Update README.md
Browse files
README.md
CHANGED
|
@@ -2,7 +2,7 @@
|
|
| 2 |
title: EXOKERN
|
| 3 |
emoji: ⚡
|
| 4 |
colorFrom: blue
|
| 5 |
-
colorTo:
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
short_description: The Data Engine for Physical AI
|
|
@@ -10,42 +10,58 @@ short_description: The Data Engine for Physical AI
|
|
| 10 |
|
| 11 |
# EXOKERN — The Data Engine for Physical AI
|
| 12 |
|
| 13 |
-
**Contact-rich. Sensor-annotated. Industrially validated.**
|
| 14 |
|
| 15 |
-
We produce high-fidelity force/torque manipulation datasets for enterprise robotics, humanoid robot manufacturers, and research institutions.
|
| 16 |
|
| 17 |
## 🎯 The Problem
|
| 18 |
|
| 19 |
Over 95% of existing robotics datasets lack force/torque sensor data. Vision-only approaches fail at contact-rich tasks like insertion, assembly, and manipulation — where precise haptic feedback is critical.
|
| 20 |
|
|
|
|
|
|
|
| 21 |
## 💡 Our Solution
|
| 22 |
|
| 23 |
EXOKERN provides industrially calibrated **6-axis force/torque (wrench) data** alongside standard observations, enabling robots to learn manipulation skills that require physical contact understanding.
|
| 24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
## 📦 Datasets
|
| 26 |
|
| 27 |
| Dataset | Task | F/T Data | Status |
|
| 28 |
|---------|------|----------|--------|
|
| 29 |
| [contactbench-forge-peginsert-v0](https://huggingface.co/datasets/EXOKERN/contactbench-forge-peginsert-v0) | Peg-in-Hole Insertion | ✅ 6-axis Wrench | Available |
|
| 30 |
|
|
|
|
|
|
|
| 31 |
## 🔧 What Makes Our Data Different
|
| 32 |
|
| 33 |
- **Force/Torque annotations** on every frame (Fx, Fy, Fz, Mx, My, Mz)
|
| 34 |
- **LeRobot v3.0 compatible** — plug directly into policy training pipelines
|
| 35 |
- **Industrially relevant tasks** — insertion, assembly, contact-rich manipulation
|
| 36 |
- **Calibrated sensors** — not estimated, not vision-derived, directly measured
|
|
|
|
|
|
|
| 37 |
|
| 38 |
## 🏗️ Built With
|
| 39 |
|
| 40 |
- NVIDIA Isaac Lab for high-fidelity physics simulation
|
| 41 |
- LeRobot v3.0 format for interoperability
|
| 42 |
-
- Franka
|
|
|
|
| 43 |
|
| 44 |
## 📬 Contact
|
| 45 |
|
| 46 |
- 🌐 Website: [exokern.com](https://exokern.com)
|
| 47 |
- 🤗 HuggingFace: [huggingface.co/EXOKERN](https://huggingface.co/EXOKERN)
|
|
|
|
| 48 |
|
| 49 |
---
|
| 50 |
|
| 51 |
-
*EXOKERN —
|
|
|
|
| 2 |
title: EXOKERN
|
| 3 |
emoji: ⚡
|
| 4 |
colorFrom: blue
|
| 5 |
+
colorTo: red
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
short_description: The Data Engine for Physical AI
|
|
|
|
| 10 |
|
| 11 |
# EXOKERN — The Data Engine for Physical AI
|
| 12 |
|
| 13 |
+
**Contact-rich. Sensor-annotated. Industrially validated. EU AI Act ready.**
|
| 14 |
|
| 15 |
+
We produce high-fidelity force/torque manipulation datasets for enterprise robotics, humanoid robot manufacturers, and research institutions — with full data provenance and compliance documentation.
|
| 16 |
|
| 17 |
## 🎯 The Problem
|
| 18 |
|
| 19 |
Over 95% of existing robotics datasets lack force/torque sensor data. Vision-only approaches fail at contact-rich tasks like insertion, assembly, and manipulation — where precise haptic feedback is critical.
|
| 20 |
|
| 21 |
+
Starting August 2026, the EU AI Act requires documented data provenance for AI training data. Most existing datasets cannot meet this standard.
|
| 22 |
+
|
| 23 |
## 💡 Our Solution
|
| 24 |
|
| 25 |
EXOKERN provides industrially calibrated **6-axis force/torque (wrench) data** alongside standard observations, enabling robots to learn manipulation skills that require physical contact understanding.
|
| 26 |
|
| 27 |
+
### Capture → Anchor → Ship
|
| 28 |
+
|
| 29 |
+
| Phase | What We Do | Output |
|
| 30 |
+
|-------|-----------|--------|
|
| 31 |
+
| **01 Capture** | Operators generate specific contact forces incl. systematic failures in calibrated environments | Raw data + Failure Taxonomy + QC Report |
|
| 32 |
+
| **02 Anchor** | Real friction logs calibrate simulation. Physics parameters validated against F/T measurements | Calibrated Sim Assets (MJCF/USD/JSON) |
|
| 33 |
+
| **03 Ship** | ML-ready exports in all major formats — clean, tagged, quality-certified | RLDS, HDF5, Zarr, MCAP, LeRobot v3.0 |
|
| 34 |
+
|
| 35 |
## 📦 Datasets
|
| 36 |
|
| 37 |
| Dataset | Task | F/T Data | Status |
|
| 38 |
|---------|------|----------|--------|
|
| 39 |
| [contactbench-forge-peginsert-v0](https://huggingface.co/datasets/EXOKERN/contactbench-forge-peginsert-v0) | Peg-in-Hole Insertion | ✅ 6-axis Wrench | Available |
|
| 40 |
|
| 41 |
+
> 🗺️ *More contact-rich datasets in development — including assembly, grasping, and deformable object manipulation. Commercial Contact Skill Packs available for enterprise customers.*
|
| 42 |
+
|
| 43 |
## 🔧 What Makes Our Data Different
|
| 44 |
|
| 45 |
- **Force/Torque annotations** on every frame (Fx, Fy, Fz, Mx, My, Mz)
|
| 46 |
- **LeRobot v3.0 compatible** — plug directly into policy training pipelines
|
| 47 |
- **Industrially relevant tasks** — insertion, assembly, contact-rich manipulation
|
| 48 |
- **Calibrated sensors** — not estimated, not vision-derived, directly measured
|
| 49 |
+
- **Full data provenance** — capture methodology, operator ID, QC reports, EU AI Act compliant
|
| 50 |
+
- **Sim-anchoring** — calibrated simulation assets bridge the Sim2Real gap
|
| 51 |
|
| 52 |
## 🏗️ Built With
|
| 53 |
|
| 54 |
- NVIDIA Isaac Lab for high-fidelity physics simulation
|
| 55 |
- LeRobot v3.0 format for interoperability
|
| 56 |
+
- Franka FR3 (7-DOF) robot platform
|
| 57 |
+
- Bota Systems SensONE 6-axis F/T sensor
|
| 58 |
|
| 59 |
## 📬 Contact
|
| 60 |
|
| 61 |
- 🌐 Website: [exokern.com](https://exokern.com)
|
| 62 |
- 🤗 HuggingFace: [huggingface.co/EXOKERN](https://huggingface.co/EXOKERN)
|
| 63 |
+
- 📧 Enterprise inquiries: info@exokern.com
|
| 64 |
|
| 65 |
---
|
| 66 |
|
| 67 |
+
*EXOKERN — Bridging the Haptic Data Gap in Physical AI*
|