# ViNT / NoMaD training mix

From the catalogue. Listed by one researcher from one source page.

- id: `vint-nomad-training-mix`
- kind: teleop
- robot or device: multiple ground robots (Jackal, TurtleBot2, Spot, ATV, Roomba-based) (class: cross embodiment)
- how collected: mix of teleop and autonomous public sets plus unreleased data
- size as stated: hundreds of hours (ViNT abstract)
- hours: not stated | hours only claimed: none | episodes: not stated
- year: 2023
- organisation: UC Berkeley
- licence: MIT (code repo); datasets carry their own terms (class: unclear)
- access: open for the public parts (class: open)
- link (dataset): https://github.com/robodhruv/visualnav-transformer
- note: Public parts: RECON, TartanDrive, SCAND, GoStanford2, SACSoN/HuRoN. Rest unreleased.

Confirm the licence at the link before relying on it.