A technician walks through the shop floor with a tablet. Points the camera at a machine — sees the wiring diagram, list of parts under the cover, disassembly animation. Without AR, it's paper manuals, catalog searches, minutes per identification. AR turns the tablet into a diagnostic tool. Our team, with 5 years of experience, has delivered 30+ AR projects for machine building, oil & gas, and energy. We guarantee positioning accuracy up to 1 mm on iPad Pro with LiDAR. AR instructions work 3–5 times faster than paper — the technician doesn't flip through catalogs, but immediately sees data on the part. An average project saves 50 hours per month per workstation.
How does an AR system recognize equipment?
The main technical challenge is anchoring virtual annotations to the real asset. We use three approaches:
| Method | Accuracy | Markers | Hardware |
|---|---|---|---|
| Image Tracking | ~1–2 mm | QR code or label | Any smartphone/tablet |
| Object Scanning | ~5–10 mm | None | iOS with LiDAR or textured objects |
| LiDAR + ICP | ~1 mm | None | iPad Pro with LiDAR |
Image Tracking is the most reliable for production: ARKit ARImageTrackingConfiguration detects the marker, positioning accuracy ~1–2 mm, works on any device. Object Scanning (ARKit ARObjectScanningConfiguration) creates an .arobject file but requires sufficient texture — a smooth metal casing won't work. LiDAR Scene Reconstruction builds a mesh of the environment and compares it to a reference via Iterative Closest Point, but is computationally expensive.
Which stacks to use for industrial AR?
For enterprise tasks, there are ready platforms. PTC Vuforia Engine is the de-facto standard: Model Target recognizes by CAD model without markers, Area Target scans a room. SDK for iOS/Android/Unity. Scope AR WorkLink is a no-code platform for creating AR instructions via a browser. Upskill Skylight is an analogue focused on checklists. Custom development is justified when deep integration with ERP/CMMS (SAP PM, IBM Maximo, Infor EAM), specialized UI, or offline work on rugged devices is required.
Comparison of platforms for industrial AR:
| Parameter | Vuforia Engine | ARKit/ARCore | Custom Development |
|---|---|---|---|
| CAD recognition | Yes (Model Target) | No (only image/object) | Can be implemented |
| Offline mode | Yes (cloud targets) | Limited | Full control |
| ERP integration | Via REST | Via REST | Direct SDKs |
| Licensing | Commercial | Free | N/A |
What difficulties arise when integrating AR with ERP?
A typical problem is data synchronization between the AR app and ERP/CMMS. For example, SAP PM requires a strict format via BAPI for creating repair orders. In offline mode, data is buffered locally and synced after network recovery to avoid conflicts. Another challenge is CAD model conversion: even after conversion to GLTF via Open CASCADE Technology or Datasmith, manual scaling and alignment to equipment coordinates are often required.
Step-by-step process for creating AR instructions
- Load CAD model in STEP/IGES format and convert to GLTF via Open CASCADE Technology.
- Attach annotations to nodes: arrow + text "loosen M8 fasteners × 4 pcs" to a specific point.
- Disassembly animation: CAD model "slides out" showing the part removal path.
- Create a checklist with auto-transition — the system repositions annotations after each step.
- Export to AR app with local storage.
What's included in our work
- Technical audit of requirements and operating conditions
- AR app development for iOS/Android (ARKit/ARCore)
- CAD model conversion to AR-compatible formats
- Integration with ERP/CMMS via REST API or GraphQL
- Testing on real equipment with accuracy measurements
- Documentation and staff training
- Post-launch technical support
To choose the recognition approach: if equipment can be tagged with QR markers — choose Image Tracking (cheap and reliable). If markers are not allowed and the equipment has texture — Object Scanning. For maximum accuracy without markers on iPad Pro — LiDAR + ICP. The specific method is selected after audit.
Integration with ERP and CMMS
Typical integrations: SAP PM (Plant Maintenance), IBM Maximo, Infor EAM. The AR app receives the equipment serial number from ERP and loads the appropriate instructions. After completing maintenance, it writes work order completion back via API. For limited internet scenarios, offline sync is implemented: data is stored locally (SQLite / Core Data / Room) and synced upon connection restoration.
Timelines and cost
MVP (image tracking + static annotations without ERP integration) — 4–6 weeks. Full solution with animated instructions, offline sync, and integration — 3–6 months. Cost is calculated individually after requirements audit. Request a consultation — we will assess your project end-to-end. Write to us to discuss details.
Get a consultation from our engineers: we will assess your project, choose the optimal stack, and propose an implementation plan.