███████╗ ██████╗ ██████╗  ██████╗ ███████╗
██╔════╝██╔═══██╗██╔══██╗██╔════╝ ██╔════╝
█████╗  ██║   ██║██████╔╝██║  ███╗█████╗
██╔══╝  ██║   ██║██╔══██╗██║   ██║██╔══╝
██║     ╚██████╔╝██║  ██║╚██████╔╝███████╗
╚═╝      ╚═════╝ ╚═╝  ╚═╝ ╚═════╝ ╚══════╝

⚒ Robotics Data Toolkit ⚒

Convert, inspect, visualize, and score robotics datasets across every major format.



RLDS ═══╗ ╔═══► LeRobot
Zarr ═══╬════⚙════╬═══► RoboDM
HDF5 ═══╝ ╚═══► RLDS

Convert between robotics dataset formats with one command. Score demonstration quality with research-backed metrics.

Format	Read	Write	Visualize	Notes
RLDS	✓	✓	✓	Open-X, TensorFlow Datasets
LeRobot v2/v3	✓	✓	✓	HuggingFace, Parquet + MP4
GR00T	✓	-	✓	NVIDIA Isaac, LeRobot v2 with embodiment metadata
RoboDM	✓	✓	✓	Berkeley's .vla format, up to 70x compression*
Zarr	✓	-	✓	Diffusion Policy, UMI
HDF5	✓	-	✓	robomimic, ACT/ALOHA
Rosbag	✓	-	✓	ROS1 .bag, ROS2 MCAP

*RoboDM requires manual installation from GitHub (see below)

See docs/model_formats.md for which models (Octo, OpenVLA, ACT, Diffusion Policy, etc.) use which format. See docs/format_reference.md for detailed format specifications.

Why Forge?

Every robotics lab has their own data format: Open-X uses RLDS, HuggingFace uses LeRobot, Diffusion Policy uses Zarr, robomimic uses HDF5. Want to train Octo on your ALOHA data? Write a converter. Want to use LeRobot on Open-X datasets? Write another.

Forge uses a hub-and-spoke architecture — one intermediate representation, O(n) format support:

Any Reader → Episode/Frame → Any Writer

Add a reader, get all writers for free. Add a writer, get all readers for free. No N×M conversion logic. See docs/architecture.md for details.

Quick Start

git clone https://github.com/arpitg1304/forge.git
cd forge
pip install -e ".[all]"

RoboDM Support (Optional)

RoboDM requires manual installation from GitHub (PyPI version has a codec bug):

git clone https://github.com/BerkeleyAutomation/robodm.git
pip install -e robodm

Usage

# See what's in a dataset
forge inspect /path/to/dataset

# Convert it
forge convert /path/to/rlds ./output --format lerobot-v3
forge convert hf://arpitg1304/stack_lego ./stack_lego_rlds --format rlds --workers 4 --visualize
forge convert hf://lerobot/pusht ./pusht_robodm --format robodm

Works with HuggingFace Hub too:

forge inspect hf://lerobot/pusht
forge convert hf://lerobot/pusht ./output --format lerobot-v3

Python API

import forge

# Inspect
info = forge.inspect("/path/to/dataset")
print(info.format, info.num_episodes, info.cameras)

# Convert
forge.convert(
    "/path/to/rlds",
    "/path/to/output",
    target_format="lerobot-v3"
)

Quality Metrics

Automated episode-level quality scoring from proprioception data alone — no video processing needed.

forge quality ./my_dataset
forge quality hf://lerobot/aloha_sim_cube --export report.json

Scores each episode 0-10 based on 8 research-backed metrics:

• Smoothness (LDLJ) — jerk-based smoothness from motor control literature (Hogan & Sternad, 2009)
• Dead actions — zero/constant action detection (Kim et al. "OpenVLA", 2024)
• Gripper chatter — rapid open/close transitions (Sakr et al., 2024)
• Static detection — idle periods where the robot isn't moving (Liu et al. "SCIZOR", 2025)
• Timestamp regularity — dropped frames and frequency jitter
• Action saturation — time spent at hardware limits
• Action entropy — diversity vs repetitiveness (Belkhale et al. "DemInf", 2025)
• Path length — wandering/hesitation in joint space

See forge/quality/README.md for full metric details, paper references, and how to add new metrics.

CLI Reference

See docs/cli.md for the full command reference including:

• forge inspect - Dataset inspection and schema analysis
• forge convert - Format conversion with camera mapping
• forge visualize - Interactive dataset viewer
• forge quality - Episode-level quality scoring (details)
• forge stats - Compute dataset statistics
• forge export-video - Extract camera videos as MP4
• forge hub - Search and download from HuggingFace

Configuration

For complex conversions, use a YAML config:

forge inspect my_dataset/ --generate-config config.yaml
forge convert my_dataset/ output/ --config config.yaml

See docs/configuration.md for details.

Roadmap

Planned features (contributions welcome!):

• Dataset merging - Combine multiple datasets into one (forge merge ds1/ ds2/ --output combined/)
• Train/val/test splitting - Split datasets with stratification (--split 80/10/10)
• Streaming reads - Process HuggingFace datasets without full download
• Episode filtering - Filter by quality score, flags, or episode IDs (forge filter --min-quality 6.0)
• Depth/point cloud support - Preserve depth streams from RLDS/Open-X
• GR00T writer - Write to NVIDIA Isaac GR00T training format (read support complete)
• Distributed conversion - Scale to 100K+ episode datasets across nodes
• Conversion verification - Automated diff between source and converted data

Development

make venv && source .venv/bin/activate
make install-dev
make test

License

MIT