CLARU
paulpacaudapache-2.0

Guardian: RLBench-Fail

Annotated robotic manipulation failure dataset generated in RLBench simulator for training Vision-Language Models on failure detection in planning and execution tasks. Contains 52 manipulation tasks with procedurally generated planning and execution failures.

Downloads74

Technical Profile

Modalities
rgblanguage
Robot Embodiments
robotic arm with gripper
Environment
simulation
Task Types
manipulationgraspingpick_and_placefailure_detectionvisual_question_answering
Data Format
JSONL
License
apache-2.0
Part of the Guardian: RLBench-Fail family

Access

View on HuggingFace

Need custom rgb data?

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

Request a Sample Pack

Related Datasets

Guardian: RLBench-Fail

A dataset of annotated robotic manipulation failures from the RLBench simulator, containing 12,358 execution failure samples and 5,808 planning failure samples across 52 tasks, designed for training Vision-Language Models on failure detection.

rgblanguage
278 downloadsFeb 2026apache-2.0

BuggyBrain

A robotics dataset created using LeRobot containing 64 episodes of robot manipulation data with multi-camera observations and joint position actions.

rgbproprioception
null downloadsApr 2026apache-2.0

UMI-3D Dataset (Curtain)

Demonstration dataset for the UMI-3D curtain manipulation task, containing RGB images and robot state paired with end-effector and gripper trajectories.

rgbproprioception
null downloadsApr 2026MIT

UMI-3D Dataset (Cup)

Demonstration data for cup manipulation task in UMI-3D, containing RGB observations, end-effector trajectories, gripper states, and action sequences.

rgbproprioception
null downloadsApr 2026mit

CRCD-dVRK-LeRobot

18 episodes of robot-assisted cholecystectomy on pig liver using the da Vinci Research Kit (dVRK) in LeRobot v3.0 format. Converted from the CRCD dataset with stereo endoscopic video and kinematic state/action data.

rgbproprioception
null downloadsApr 2026cc-by-4.0

Xperience-10M

A large-scale egocentric multimodal dataset of human experience containing 10 million interactions and 10,000 hours of synchronized first-person recordings with six video streams, audio, stereo depth, camera pose, hand mocap, full-body mocap, IMU, and hierarchical language annotations for embodied AI, robotics, and world modeling research.

rgbaudiodepthproprioception+3
2.3M downloadsMar 2026other