larsvandorpapache-2.0
Ψ₀ SIMPLE Evaluation Rollouts (G1 locomotion-pick)
Closed-loop evaluation rollout videos of the Ψ₀ (Psi-0) humanoid vision-language-action policy run in the SIMPLE simulator on a Unitree G1 robot, consisting of 10 evaluation clips on a locomotion-pick task with 4 successful and 6 failed episodes.
Downloads79
Episodes10
Why This Matters for Physical AI
This dataset provides evaluation rollout data for a vision-language-action humanoid policy, demonstrating closed-loop performance on complex locomotion and manipulation tasks in simulation, which is relevant for understanding generalization and robustness challenges in embodied AI systems.
Technical Profile
- Modalities
- rgblanguage
- Robot Embodiments
- humanoidUnitree G1
- Action Space
- end_effector_delta
- Environment
- simulation
- Task Types
- locomotionpick_and_placemanipulation
- Episodes
- 10
- Data Format
- video
- Annotation Types
- reward_labelslanguage_instructions
- License
- apache-2.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.
Request a Sample Pack