haw-ai-i2026cc-by-4.0

Fiatlux policy baselines

Recorded and scored runs of reference policies on twelve Fiatlux humanoid subtasks, including zero-action, random-action, and GR00T N1.7 zero-shot baselines with benchmark scorer outputs.

Downloads56
Episodes12

Why This Matters for Physical AI

Provides standardized benchmark baselines and scoring infrastructure for evaluating humanoid control policies on diverse manipulation and locomotion subtasks.

Technical Profile

Modalities
rgbproprioceptionforce_torque
Robot Embodiments
humanoid
Action Space
joint_positions
Environment
simulation
Task Types
manipulationlocomotion
Episodes
12
Data Format
HDF5
Annotation Types
reward_labelsaction_labels
License
cc-by-4.0
Part of the Fiatlux family

Access

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