armteam/crab-manipulation-dataset

Crab Manipulation Dataset

Teleoperated demonstration dataset for contact-rich manipulation on the Crab bimanual mobile manipulator (dual SO-101 arms). Collected at 15 Hz with 3 RGB cameras (640×480) and dual 10×10 tactile force sensor matrices.

Dataset Structure

Clean demonstrations (27 sessions, used for training)

Negative examples (7 sessions, used for RWFM reward weighting)

Data Format

Each session contains:

session_name/
  ├── data/          # Action and state parquet files
  ├── meta/          # Episode metadata
  ├── videos/        # Compressed video streams (AV1)
  └── images/        # (if applicable)

Observation Space

• —3 cameras: main (overhead), left arm, right arm — 640×480 RGB
• —Proprioception: 12-DOF joint positions (6 per arm)
• —Tactile: 2×10×10 force matrices (left/right fingers of parallel gripper)

Action Space

• —12-DOF absolute joint positions (6 per arm)
• —Recorded at 15 Hz via teleoperation

Related Models

• —crab-smolvla-left-arm — Left arm baseline
• —crab-smolvla-right-arm — Right arm baseline
• —crab-smolvla-rwfm — SA-RWFM teacher (tactile)
• —crab-smolvla-hapticsvla — HapticVLA (distilled)

Citation

If you use this model, please cite our paper:

@article{gubernatorov2026hapticvla,
  title={HapticVLA: Contact-Rich Manipulation via Vision-Language-Action Model without Inference-Time Tactile Sensing},
  author={Gubernatorov, Konstantin and Sannikov, Mikhail and Mikhalchuk, Ilya and Kuznetsov, Egor and Artemov, Makar and Ouwatobi, Ogunwoye Faith and Fernando, Marcelino and Asanov, Artem and Guo, Ziang and Tsetserukou, Dzmitry},
  journal={arXiv preprint arXiv:2603.15257},
  year={2026}
}