RealSense and NVIDIA Team Up to Push Physical AI Into the Mainstream

In the race to bring humanoids and autonomous mobile robots out of the lab and into the real world, RealSense and NVIDIA are joining forces. The companies announced a deep integration effort that fuses RealSense’s depth-sensing hardware with NVIDIA’s most advanced robotics platforms, aiming to shrink development cycles and supercharge the adoption of physical AI.

At the heart of the collaboration is RealSense’s new D555 depth camera. It debuts the company’s v5 Vision Processor and comes equipped with features that robotics engineers have long wanted in a single package: a ruggedized chassis, global shutter, on-chip Power over Ethernet, and embedded support for ROS 2. Crucially, it also delivers native streaming to NVIDIA’s Holoscan Sensor Bridge, giving robots ultra-low-latency perception data and on-camera neural network processing.

On the compute side, NVIDIA is bringing in its new Jetson Thor module, powered by the Blackwell GPU. With 128GB of memory and 2,070 FP4 teraflops of AI performance inside a 130-watt power envelope, Jetson Thor represents a massive leap from the Orin generation—up to 7.5 times more compute and 3.5 times the energy efficiency. That means humanoids and AMRs can run large generative AI models at the edge without a data-center-level power bill.

The integration of RealSense sensors with NVIDIA’s Isaac Sim and Holoscan stack is designed to reduce the gap between prototype and production. Developers can stream high-fidelity depth and image data directly into Isaac’s digital twin environment, iterate faster, and validate real-time robotics workloads in simulation before deploying them in the physical world.

“This initiative cements RealSense’s role as the perception platform of choice for AMRs and humanoids,” said Nadav Orbach, CEO of RealSense. “By providing native integration and performance optimizations with NVIDIA Thor and Holoscan Sensor Bridge, we are accelerating the mainstream adoption of physical AI. Together, we are enabling the robotics industry to unlock the extraordinary potential of physical AI and drive the future of intelligent machines.”

For NVIDIA, the partnership underscores its strategy of building not just the AI brain for robots but the entire ecosystem—compute, simulation, and sensor fusion. For RealSense, it positions the company as a crucial bridge between raw perception data and the AI algorithms that will control next-generation machines.

With forecasts placing the robotics market in multi-trillion-dollar territory, both companies are betting that faster iteration, validated system designs, and native integration will be the key accelerants. The combination of RealSense’s vision technology and NVIDIA’s compute muscle could mark a turning point where humanoids and AMRs finally scale safely into everyday environments—factories, warehouses, and even homes.