Solution — Digital twin
Commission the line twice. The first time costs nothing to fix.
Physics-accurate factory and robot twins in NVIDIA Omniverse and Isaac Sim — used for virtual commissioning, cycle-time proof, operator training and synthetic data generation.
Approach
A twin is only useful if something is at stake in it.
A rendering is not a digital twin. Ours run the actual robot programs against an emulated controller, with real I/O timing, real interlocks and injected faults. When the virtual line jams, the same logic error would have jammed the real one — three weeks later, in front of your customer.
We build in OpenUSD so the twin composes: mechanical CAD, robot kinematics, conveyor logic and the control emulation live as layers that different teams can edit without overwriting each other. That is what keeps a twin alive after the project ends instead of becoming a stale demo file.
For vision projects, the same scene generates labelled synthetic imagery — defects at controlled scale, position and lighting — which is often the only practical way to train a model for a defect that occurs twice a month.
Scope
What is included.
Delivered as a defined scope with acceptance criteria, not as a time-and-materials estimate that drifts.
- 01Factory & line simulationthroughput, buffers, bottlenecks
- 02Robot simulationreach, collision, cycle time
- 03Virtual commissioningPLC-in-the-loop, fault injection
- 04Controller emulationPLCSIM Advanced, Emulate3D
- 05OpenUSD scene assemblylayered, multi-team editable
- 06Synthetic data generationdomain-randomised training sets
- 07Operator trainingsafe rehearsal of recovery procedures
- 08Twin synchronisationlive OPC UA feed into the scene
- 09Sim-to-real validationmeasured drift against the cell
- 10Isaac Lab policy trainingreinforcement and imitation
Applications
Where this is used.
Virtual commissioning
Robot and PLC code validated against an emulated controller before hardware is powered — the single largest schedule-risk reduction available.
Throughput simulation
Discrete-event models that size buffers and expose the real constraint before you buy equipment to fix the wrong one.
Synthetic data
Domain-randomised, pixel-perfect labelled imagery for defects that are too rare to photograph in production.
Operator training twins
Crews rehearse startup, changeover and fault recovery on the twin while the panel is still being wired.
Living twin
The commissioned scene fed by live OPC UA data, so drift between design intent and reality becomes visible.
Robot learning environments
Isaac Lab setups for training manipulation policies with domain randomisation and sim-to-real transfer.
Stack
Tools and platforms.
Related
Where this connects.
Very few problems are solved by one discipline. These are the capabilities most often combined with this one.
FAQ
Questions about digital twin
What is the difference between a simulation and a digital twin?
A simulation is a model of how something would behave. A digital twin is connected to the real asset — it receives live data, and its state is expected to track the physical system. Both are useful; virtual commissioning uses simulation, while condition monitoring and drift detection need a true twin.
How accurate is a virtual commissioning model?
For logic, interlocks and sequence, it is functionally exact — the same code runs. For cycle time, we typically land within a few percent of measured, with the residual coming from process variables such as weld dwell or adhesive cure that are governed by physics we model approximately.
Do we need NVIDIA hardware to use a twin you build?
For interactive use of a large Omniverse scene, an RTX-class GPU is required. Discrete-event throughput models and controller emulation run on standard workstations. We size the requirement in the proposal so there is no surprise.
Can you build a twin of a line we already have?
Yes. We reconstruct from layout drawings, controller programs and site survey — including laser scanning where the as-built has diverged from the drawings, which it usually has.
Next step
Ready to scope your next automation programme?
Send us a drawing, a cycle-time target or a line layout. Our engineers respond with a technical assessment — not a brochure.