Skip to content

installation.md

Latest commit

 

History

History
70 lines (45 loc) · 1.85 KB

installation.md

File metadata and controls

70 lines (45 loc) · 1.85 KB

Installation

Environment Setup

conda create -n rise python=3.11.14 -y
conda activate rise

cd /path/to/RISE
bash install.sh

For deployment setup, please refer to the deployment README.

Data Preparation

The training code expects LeRobot-style data.

The data is first collect on Piper in hdf5 format and then converted to the lerobot format.

Raw Data Collection

Please refer to the data collection README for details.

Lerobot Conversion

Convert raw HDF5 logs collected on Piper into LeRobot-format data with:

cd /path/to/RISE/policy_and_value/policy_offline_and_value

python examples/aloha_real/convert_to_lerobot.py \
  --data-dir /path/to/raw_dataset \
  --repo-ids aloha_mobile_dummy \
  --prompt "TASK_PROMPT" \   # e.g., "Pick up the block"
  --save-dir /path/to/lerobot_output_root \
  --save_repoid output_dataset_name \

Arguments:

  • --data-dir: Root directory of one raw dataset collection. The script expects HDF5 files under <data-dir>/<repo-id>/ and videos under <data-dir>/<repo-id>/video/.
  • --repo-ids: One or more raw subdirectories to convert. For most RISE data, this is aloha_mobile_dummy.
  • --prompt: Task description stored in the converted LeRobot dataset.
  • --save-dir: Parent directory for converted output.
  • --save_repoid: Name of the output dataset folder created under --save-dir.
  • --overwrite: Optional. Remove an existing output folder and regenerate it.

The generated output follows the standard LeRobot layout shown below.

Expected Lerobot Data Layout

<dataset_name>/
├── data/chunk-000/episode_*.parquet
├── meta/
│   ├── info.json
│   ├── episodes.jsonl
│   ├── episodes_stats.jsonl
│   └── tasks.jsonl
└── videos/chunk-000/*.mp4