the product does not.

For high-mix, low-volume manufacturing, the goal is therefore not simply to make the robot weld faster. It is to reduce the amount of work required to prepare the robot for the next different workpiece.

A workflow built around offline trajectory generation, laser seam tracking, robot compensation and extended-axis motion provides one possible approach.

For manufacturers processing structural parts, ribbed components and other changing welded assemblies, robotic automation can therefore be evaluated not only by cycle time, but also by another increasingly important question:

How quickly can the system adapt when production changes?

At Xunchi, our approach to robotic welding starts with the application, the workpiece and the production workflow before defining the automation configuration.

For high-mix welding projects, the right solution is rarely just a robot.

It is the engineering around it.

Have a welding application that is difficult to automate?

Share your workpiece drawing, 3D model, joint type, material and production requirements with our team. We can evaluate the application and discuss a suitable robotic welding configuration based on your actual production needs.

Explore how teach-free robotic welding could fit your production workflow.