myconnects2024MIT

MotIF-Actions

Action-space companion dataset to MotIF-1K that provides proprioceptive joint states and end-effector trajectories for robot teleoperation demonstrations on the Stretch mobile manipulator. Contains 237 trajectories with joint state logging, 202 with task/motion language labels, and 175 with 2D end-effector pixel coordinates.

Downloads0
Episodes237

Why This Matters for Physical AI

This dataset enables training robot policies with grounded action spaces and language instructions, combining proprioceptive feedback with visual observations for improved imitation learning on mobile manipulation tasks.

Technical Profile

Modalities
rgbproprioceptionlanguage
Robot Embodiments
Stretch
Action Space
joint_deltas
Task Types
manipulationmobile_manipulation
Episodes
237
Data Format
RLDS
Annotation Types
language_instructionsaction_labels
License
MIT
Part of the MotIF family

Access

Need custom rgb data?

Claru builds purpose-built datasets for any environment applications with dense human annotations and quality assurance.

Request a Sample Pack

Related Datasets