---
pretty_name: "EXOKERN ContactBench v0 - Peg Insertion with Force/Torque"
license: cc-by-nc-4.0
language:
  - en
task_categories:
  - robotics
size_categories:
  - 1K<n<10K
tags:
  - robotics
  - force-torque
  - contact-rich
  - manipulation
  - insertion
  - lerobot
  - isaac-lab
  - forge
  - franka
  - panda
  - simulation
  - benchmark
configs:
  - config_name: default
    data_files:
      - split: train
        path: "data/**/*.parquet"
dataset_info:
  features:
    - name: observation.state
      dtype: float32
      shape:
        - 24
    - name: observation.wrench
      dtype: float32
      shape:
        - 6
    - name: action
      dtype: float32
      shape:
        - 7
    - name: timestamp
      dtype: float32
    - name: frame_index
      dtype: int64
    - name: episode_index
      dtype: int64
    - name: index
      dtype: int64
    - name: task_index
      dtype: int64
  splits:
    - name: train
      num_examples: 330929
---

# EXOKERN ContactBench v0 - Peg Insertion with Force/Torque

`contactbench-forge-peginsert-v0` is the fixed-condition baseline dataset in the EXOKERN ContactBench series. It is designed for clean ablations of force/torque sensing before adding domain randomization.

## Quick Facts

| Item | Value |
| --- | --- |
| Episodes | 2,221 |
| Frames | 330,929 |
| Control frequency | 20 Hz |
| Format | LeRobot v3.0-compatible parquet |
| Robot | Franka Emika Panda |
| Simulator | NVIDIA Isaac Lab |
| Task | `Isaac-Forge-PegInsert-Direct-v0` |
| Access | Gated download on the Hugging Face Hub |

## Why This Release Exists

Most open manipulation datasets focus on geometry and vision while leaving contact dynamics implicit. v0 exists as a controlled reference dataset where the wrench signal is explicit and the environment is mostly fixed, making it easier to isolate the effect of force/torque input during policy training.

This dataset is the paired data source for [EXOKERN/skill-forge-peginsert-v0](https://huggingface.co/EXOKERN/skill-forge-peginsert-v0).

## Data Schema

| Field | Shape | Description |
| --- | --- | --- |
| `observation.state` | `(24,)` | Collapsed policy state vector exported by the collection task |
| `observation.wrench` | `(6,)` | End-effector wrench `[Fx, Fy, Fz, Mx, My, Mz]` in N / N*m |
| `action` | `(7,)` | Normalized controller action recorded during data collection |
| `timestamp` | scalar | Per-frame timestamp in seconds |
| `frame_index` | scalar | Frame index within an episode |
| `episode_index` | scalar | Episode identifier |
| `index` | scalar | Global frame index |
| `task_index` | scalar | LeRobot task mapping index |

## State Tensor Semantics

The `observation.state` vector is a 24D flat tensor exported by the FORGE collection pipeline. The current mapping is:

| Index Range | Field | Dim | Unit |
| --- | --- | ---: | --- |
| `0:3` | `fingertip_pos` | 3 | m |
| `3:6` | `fingertip_pos_rel_fixed` | 3 | m |
| `6:10` | `fingertip_quat` | 4 | unitless |
| `10:13` | `ee_linvel` | 3 | m/s |
| `13:16` | `ee_angvel` | 3 | rad/s |
| `16:22` | `force_sensor_smooth` | 6 | N / N*m |
| `22` | `force_threshold` | 1 | N |
| `23` | `ema_factor` | 1 | unitless |

`observation.wrench` duplicates `observation.state[16:22]` as a dedicated column for direct access.

## Wrench Convention

| Property | Value |
| --- | --- |
| Frame | End-effector body frame |
| Order | `[Fx, Fy, Fz, Mx, My, Mz]` |
| Force unit | Newtons |
| Torque unit | Newton-meters |

## Why v0 Is Useful

- It is the cleanest EXOKERN release for force/torque ablations.
- The collection regime is simple enough that contact-quality differences are easy to interpret.
- It pairs directly with a public baseline model repo.

Reference benchmark from [EXOKERN/skill-forge-peginsert-v0](https://huggingface.co/EXOKERN/skill-forge-peginsert-v0):

| Condition | Success Rate | Avg Contact Force (N) |
| --- | ---: | ---: |
| `no_ft` | 100.0 | 5.2 |
| `full_ft` | 100.0 | 3.2 |

![EXOKERN ContactBench v0 contact regime](https://huggingface.co/datasets/EXOKERN/contactbench-forge-peginsert-v0/resolve/main/figures/contact_regime.png)

*Figure: representative wrench trace plus episode-wise mean and peak force coverage computed from the published parquet rollout data.*

## Loading

```python
from lerobot.datasets.lerobot_dataset import LeRobotDataset

ds = LeRobotDataset("EXOKERN/contactbench-forge-peginsert-v0")
print(ds.num_episodes, len(ds))

frame = ds[0]
state = frame["observation.state"]
wrench = frame["observation.wrench"]
action = frame["action"]

print(state.shape, wrench.shape, action.shape)
```

```python
from datasets import load_dataset

ds = load_dataset("EXOKERN/contactbench-forge-peginsert-v0", split="train")
print(ds[0].keys())
```

## Related Work

- FORGE: [Force-Guided Exploration for Robust Contact-Rich Manipulation under Uncertainty](https://arxiv.org/abs/2408.04587)
- Factory: [Fast Contact for Robotic Assembly](https://arxiv.org/abs/2205.03532)
- Diffusion Policy: [Visuomotor Policy Learning via Action Diffusion](https://arxiv.org/abs/2303.04137)

## Citation

```bibtex
@dataset{exokern_contactbench_peginsert_v0_2026,
  title   = {EXOKERN ContactBench v0: Peg Insertion with Force/Torque},
  author  = {{EXOKERN}},
  year    = {2026},
  url     = {https://huggingface.co/datasets/EXOKERN/contactbench-forge-peginsert-v0},
  license = {CC-BY-NC-4.0}
}
```

## Limitations

- Simulation only. The dataset is not a real-robot capture.
- This is a mostly fixed-condition release, so it is not the right dataset for robustness claims under large physical variation.
- The task scope is a single peg-insertion setup.
- Access is gated on the Hub.

## Related Resources

- Baseline model: [EXOKERN/skill-forge-peginsert-v0](https://huggingface.co/EXOKERN/skill-forge-peginsert-v0)
- Domain-randomized successor: [EXOKERN/contactbench-forge-peginsert-v0.1.1](https://huggingface.co/datasets/EXOKERN/contactbench-forge-peginsert-v0.1.1)
- Organization page: [huggingface.co/EXOKERN](https://huggingface.co/EXOKERN)