shareef14MIT

Robot Zoo

An open-source multi-embodiment robotics dataset for cross-embodiment research built with Isaac Lab and Isaac Sim, containing 13,058 episodes across three robot platforms performing manipulation and navigation tasks.

Downloads3
Episodes13058

Why This Matters for Physical AI

This dataset enables research into cross-embodiment imitation learning and transfer by providing standardized task demonstrations across different robot morphologies (fixed-arm, mobile manipulators) in simulation, revealing how embodiment affects task executability and strategy.

Technical Profile

Modalities
proprioceptionstate
Robot Embodiments
Franka PandaStretch 3Fetch
Action Space
joint_positions
Environment
simulation
Task Types
pick_and_placegraspingopeningreachingpushingmanipulation
Episodes
13058
Data Format
npz
Annotation Types
reward_labelsaction_labels
License
MIT
Part of the Robot Zoo family

Access

Need custom proprioception data?

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

Request a Sample Pack

Related Datasets