OpenRAL/rskill-act-aloha-aloha_insertion-fp32
rskill-act-aloha-aloha_insertion-fp32
OpenRAL rSkill (custom example) — ACT (Action Chunking Transformer) finetuned on the ALOHA bimanual peg-insertion task, packaged for OpenRAL.This package wraps `lerobot/act_aloha_sim_insertion_human` with a rskill.yaml manifest that adds capability checking, license surfacing, latency budgets, and local registry integration. It does not copy model weights.
It is the harder sibling of `rskill-act-aloha-aloha_transfer_cube-fp32` (cube transfer) and demonstrates how a single packaging format covers multiple task-specific checkpoints from the same paper. The runnable demo lives at scenes/benchmark/aloha_insertion.yaml and is wired into the top-level just sim-custom recipe.
Upstream model
Why no eval/ block?
This skill is shipped as a custom-example package, not as a reproduced benchmark entry. The paper's headline number for sim ALOHA insertion is markedly lower than the cube-transfer figure (the task is harder and the upstream protocol uses different camera intrinsics). We deliberately omit eval/ rather than copy paper numbers without an internal reproduction; per CLAUDE.md §6.4 that omission must be documented — this section is that documentation. Add eval/aloha_insertion.json once a local reproduction lands.
Supported robots
Same physical embodiment as the `act-aloha` sibling (cube transfer); the only difference is the task contact dynamics — peg insertion is harder than cube pick-and-place.
Sensors required
Single top-down RGB stream like the cube-transfer sibling — the checkpoint does not consume wrist or third-person views.
Manifest summary
Full schema: `openral_core.schemas.RSkillManifest`.
Run it
just sim-custom…which is equivalent to:
MUJOCO_GL=egl uv run --group sim openral sim run \
--config scenes/benchmark/aloha_insertion.yaml \
--save-video example_videosLicense
This rSkill package (rskill.yaml, README.md) is MIT to match the upstream weights. Commercial use is allowed (commercial_use_allowed: true).