Sample Lab

Interactive Robot Data Samples

Inspect a real teleoperation episode end-to-end: synchronized cameras, action vs. state curves, end-effector trajectory, force/torque, language annotations, and dataset health — before you ever pull a byte of training data.

Download Research Preview24 MB · 11 samplesSelf-contained zip: every sample bundle, a data card, license, and a Python loader stub. Loads in ~5 s, no auth, no signup.
SignIQ · curated sampleInteractive

SignIQ ALOHA Bimanual — Coffee Prep

Real bimanual coffee-prep episode at 50 Hz with four synchronized camera views (overhead, low-angle, and two wrist cameras), full leader/follower joint streams, and per-frame action targets.

Robot
ALOHA dual-arm (14-DOF leader/follower bimanual)
Task family
Bimanual tabletop manipulation
License
Commercial — sample for inspection
Format
LeRobot, HDF5, MCAP
video × 4 cams14-D joint state14-D actiontask instruction
Best for: ACT · Diffusion Policy · openpi · VLA fine-tuning
SignIQ · curated sampleInteractive

SignIQ Humanoid — Warehouse Manipulation

Real humanoid warehouse episode with dual-camera capture, 19-D body proprioception, and a 40-D whole-body action vector ready for GR00T or openpi humanoid fine-tuning.

Robot
H1-class humanoid (whole-body upper torso, dual 7-DOF arms)
Task family
Humanoid whole-body manipulation
License
Commercial — sample for inspection
Format
LeRobot, GR00T-LeRobot, HDF5
video × 2 cams (left/right)19-D body state40-D actiontask instruction
Best for: GR00T · openpi humanoid · Whole-body policy · Helix-style
SignIQ · curated sampleInteractive

SignIQ Collaborative Dual-Arm — Contact-Rich Threading

Bimanual contact-rich threading episode with four-view capture and high-frequency action streams. Demonstrates the schema we ship for assembly and dexterous manipulation collections.

Robot
Collaborative dual-arm (2× 7-DOF + parallel grippers)
Task family
Contact-rich dual-arm threading
License
Commercial — sample for inspection
Format
LeRobot, HDF5, MCAP
video × 4 cams14-D joint state14-D actiontask instruction
Best for: ACT · Diffusion Policy · GR00T · Dexterous BC
SignIQ · curated sampleInteractive

SignIQ Collaborative Single-Arm — Pick & Place

Single-arm pick-and-place episode with three synchronized cameras (scene + hand-mount + side), proprioceptive state, and language instruction — ideal for OpenVLA-style fine-tuning.

Robot
Collaborative single-arm (7-DOF + parallel gripper)
Task family
Single-arm tabletop pick-and-place
License
Commercial — sample for inspection
Format
LeRobot, HDF5
video × 3 cams8-D joint state7-D actiontask instruction
Best for: openpi · Diffusion Policy · OpenVLA · VLA fine-tuning
SignIQ · curated sampleInteractive

SignIQ Sim — Bimanual Cube Transfer

Simulation-paired bimanual transfer episode. Same embodiment tag and modality map as our real ALOHA bundle — drop into sim/real co-training without schema reconciliation.

Robot
ALOHA bimanual (simulated, 14-DOF leader/follower)
Task family
Simulation-paired bimanual manipulation
License
Commercial — sample for inspection
Format
LeRobot, GR00T-LeRobot, HDF5
video × 1 cam14-D joint state14-D actionmodality.json
Best for: Sim-to-real · Domain randomization · GR00T · Diffusion Policy
SignIQ · curated sampleInteractive

SignIQ Mobile ALOHA — Elevator Navigation

Mobile bimanual robot performing a long-horizon elevator-traversal task: dual-arm interaction with elevator buttons combined with mobile-base navigation.

Robot
Mobile ALOHA (mobile base + dual 7-DOF arms + parallel grippers)
Task family
Mobile manipulation with whole-body navigation
License
Commercial — sample for inspection
Format
LeRobot, HDF5, MCAP
video × 3 cams14-D joint state + base velocity14-D actiontask instruction
Best for: Mobile manipulation policies · Long-horizon BC · GR00T mobile
SignIQ · curated sampleInteractive

SignIQ Egocentric Human — Fold Laundry

First-person human capture during a laundry-folding session — different scene, different object class (deformable cloth vs. rigid box), same hand-pose schema as the pick-and-pack sample. Useful for testing generalization across object categories.

Robot
Egocentric human capture (head-mounted camera + 150-D hand pose + IMU)
Task family
Egocentric human task footage for robot learning
License
Commercial — sample for inspection (consent + face blur)
Format
LeRobot, MP4 + JSON, Parquet
video × 1 (1080p egocentric)150-D hand pose40-D finger angles26-D gestures6-DoF IMU
Best for: VLA reasoning pretraining · World models · Affordance priors · Cloth manipulation transfer
SignIQ · curated sampleInteractive

SignIQ Egocentric Human — Fold Laundry (quick session)

Compact 18-second laundry-folding clip — useful for quick subtask-boundary tests and short-horizon affordance experiments where the long-form sample would be too dense.

Robot
Egocentric human capture (head-mounted camera + 150-D hand pose + IMU)
Task family
Egocentric human task footage for robot learning
License
Commercial — sample for inspection (consent + face blur)
Format
LeRobot, MP4 + JSON, Parquet
video × 1 (1080p egocentric)150-D hand pose40-D finger angles26-D gestures6-DoF IMU
Best for: Short-horizon affordance learning · Subtask boundary detection · Hand-pose pretraining
SignIQ · curated sampleInteractive

SignIQ Policy Rollouts — Single-Arm with Reward Labels

Policy-rollout episode at 5 Hz with `next.reward` per frame — illustrates the schema we use when capturing rollouts from your deployed policy in our cell, including success/failure labeling and operator-rated severity.

Robot
Franka-class single-arm (24-D state, 7-D action, dual cam)
Task family
Scripted policy rollouts with per-step reward
License
Commercial — sample for inspection
Format
LeRobot, HDF5
video × 2 cams (scene + wrist)24-D state7-D actionper-step rewardoutcome label
Best for: Post-training · Evaluation suites · Recovery skill learning · Reward-aware BC
SignIQ · curated sampleInteractive

SignIQ Egocentric Human — Pick-and-Pack Activity

First-person human capture during a pick-and-pack activity, with full hand articulation (150-D pose + 40-D finger angles), IMU, and auto-derived motion-energy spans. Consent-cleared, face-blurred.

Robot
Egocentric human capture (head-mounted camera + 150-D hand pose + IMU)
Task family
Egocentric human task footage for robot learning
License
Commercial — sample for inspection (consent + face blur)
Format
LeRobot, MP4 + JSON, Parquet
video × 1 (1080p egocentric)150-D hand pose40-D finger angles26-D gestures6-DoF IMU
Best for: VLA reasoning pretraining · World models · Affordance priors · Human-to-robot transfer
SignIQ · curated sampleInteractive

SignIQ Humanoid — Multi-Agent Coordination

Two humanoids performing a coordinated greeting motion — capture pipeline showcases retargeted human motion priors and synchronized multi-agent state streams.

Robot
H1-class humanoid pair (whole-body upper torso, dual arms)
Task family
Humanoid coordination & retargeted motion
License
Commercial — sample for inspection
Format
LeRobot, GR00T-LeRobot, HDF5
video × 2 camswhole-body proprioceptionretargeted actionmulti-agent context
Best for: GR00T · Whole-body humanoid · Retargeted motion priors · Multi-agent coordination