Hugging Face's logo

Datasets:
eloisezeng
/
playground-test-v5-demos

Tasks:
Robotics
Reinforcement Learning
Size:
1K<n<10K
Tags:
quadruped
locomotion
navigation
imitation-learning
mujoco
License:
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    ValueError
Message:      Dataset 'episode_starts' has length 10407 but expected 8654969
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 99, in get_rows_or_raise
                  return get_rows(
                         ^^^^^^^^^
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                         ^^^^^^^^^^^^^^^^^^^^^
                File "/src/services/worker/src/worker/utils.py", line 77, in get_rows
                  rows_plus_one = list(itertools.islice(ds, rows_max_number + 1))
                                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2690, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2227, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2251, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 494, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 384, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/hdf5/hdf5.py", line 80, in _generate_tables
                  num_rows = _check_dataset_lengths(h5, self.info.features)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/hdf5/hdf5.py", line 359, in _check_dataset_lengths
                  raise ValueError(f"Dataset '{path}' has length {dset.shape[0]} but expected {num_rows}")
              ValueError: Dataset 'episode_starts' has length 10407 but expected 8654969

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

V5 forest navigation demonstrations (Unitree Go1)

Expert demonstrations for the V5 75D navigation observation (2D goal + 64D LIDAR + 9D action history) collected on the 30x30 m forest task at training-time obstacle density δ = 0.04 trees/m².

The expert is an A* planner + Pure Pursuit controller over a privileged obstacle map; the demos are intended to train LIDAR-only students (PIL-Control, PIL-Trajectory, IQL, CQL).

Files

File Size What
v5_clean.hdf5 ~1.2 GB Per-step (75D obs, 3D velocity action) pairs from successful episodes.
v5_clean_trajectory.hdf5 ~1.5 GB Same episodes converted to horizon-10 waypoints [x1,y1,...,x10,y10] (meter-valued, un-normalized).
scenarios_30x30_v5.pkl ~62 MB Plannable start/goal/obstacle layouts used to generate the demos (11k attempted, 10.4k plannable = 94.6%).
v5_clean.config.yaml <1 KB Sidecar with the collector's exact env parameters.
scenarios_30x30_v5.config.yaml <1 KB Sidecar with the scenario generator's parameters (δ=0.04, boundary_safety_margin=0.5, room=30x30, base_seed=42).

Provenance

  • Generated: 2026-06-01
  • Source repo: https://github.com/Jaeha0526/playground_test
  • Code revision: 637dbe20a848bbfba4bf1c322245fa45c4c3e341 (at upload time)
  • Pipeline (see CLAUDE.md in the source repo):
    1. src/navigation/imitation/forest/generate_scenarios_parallel.py — scenario layouts
    2. src/navigation/imitation/forest/collect_forest_demos_vmap.py — expert rollouts
    3. src/navigation/imitation/forest/convert_to_trajectory_dataset_from_positions.py --horizon 10 — waypoint conversion

Quick start 

from huggingface_hub import hf_hub_download

hdf5 = hf_hub_download(
    repo_id="eloisezeng/playground-test-v5-demos",
    filename="v5_clean.hdf5",
    repo_type="dataset",
)
# hdf5 is a local cached path; reuse on subsequent calls is free.

To pull all five files at once into a local directory:

from huggingface_hub import snapshot_download

local_dir = snapshot_download(
    repo_id="eloisezeng/playground-test-v5-demos",
    repo_type="dataset",
    local_dir="data/imitation/forest",
)

Set HF_XET_HIGH_PERFORMANCE=1 in your shell for ~10x faster downloads on multi-GB files (uses HF's Xet content-addressed transport).

License

MIT — same as the source repo.

This README was auto-generated by autoresearch_round5/tools/upload_v5_demos_to_hf.py on 2026-06-02.
Downloads last month
31
Total file size:
2.78 GB