FORT Robotics has extended its safety platform for autonomous machines with a new capability called Outside-In Safety, developed in collaboration with NVIDIA as part of the NVIDIA Halos for Robotics ecosystem. The companies presented the work in the Humanoid Robotics Pavilion at the Automate conference on June 23, 2026.
Beyond onboard sensors
The blueprint pairs FORT's Trust Layer with NVIDIA's Outside-In Safety architecture to push robot perception beyond a machine's own cameras and lidar. Using external infrastructure sensors and visual AI agents, the system delivers real-time, safety-certifiable functional safety, running on NVIDIA IGX Thor and the Holoscan Sensor Bridge for AI compute and sensor connectivity.
Safety that keeps throughput high
Rather than forcing blanket slowdowns, Outside-In Safety modulates robot speed dynamically as conditions change, letting machines run in high-efficiency modes around people while still preventing incidents. FORT says that cuts the costly slowdowns that erode the business case for automation in warehousing and manufacturing, sectors also chasing the physical AI gains shown by ABB and Roboteon at Automate.
An open blueprint
The Outside-In Safety blueprint is open source, and FORT demonstrated an agentic safety application built on it for new and existing customers. The collaboration deepens NVIDIA's push to standardise safety primitives across the robot industry, complementing the compute and models it is supplying to partners like Richtech Robotics.
Reporting based on coverage from The Robot Report, PR Newswire and FORT Robotics.
