Robot Safety

Industrial environments, such as warehouses and factory floors, require safe autonomous systems with “inside-out” protections on the robot itself and “outside-in” systems that monitor the broader workspace. 

Over 80% of U.S. manufacturing companies have already deployed industrial robots or are in the early stages of their deployment journey. This adoption highlights an immediate need to ensure robots are operating with AI-level intelligence. 

• Proactive Safety: Prevent incidents in real time and through proactive situational awareness.
• Functional Safety Standards Compliance: Implement solutions developed, tested, and inspected by TÜV Rheinland to comply with current and emerging functional safety standards.
• Enhanced Productivity: Automatically modulate robot efficiency across dynamic environments to optimize actions.
• Reduce Downtime: Negate costly operation slowdowns and improve safety without sacrificing productivity using closed-loop training and optimization. 
• Maximize Sensor Infrastructure: Use existing building-mounted cameras to their full potential for real-time analytics, alerts, and adaptive reasoning.

Both traditional and smart robots rely on onboard sensors and cameras to see, perceive, and act based on the conditions of the world around them, referred to as inside-out safety. A robot monitors its immediate surroundings to detect workers, objects, or hazards and then slows down, stops, or adjusts its actions to avoid incidents. 

Onboard sensors have a limited field of view when dealing with occlusions like static walls, payloads, and moving objects such as workers and vehicles. With an outside-in safety agent, robot awareness expands beyond the robot's eyes by communicating with sensors and cameras placed throughout the facility. 

This broader 365° view helps agents understand, predict, and monitor incredibly complex environments that onboard sensors might miss. The result is more proactive protection, fewer slowdowns, and safer, more efficient operations.

Earning safety certification is crucial for industrial automation companies to demonstrate readiness for real-world deployments, meeting rigorous safety and AI integrity requirements. This compliance gives warehouse operators confidence, ensuring deployments have been tested and evaluated against best practices for safe and secure operations.

The NVIDIA Halos AI Systems Inspection Lab offers accreditation tools, providing hardware and connection to platform companies with compliance recognition such as ANAB, TÜV Rheinland, and other third‑party agencies.

Running on the NVIDIA IGX™ platform, outside-in functional safety agents fuse low-latency detections with safety monitoring and decision logic to supervise multiple robots simultaneously. 

The outside-in safety workflow uses NVIDIA Halos, NVIDIA Metropolis-based video analytics AI agents, and the NVIDIA Isaac Sim™ open simulation framework to create a closed-loop training platform. Together, these technologies enable robots to think, perceive, and act autonomously.

The agent interacts with the robot to slow or stop the moment an object enters a protected area, while allowing higher speeds and closer collaboration when zones are clear. This approach helps manufacturers reduce safety incidents and false stops, increase robot throughput, and more easily navigate occlusions. 

In this safety concept, successfully inspected by TÜV Rheinland, outside-in cameras establish a virtual fence and monitor the trailer interior and docking zones for occlusion alerts. Robots operate at full speed or high-efficiency mode when no workers are in the region of interest, and engage safety functions the moment a person enters to prevent incidents from occurring.

Virtual tripwires and dynamic zones manage multiple AMR robots and people in shared aisles and intersections, including around blind corners or high-rack areas. Especially in areas with material handling, alerts and occlusions detection are imperative when a person or robot enters a hidden zone or is obscured by an object, maintaining safety even in blind spots.