Nomadic publishes VLA-ready robotics video annotation workflow with Hugging Face Buckets

Mar 21, 2026 · Capability · Nomadic · Data Infrastructure

Nomadic company media
Company media · Nomadic

Evidence notes

  • Nomadic published a workflow showing how raw robotics video stored in Hugging Face Buckets can be annotated and converted into VLA-training-ready data using its platform.
  • The article described Nomadic''s ability to segment robotic actions, identify objects, infer 3D positions from RGB video, and generate timestamped outputs that map into LeRobot v3-compatible dataset artifacts.
  • Nomadic also showed direct integration paths through both its UI and SDK, including native support for hf://buckets URIs.

Company context

Nomadic builds visual data infrastructure for robotics and autonomous-vehicle teams. Its platform helps teams search, inspect and learn from large volumes of robot and vehicle video.

More from Nomadic

Links