CLARU
RoMALabMIT

Video2VLA RoboTwin Gemini subtask plans (15 tasks)

Dataset containing 7,500 Gemini Robotics subtask plans for 15 RoboTwin tasks, with 500 unique episode plans per task (450 train, 50 validation).

Downloadsnull
Likes1

Technical Profile

Modalities
language
Environment
simulation
Task Types
manipulationpick_and_place
Data Format
tar
License
MIT
Part of the Video2VLA RoboTwin Gemini subtask plans (15 tasks) family

Access

View on HuggingFace

Need custom language data?

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

Request a Sample Pack

Related Datasets

LIBERO datasets pre-encoded for LingBot-VA

Three community-preprocessed LIBERO suites (spatial, goal, object) optimized for the LingBot-VA pipeline with precomputed dual-camera visual latents and LeRobot v2.1 formatting.

rgblanguage
null downloadsJul 2026cc-by-4.0

RoboEval → openpi LeRobot dataset (roboeval_all)

Vision-language-action dataset for fine-tuning π0.5 on the RoboEval bimanual-manipulation benchmark, containing 4135 episodes of rendered observations from lightweight demos with a camera.

rgbproprioceptionlanguage
null downloadsJul 2026apache-2.0

Kinetic

A lightweight modular Vision-Language-Action (VLA) system for robotics that uses specialized modules communicating through a World Model rather than a single end-to-end transformer.

rgblanguage
null downloadsJul 2026apache-2.0

SO101 Bench Sim 4 — Sim-Real Correspondence Dataset

A curated simulated robot-manipulation dataset for the SO-101 Bench sim-real correspondence study, containing tabletop object manipulation trajectories with language annotations.

rgbproprioceptionlanguage
null downloadsJul 2026

MiniCPM-RobotManip LIBERO

The four LIBERO suites converted to LeRobot v3 format for vision-language-action model training with MiniCPM-RobotManip. Contains 1,636 simulation episodes across four benchmark suites (LIBERO-10, LIBERO-Goal, LIBERO-Object, LIBERO-Spatial).

rgblanguage
null downloadsJul 2026apache-2.0

OmniAction

A large-scale multimodal dataset for proactive robot manipulation comprising 141,162 episodes with cross-modal contextual instructions derived from spoken dialogue, environmental sounds, and visual cues rather than explicit commands.

rgbaudiolanguage
162K downloadsMar 2026cc-by-nc-4.0