Neurealm has adapted its vision-based safety system for industrial robots to run on Advantech’s MIC-735-IT Edge computing platform, using NVIDIA technology to detect potential hazards and trigger emergency stops.
The system combines Advantech’s industrial-grade platform, built on NVIDIA IGX Thor, with the NVIDIA Halos Outside-In Safety Blueprint and IGX Thor Safety Extension Package. It is designed to provide an additional layer of protection for robots working alongside people by monitoring their surroundings through external cameras rather than relying solely on onboard sensors.
Neurealm and Advantech will demonstrate the technology at the International Robot Safety Conference (IRSC) in Detroit from 4th November.
Conventional robot safety systems typically use sensors mounted on the robot to monitor its immediate operating environment. The outside-in approach instead uses cameras positioned around the workspace to create a broader view of activity and identify potential hazards that might fall outside the robot’s own field of perception.
The system processes camera feeds through NVIDIA’s Halos pipeline. AI software running on the IGX Thor central complex (CCPLEX) detects potentially unsafe events and passes triggers to the platform’s Functional Safety Island (FSI).
The FSI aggregates these signals and monitors the operation of the main computing complex before generating hardware emergency stop signals to bring machinery to a safe state.
Separating AI-based perception from the safety response is intended to allow manufacturers to use vision-based monitoring without relying on the main AI computing system alone to execute safety functions.
“Vision AI has given manufacturers a far richer picture of what is happening around a robot. The next step is making sure that insight can trigger a dependable safety response,” said Sanjay Jayakumar, President of Semiconductor, Embedded and Automotive Business at Neurealm.
Advantech said the MIC-735-IT provides the computing and hardware foundation for developing outside-in safety architectures, with support for Safety MCU integration. Its design and manufacturing services are intended to help customers develop systems around application-specific requirements, including those derived from IEC 61508 and ISO 13849.
