Through the support of the NEPHELE Open Call, Secmotic has successfully deployed its AIoT-based safety solution, AIoTGuard, in active industrial warehouses, moving from controlled testing conditions to real-life operational contexts.

This deployment marks a significant milestone in validating the system’s capacity to enhance occupational safety in settings with constant interaction between workers and automated vehicles, such as forklifts.

From validation to real‑world trials

After successfully validating our core use‑case in Stage 2, we deployed AIoTGuard in active distribution centers where thousands of operations occur daily. Working side‑by‑side with warehouse personnel, we gathered invaluable feedback:

• Instant hazard detection: the system issued alerts in situations where workers were at risk of being struck by moving vehicles.
• User confidence: operators quickly understood the value of the system and expressed trust in its ability to prevent incidents.
• Seamless integration: minimal operational disruption—camera and projector setup on the existing network was completed in a short amount of time.

Key technical breakthrough: VO‑based scalability

A key development from Stage 2 in NEPHELE Open Call —replacing a single projector with a Virtual Object (VO) architecture using relays and switches—proved highly effective. This enhancement enabled:

1. Connected multiple camera‑projector pairs through one of eight switch ports.
2. Deployed new VOs on‑the‑fly, dynamically instantiating analytics pipelines.
3. Left six ports free, laying the groundwork for future multi‑angle coverage and new sensors.

This VO‑oriented model proved that AIoTGuard isn’t a one‑off demo—it’s a truly modular, plug‑and‑play architecture ready to expand.

Seamless Interoperability & Edge Intelligence

AIoTGuard fuses three pillars:

• Edge‑based AI: real‑time image analysis without constant cloud round‑trips.
• Modular hardware: cameras, projectors, switches and IoT sensors that snap together.
• Open standards: FIWARE‑compatible VOs orchestrated via NEPHELE for end‑to‑end interoperability.

Together, they deliver an intelligent operational layer that evolves from simple event detection (absence of PPE, falls) to predictive insights—identifying high‑risk zones before incidents occur.

Measurable impact

Our trials tracked:

• Detection accuracy: > 95 % true‑positive rate.
• Latency: < 200 ms from image capture to alert.
• Worker trust: 90 % of surveyed staff reported reliance on system alerts.

These metrics not only met but often exceeded our Stage 1 targets, confirming AIoTGuard’s technical robustness and operational value.

5. What’s next?

Following this successful deployment, Secmotic’s roadmap includes:

• Development of new Virtual Objects with expanded features
• Hybrid architectures combining edge and cloud capabilities
• Pilot programs across various industrial sectors

By bringing advanced AIoT technology to real operational settings, AIoTGuard redefines workplace safety through intelligent automation.

Ready to see AIoTGuard in action?
📽️ Watch our behind‑the‑scenes videos documenting the trial: