Blog

  1. Serving 256 robots from a single node

    Cloud inference for robots only makes economic sense if each GPU serves many robots. We describe how one 8×H100 node serves the π0.5 model to 256 robots, with 99% of requests answered within 50 ms.

  2. Can it Excavate?

    An experiment of how well state-of-the-art VLAs and WAMs can control an excavator.

  3. Transport V2

    On a home Wi-Fi link, Transport V2 reduced p99 network latency from 420 ms to 160 ms for a robot running inference in the cloud.

  4. Evaluating robot models in the cloud

    HUD partners with Dreamscale Labs for cloud inference, powering massively parallel sim evals, world action models and real-world robot rollouts.

  5. Why Robot Brains Will Live in the Cloud

    The case for off-board robotics inference