Ouster has introduced the Stereolabs ZED X Nano, describing it as a wrist-watch camera for robots designed for manipulation tasks, imitation learning, and high-throughput data collection. The device is built to mount directly on the wrist of robotic arms, with a compact design that is smaller than comparable systems. It uses a 1920×1200 global shutter sensor to capture RGB and depth data at up to 120 frames per second, targeting applications that require fast and precise visual feedback.
The ZED X Nano replaces conventional USB-based camera setups with a direct sensor-to-GPU data pipeline, eliminating intermediate data copying. This architecture allows image frames to be used simultaneously for encoding and AI inference, reducing latency and increasing processing efficiency.
The camera includes an onboard inertial measurement unit and uses GMSL2 connectivity designed for industrial environments. The connection supports cable lengths of up to 15 meters and is intended to withstand vibration and repeated motion associated with robotic arms. Depth sensing is powered by a neural stereo vision system developed by Stereolabs. The system provides sub-millimeter accuracy along the depth axis and improved lateral positioning compared with structured-light and time-of-flight systems. The camera can detect objects at distances as close as 3 centimeters.
The ZED X Nano integrates with robotics and simulation platforms including NVIDIA Isaac Sim and ROS, enabling workflows that span data capture, simulation-based training, and deployment on physical systems.
