GeorgiaTechother

bones_seed_sonic_full_129785 Reference Arrays

Training-shaped reference arrays for humanoid motion tracking derived from motion-capture data, containing 129,785 trajectories with 47,491,234 transitions in optimized tensor layout for imitation learning.

Downloads0
Episodes129785

Why This Matters for Physical AI

This dataset provides pre-processed reference arrays for humanoid motion imitation learning, enabling efficient training of embodied AI systems through optimized tensor layouts and motion-capture-derived trajectories.

Technical Profile

Modalities
proprioceptionmotion_capture
Robot Embodiments
humanoid
Action Space
joint_positions
Environment
simulation
Task Types
motion_trackingimitation_learning
Episodes
129785
Data Format
zarr
License
other
Part of the GeorgiaTech/g1_bones_seed_sonic_129k_50hz 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