mulligan2026MIT
Thread Nut: Round 1 Evaluation
Real-robot held-out evaluation dataset for the Mulligan paper, containing episodes from evaluation sessions with multiple policy checkpoints (HG-DAgger, HG-DAgger+Mulligan, HG-DAgger+Mulligan no-CF) tested on thread-nut tasks.
Downloads979
Episodes150
Hours0.649
Technical Profile
- Modalities
- rgb
- Robot Embodiments
- franka
- Action Space
- 7-DoF: vel_x, vel_y, vel_z, vel_roll, vel_pitch, vel_yaw, gripper_action
- Environment
- lab
- Task Types
- manipulation
- Episodes
- 150
- Total Hours
- 0.649
- Data Format
- parquet
- License
- MIT
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack