Belden and Accenture have developed physical AI systems for worker safety in factories and warehouses.

Belden’s virtual safety fence solution was developed with Accenture’s “Physical AI Orchestrator,” which uses NVIDIA Omniverse, NVIDIA Metropolis libraries for digital twins and agentic AI from Accenture. It applies computer vision and closed-loop control to monitor worker movements and automatically pause robotic operations when humans enter hazardous zones.

Following a pilot project completed in September, the virtual safety fence system is expected to be commercially deployed at an automotive manufacturer for pedestrian safety in warehouse environments later this year.

The system is depends on Belden’s Time Sensitive Networking (TSN), which allows microsecond-precise synchronisation of video streams from multiple cameras across factory floors. This synchronisation allows accurate tracking of people and equipment movement in real-time safety applications.

“Industrial-grade networking protocols are critical for safety applications that require precise timing,” said German Fernandez, VP of Ecosystem Partner Programmes at Belden. “Our TSN technology enables real-time, closed-loop control between AI systems and production equipment, which is fundamental for both safety and quality applications.”

“The physical AI systems we help Belden build on Accenture’s Physical AI Orchestrator platform address the critical need to protect workers while maintaining the operational flexibility that modern manufacturing demands,” said Ramalingam Hariharan, a managing director at Accenture.