PranayTest2026apache-2.0
US ER2 policy evals — ABC vs π0.5
Policy rollouts from a bimanual AgileX PiPER robot comparing ABC and π0.5 policies on a plate handover task, with detailed telemetry and execution metrics from 42 evaluation runs.
Downloads39
Episodes42
Why This Matters for Physical AI
This dataset provides real-world bimanual robot telemetry and policy execution traces comparing different training approaches (ABC vs π0.5) with original and time-compressed timing, enabling analysis of how timing affects learned policies on complex manipulation tasks.
Technical Profile
- Modalities
- proprioceptionjoint_positionsjoint_velocitiesjoint_currents
- Robot Embodiments
- AgileX PiPERbimanual
- Action Space
- joint_positions
- Environment
- lab
- Task Types
- bimanual_handovermanipulation
- Episodes
- 42
- Data Format
- parquet
- Annotation Types
- action_labelsreward_labels
- License
- apache-2.0
Access
Need custom proprioception data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack