Robotics — Compute & Capture Infrastructure

The flagship on-robot compute for humanoid and multi-arm platforms. 2,070 FP4 TFLOPS and 128 GB of LPDDR5X means a large vision-language-action mod...

On-robot compute for a full-size rig. 275 TOPS and 64 GB of LPDDR5 is enough to run multi-camera perception, an Isaac ROS graph and policy inferenc...

A deskside training box with 128 GB of coherent unified memory. In a robot-learning programme this is the machine that sits beside the capture rig:...

The entry point into on-robot inference. 67 TOPS in a 100 × 79 mm board at 7-25 W — enough for stereo perception and a small policy on a teleoperat...