Coordinated Two-Hand Data for Dual-Arm Robots
Coordinated two-hand demonstrations with per-hand action and contact labels for dual-arm and humanoid manipulation. Discuss bimanual task design with Shaip.
Dataset Specifications
| Target Buyer Teams | Humanoid Manipulation Lead; Dual-Arm Robotics Team; Dexterity Research; VLA Data Lead |
| Representative Tasks | Handoffs, folding, packaging, cable handling, assembly and coordinated carrying |
| Capture / Modalities | Synchronized ego/exo RGB-D; left/right hand or arm trajectories; optional tactile data |
| Annotations / Metadata | Per-hand actions, object state, coordination phase, contact events, success and recovery |
| Delivery Formats | MP4/RGB-D + trajectory JSON/Parquet; LeRobot/RLDS for robot episodes only |
| Scale (Illustrative)* | 300–1,500 demonstration hours |
| Participants / Operators | 100–400 participants |
| Tasks / Scenarios | 25–100 tasks |
| Objects / Sites / Views | 50–300 objects |
| Recommended Engagement Scope | 75–200 hours across 10–20 tasks and 25–75 objects |
A standardized extension of Shaip's VR motion and synchronized multi-view capture capabilities.
View the related case study →Where This Dataset Is Used
This bimanual manipulation dataset is designed for teams building dual-arm robots, humanoid platforms and VLA foundation models that require coordinated two-hand skills. Robotics companies and embodied AI labs use it as a dual arm robot dataset for imitation learning, policy training and benchmarking tasks such as folding, handover, insertion and tool passing. If you’re sourcing two-hand manipulation data, coordinated manipulation demonstrations or bimanual robot training data with per-hand action and contact labels, this VLA manipulation dataset supports training from pilot scale to production.
Request a Sample of the Bimanual Manipulation Demonstration Dataset
Tell us about the robot product or Physical AI system you’re building and we’ll respond with sample access, specifications and a proposal matched to your task.
"*" indicates required fields
