Skip to content
humanoidrobots.training

Starting viewport

Stanford University · humanoid

ToddlerBot

Low-cost, fully 3D-printed open-source small humanoid (0.56 m, 3.4 kg, 30 DOF) for scalable policy learning research.

Frames · ToddlerBot

6 items · credit and license on each
  • ToddlerBot studio render, three-quarter front view
    Studio render of the published MJCF, keyframe pose from the model. Geometry only; colours are the model's own materials.Rendered by Hyperspawn from mujoco_menagerie/toddlerbot_2xc (MIT) · Hyperspawn render · source
  • ToddlerBot studio render, side view
    Side view of the same MJCF render, keyframe pose from the model.Rendered by Hyperspawn from mujoco_menagerie/toddlerbot_2xc (MIT) · Hyperspawn render · source
  • 4 s turntable of the MJCF studio render.Rendered by Hyperspawn from mujoco_menagerie/toddlerbot_2xc (MIT) · Hyperspawn render · source
  • Three ToddlerBot humanoids standing side by side with coloured chest panels
    ToddlerBot banner from the project repo.Haochen Shi et al. (ToddlerBot) · MIT · source
  • ToddlerBot: Open-Source ML-Compatible Humanoid Platform for Loco-ManipulationHaochen Shi · official embed · source
  • ToddlerBot 2.0 Arm OverviewHaochen Shi · official embed · source

Provenance

Model
ToddlerBot 2XC
Repo
google-deepmind/mujoco_menagerie
Commit
c96a32d28fb5
Entry
toddlerbot_2xc/scene.xml
Files
49 · 65 MB, fetched from raw.githubusercontent.com, cached in your browser
License
MIT
Engine
MuJoCo 3.14.0, official WebAssembly build · three.js renderer
Controller
PD hold on the stand keyframe with an ankle-strategy tilt term; not a learned policy
Caveat
Loads and runs, but our generic PD stand hold does not keep it up; it needs its own balance controller or a trained policy.

Policies listed for ToddlerBot

  • walk (omnidirectional, 2XC)

    MuJoCo MJX + RSL-RL PPO, ONNX Runtime on Jetson · not replayed in the tab

    source

  • get up from the ground

    MuJoCo MJX + RSL-RL PPO · not replayed in the tab

    source

  • crawl (full / only / rotate)

    MuJoCo MJX + RSL-RL PPO · not replayed in the tab

    source

  • crawl up / down stairs

    MuJoCo MJX + RSL-RL PPO · not replayed in the tab

    source

  • climb up / down box, climb wall

    MuJoCo MJX + RSL-RL PPO · not replayed in the tab

    source

  • multi-skill whole-body locomotion (Locomotion Beyond Feet)

    RL skill policies + depth skill classifier (FoundationStereo) · not replayed in the tab

    source

  • push-up (keyframe replay)

    open-loop keyframe replay (run_policy.py --policy replay) · not replayed in the tab

    source

  • cartwheel (DeepMimic-style)

    MuJoCo MJX + RSL-RL PPO · not replayed in the tab

    source

  • bimanual manipulation (diffusion policy)

    Diffusion Policy (PyTorch) · not replayed in the tab

    source

Train on GPUs

mjlab ships no stock task for ToddlerBot. The training page gives you the pinned model and the steps to write a robot config.

Build the job →