Everloom2025MIT

Active Perception AAWR Dataset in GRASP Lab Mock Kitchen

A real-world reinforcement learning dataset of active perception behaviors collected in a mock kitchen environment with 4 scenes. The dataset contains 13.5GB of robot demonstrations in DROID format with cleaned, idle-frame-filtered episodes.

Downloads7K

Why This Matters for Physical AI

This dataset demonstrates real-world active perception learning through reinforcement learning, showing how robots can learn to effectively control their viewpoints and sensing strategies to accomplish manipulation tasks in realistic kitchen environments.

Technical Profile

Modalities
rgbproprioception
Robot Embodiments
xarm
Action Space
joint_positions
Environment
kitchenlab
Task Types
manipulationactive_perceptionlifting
Data Format
DROID
License
MIT
Part of the AAWR family

Community Signals

Access

Need custom rgb data?

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

Request a Sample Pack

Related Datasets