CLARU
Hannibal52Barca2026MIT

ICL Project — Manual Progress Annotations

Human progress annotations for ICL robot demonstrations, providing per-episode subtask milestones and dense per-frame progress labels for value-model agreement evaluation.

Downloads33
Episodes243

Why This Matters for Physical AI

This dataset provides human reference progress annotations that enable evaluation of value models' agreement with human judgment on task completion, supporting the development of better reward learning for robotic agents.

Technical Profile

Modalities
language
Episodes
243
Annotation Types
language_instructionsreward_labels
License
MIT
Part of the ICL Project family

Access

View on HuggingFace

Need custom language data?

Claru builds purpose-built datasets for any environment applications with dense human annotations and quality assurance.

Request a Sample Pack

Related Datasets

ICL Project — Value/Reward Model Evaluation Results

Per-frame progress predictions and evaluation outputs for seven value/reward models (IC-VFE, RECAP, Robometer, Robo-Dopamine, TOPReward, GVL, SARM) compared in the VICTR paper on the ICL demo dataset.

rgb
51 downloadsOct 2026MIT

Robo-MLT: Medical Laboratory Manipulation Datasets

Real-robot teleoperated demonstrations for two long-horizon medical laboratory workflows (blood gas analysis and tube sorting) collected on an Agilex dual-arm Static-Aloha platform with vision-language-action supervision.

rgbproprioceptionlanguage
null downloadsJul 2026apache-2.0gated

ACE-Data-0

A human-centric ambient capture dataset recording real home environments with synchronized multi-modal streams including egocentric video, exocentric video, human motion, object state, audio, and tactile sensing for embodied AI research.

rgbdepthimuaudio+3
251K downloadsOct 2026ace-data-0-research-license

In-The-Wild Humanoid Dataset

A large-scale dataset of whole-body humanoid robot learning in natural home environments, capturing human teleoperation demonstrations on Unitree G1 across real homes in Southeast Asia with varying layouts, lighting, and clutter.

rgbstereo_irproprioceptionimu+2
151K downloadsSep 2026

In-The-Wild Humanoid Dataset

A large-scale dataset of 500+ hours and 23K+ episodes of whole-body humanoid robot learning captured through human teleoperation demonstrations on Unitree G1 across real homes in Southeast Asia. The dataset includes synchronized visual observations, robot states, actions, and metadata for research on mobile manipulation, bimanual interaction, and long-horizon household skills.

rgbinfraredproprioceptionimu+2
151K downloadsSep 2026

HiFi-UMI-2K: High-Fidelity Robot-Free Manipulation Data

A large-scale dataset of 2,000 hours of high-fidelity bimanual human manipulation demonstrations collected robot-free using the HiFi-UMI capture system, designed for action-grounded supervision for manipulation policy pre-training and deployment without requiring a robot during data collection.

rgbimuproprioceptionlanguage
117K downloadsJul 2026cc-by-4.0