RealSense has announced a collaboration with NVIDIA aimed at advancing the development and deployment of physical AI in humanoid robots and autonomous mobile robots (AMRs). The partnership integrates RealSense’s AI-enabled depth camera technology with NVIDIA’s robotics computing platforms, including the Jetson Thor Series, Isaac Sim, and Holoscan Sensor Bridge.
The initiative includes the integration of the newly launched RealSense D555 depth camera, which features the company’s v5 Vision Processor and on-chip Power over Ethernet. The D555 also offers native support for NVIDIA Holoscan Sensor Bridge streaming and on-camera neural network capabilities. Its global shutter, inertial measurement unit (IMU), and compatibility with ROS 2 are designed to support applications requiring high-fidelity perception data.
NVIDIA’s Jetson Thor, powered by the Blackwell GPU architecture and equipped with 128GB of memory, provides up to 2070 FP4 teraflops of AI compute. It is designed to run generative AI models within a 130-watt power envelope and delivers an increase in AI performance and energy efficiency compared to the earlier Jetson Orin platform.
According to the companies, the integration of RealSense depth perception with NVIDIA’s compute and simulation platforms is intended to accelerate robotics development, reduce time to market, and support scaling into production environments. Native support for NVIDIA Isaac Sim allows for simulation-based testing and development, while integration with the Holoscan Sensor Bridge is expected to enable real-time sensor fusion and low-latency data streaming.
The companies stated that the collaboration will provide validated system architectures optimized for Jetson Thor, offering developers tools to build and deploy robotics and physical AI solutions more efficiently.
RealSense originated as a technology initiative within Intel, where it developed 3D depth-sensing cameras for consumer and industrial applications. In 2023, the unit was spun out from Intel as an independent company following a $50 million funding round led by strategic investors. As an independent entity, RealSense has focused on developing AI-powered vision systems for robotics, biometrics, and industrial automation, expanding its reach into physical AI and intelligent machines. RealSense is headquartered in Santa Clara, California, and operates globally.
