ODE Robots
AI Assistant · Online
// Powered by ODE AI · Llewellyn Systems Inc
Every real-world robot generates raw sensor streams, action logs, and telemetry. RobotDAQ captures, classifies, scores, and packages that activity into structured, training-ready datasets — automatically.
Five automated stages — from raw hardware output to deployment-ready training data.
Connects to any robot via ROS2, MQTT, REST, or OPC-UA. Captures raw sensor streams, motor commands, camera frames, and telemetry continuously.
Tags every data segment by embodiment type, task class, environment context, and hardware profile — automatically, without manual labeling.
Each data segment is scored for quality: completeness, temporal alignment, sensor coverage, action diversity, and ground-truth confidence.
Exports curated datasets in HDF5, RLDS, LeRobot, or MCAP format. Includes metadata manifests, episode boundaries, and provenance records.
Push directly to Hugging Face, your private S3 bucket, or ODE's managed data vault. Versioned, signed, and API-accessible from day one.
Any team that builds or operates real robots and wants to turn runtime data into a strategic asset.
Capture fleet-wide operational data at scale. Build proprietary datasets that improve your next product generation.
Source embodied interaction data across diverse hardware. Fill the physical domain gap in your training corpus.
Turn warehouse, logistics, or manufacturing robot fleets into continuous data pipelines without custom engineering.
Publish reproducible, structured robot datasets under standardized schemas. Meet data-sharing requirements with zero overhead.
Air-gapped deployment available. Classified data paths with full audit trails. FedRAMP-aligned architecture.
Replace ad-hoc data collection scripts with a production-grade pipeline on day one. Prove data moat to investors.
Add RobotDAQ to any deployment as a managed data layer. Reseller terms available.
Every ODE robot kit can optionally stream to RobotDAQ. Contribute to the community dataset pool and earn dataset credits.
Six product tiers for every scale of robot data operation. All pricing is project-based — no surprise overages.
One-time project fee
One-time project fee
Annual license
Managed data platform — annual
Per dataset published
Per robot — add-on to any ODE kit tier
10-layer metadata schema for every dataset published through the RobotDAQ pipeline.
| Layer | Field Name | Type | Description |
|---|---|---|---|
| 01 | embodiment.type | enum | Robot morphology: wheeled, quadruped, arm, drone, humanoid, other |
| 02 | embodiment.dof | int | Degrees of freedom. Drives downstream model selection. |
| 03 | task.class | enum | Primary task category: manipulation, navigation, inspection, interaction, transport |
| 04 | task.subtask | string | Free-form task descriptor. E.g. 'box pick-and-place', 'corridor nav' |
| 05 | environment.type | enum | indoor-structured / indoor-unstructured / outdoor / simulation / lab |
| 06 | data.modalities | string[] | Active sensor streams: rgb, depth, lidar, imu, proprio, audio, force |
| 07 | data.hz | float | Primary control loop frequency in Hz |
| 08 | quality.score | float 0–1 | Composite quality score: completeness × alignment × diversity |
| 09 | license.type | enum | CC-BY-4.0 / CC-BY-NC / Research-Only / Proprietary / Custom |
| 10 | provenance.hash | sha256 | Cryptographic fingerprint of raw data package. Immutable. |
RobotDAQ Standard is an open specification. Any team may adopt it independently. Datasets certified by ODE carry a quality score badge and are listed in the RobotDAQ registry.
Every dataset is classified across four orthogonal axes for precise discovery and model routing.
| Code | Type | Typical DOF |
|---|---|---|
| WHL | Wheeled / Differential | 2–4 |
| QRD | Quadruped | 12–20 |
| ARM | Robotic Arm | 4–7 |
| DRN | Aerial / Drone | 4–6 |
| HMN | Humanoid | 24–44 |
| HEX | Hexapod | 18–24 |
| OTH | Other / Custom | Variable |
| Code | Class | Example |
|---|---|---|
| MAN | Manipulation | Pick-and-place, assembly, packing |
| NAV | Navigation | Point-to-point, mapping, patrolling |
| INS | Inspection | Visual QC, structural scan, anomaly detection |
| INT | Interaction | Human-robot handover, social nav |
| TRN | Transport | Payload delivery, warehouse AMR |
| LRN | Learning Demo | Teleoperation recordings for imitation |
| Grade | Score Range | Meaning |
|---|---|---|
| A | 0.90 – 1.00 | Publication-ready. All modalities complete, high diversity. |
| B | 0.75 – 0.89 | Training-ready. Minor gaps acceptable. |
| C | 0.55 – 0.74 | Exploratory. Good for pre-training or augmentation. |
| D | 0.00 – 0.54 | Below threshold. Not published. Flagged for review. |
| Type | Use Case | Commercial? |
|---|---|---|
| CC-BY-4.0 | Open research, academic sharing | Yes |
| CC-BY-NC | Academic only, no commercial use | No |
| Research-Only | Internal lab use, private distribution | No |
| Proprietary | Vendor lock, internal training only | Restricted |
| Custom | Negotiated terms via ODE legal | Varies |
Pilots start at $7,500. Enterprise deployments are scoped individually.
Contact us to discuss your robot fleet, data goals, and timeline.
ODE Robots
AI Assistant · Online
// Powered by ODE AI · Llewellyn Systems Inc