ProductSciencePlatformCompanyProjectsCareers
Sign inBook pilot

The intelligencelayer forrobotic welding.

A year of qualification. Now in days.

Book a pilot
280×faster qualification
≥50%reduction in price per weld meter
≈2 tCO₂e avoided per qualification
90%prediction accuracy
4validated pilots
280×faster qualification
≥50%reduction in price per weld meter
≈2 tCO₂e avoided per qualification
90%prediction accuracy
4validated pilots

Line illustration of a robotic welding cell, first as an exploded view of its components: a six-axis collaborative robot arm, a MIG/MAG welding torch, a welding power source, and a T-joint workpiece on a fixture table. Scrolling steps through seven engineering questions, assembles the parts into one working cell connected to RobTrack, and the robot welds the seam.

Intelligence, integrated.

We don't build robots or power sources. We engineer how they work together, so RobTrack's process intelligence runs on real equipment.

Fig. 1 — Illustrative robotic MIG/MAG cell. Generic equipment, drawn for explanation.Into the real cell

The real cell.

Our robotic welding setup, recorded in operation: robots, torches, fixturing, and fume extraction working as one cell.

3D-Components robotic welding setup, recorded in operation.

Trial and error breaks at scale.

Every new robot, material, geometry, and machine creates a new process space. Conventional parameter development burns expert time, machine hours, and material before the real production work can begin.

0+
coupled process variables

Too many variables

Heat input, travel speed, wire feed, layer height, shielding, cooling, geometry, material response, and many more variables all interact.

Millions
of possible parameter combinations to search
Multi-day
to qualify one setting: test weld + lab cycle1
0–0%
of revenue lost to poor quality2

Too much waste

Each bad run consumes machine capacity, material, inspection effort, project schedule, and scarce welding expertise.

€0k–€0k
to cut out and re-weld a single bad joint
€0k–€0k/day
delay penalties can stack on top
0
change voids the whole qualification3

Too little transfer

A setup proven on one machine, material, or cell rarely transfers cleanly. Change one essential variable beyond its approved range and the procedure must be requalified.

Vendor lock-in
arc modes like CMT do not transfer across power sources3
Days – months
destructive + NDT testing loop restarts each time1
Why this matters now
0
Average welder age4
0 k
new welders are needed only in the U.S.: only 2 enter for every 5 who retire4
Group 1
Welding fumes classed a known human carcinogen by IARC5
0M
Full-time welders exposed to welding fumes worldwide6
Sources1 Procedure qualification turnaround — Norman QC / WH Labs2 Cost of poor quality & rework — Therness3 LRQA, "What's new with EN ISO 15614-1"4 Welder age & 2028 workforce gap — Novarc / AWS5 Welding fume Group 1 carcinogen — IARC / NIH6 11M full-time welders exposed — IARC / AIHA

RobTrack turns welding into an optimization problem.

RobTrack combines process data, sensor data, materials knowledge, and physics-based simulation to converge on the parameter region most likely to produce stable, repeatable, high-quality results.

robtrack.process-window.ts
1setup = {2  robot: "industrial arm",3  process: "robotic welding",4  material: "steel alloy",5  geometry: "production part"6}
Optimization engine ready
Process contour mapTravel speed × wire feed × heat input
  1. 01Recommended parameter set
  2. 02Process-window boundaries
  3. 03Stability indicators
  4. 04Expected bead geometry
  5. 05Validation report
  6. 06Deployment guidance

Built for the equipment you already own.

RobTrack adapts to the robot, machine, CAD, simulation, and process tools already present in the cell.

KUKA
Robot OEM
ABB
Robot OEM
FANUC
Robot OEM
Yaskawa Motoman
Robot OEM
Universal Robots
Collaborative robot
Comau
Robot OEM
Fronius
Welding source
ESAB
Welding source
Lincoln Electric
Welding source
Miller
Welding source
TRUMPF
Metal AM machine
KUKA
Robot OEM
ABB
Robot OEM
FANUC
Robot OEM
Yaskawa Motoman
Robot OEM
Universal Robots
Collaborative robot
Comau
Robot OEM
Fronius
Welding source
ESAB
Welding source
Lincoln Electric
Welding source
Miller
Welding source
TRUMPF
Metal AM machine
01

Any robot

Industrial arms, cobots, and mobile robotic cells.

02

Any machine

Welding systems, DED, WAAM, LMD, and metal AM setups.

03

Any material

Steels, aluminium, high-performance alloys, and emerging compositions.

04

Any process

Welding, repair, deposition, qualification, and AM workflows.

Siemens NX
CAD / CAM
Simufact Welding
FEM welding simulation
Ansys Mechanical
FEM simulation
Abaqus
FEM simulation
SolidWorks
CAD
Autodesk Fusion
CAD / CAM
PTC Creo
CAD
Altair HyperWorks
Simulation
RoboDK
Robot programming
ABB RobotStudio
Offline programming
KUKA.Sim
Robot simulation
Process Simulate
Manufacturing simulation
Siemens NX
CAD / CAM
Simufact Welding
FEM welding simulation
Ansys Mechanical
FEM simulation
Abaqus
FEM simulation
SolidWorks
CAD
Autodesk Fusion
CAD / CAM
PTC Creo
CAD
Altair HyperWorks
Simulation
RoboDK
Robot programming
ABB RobotStudio
Offline programming
KUKA.Sim
Robot simulation
Process Simulate
Manufacturing simulation
What people say

"Our project connects real telemetry and near-real-time parameter control around RobTrack."

J

Jayant Singh

Technology Manager - Robotics & AI, Mechatronics Innovation Lab

Key Result

Deployed, user-validated project workflow

Built by domain experts in materials science, robotics, and AI.

RobTrack is built by 3D-Components, a deep-tech team from Oslo combining materials science, robotic manufacturing, simulation, and AI software.

Leadership
Amin S. Azar

Amin S. Azar

Founder & CEO

Daniel Haas

Daniel Haas

AI Engineer & CTO

Virgilio Gomez

Virgilio Gomez

Lead Robotics Engineer

Partners & Collaborations

Bring your setup.
RobTrack finds the process window.

Tell us about your challenges and current bottleneck. Together, we will assess how RobTrack can deliver a faster path to validated welding or metal AM parameters.

Book a pilot