Video · Multimodal / AI & Machine Learning
Custom Manipulation Data Collection | Egocentric Wrist-Camera Video + IMU Recorded to Your Task Spec | LeRobot v3.0
Advertised by providerLicensing rights not independently verifiedPrice on request
Overview
Custom Manipulation Data Collection | Egocentric Wrist-Camera Video + IMU Recorded to Your Task Spec | LeRobot v3.0 is a multimodal dataset. Typical applications include LLM pre-training, alignment & RLHF and robotics & embodied AI. Relevant sectors: AI & Machine Learning, Education, Manufacturing & Industrial and Technology & IT. Pricing is not published and is provided on request.
Primary applications
Off-the-shelf manipulation datasets rarely match the exact task, objects, or environment your policy needs. We run collection on demand: you specify the manipulation task — household, workshop, or industrial hand work — and we collect it with our wrist-worn AI smartwatch, delivered in LeRobot v3.0 so your existing ingestion pipeline works unchanged. What you get per episode: egocentric wrist-camera video (640x480, 30 fps), frame-aligned 6-axis IMU state in SI units, raw 104 Hz IMU streams with per-episode alignment metadata, derived motion-intensity signals, and full LeRobot metadata. Collection is natural and uninstrumented. Participants wear the watch and do the task — no rigs, no teleoperation, no staged execution. All participants provide written consent covering commercial redistribution, and every frame is screened for faces, screens, and personal information before delivery. How it works: (1) you describe the task, object set, and target episode count; (2) we confirm feasibility, scope and price within 24 hours; (3) a pilot batch for format and quality sign-off; (4) full delivery, typically about a week from confirmation. Pricing is by quote rather than a list price. What a collection costs depends on the manipulation, the episode count, and the environment it has to be shot in, so we scope it rather than guess. Scope is confirmed before you pay. PROVEN EFFECT On DexGarmentLab (NeurIPS 2025, Isaac Sim), pretraining a Diffusion Policy's vision encoder on our wrist data
Related datasets
More in Video & Physical AI →Video · MultimodalAdvertised
Egocentric Hand Manipulation Dataset | Cloth-Folding Episodes with Wrist-Camera Video & IMU | LeRobot v3.0, Global Use
This dataset contains ~20-second episodes of a person naturally folding cloth, captured by the provider’s wrist-worn AI smartwatch.
- Language
- —
- Volume
- —
- Domain
- AI & Machine Learning +2
- Use case
- LLM pre-training
Price on request