RoboMIND 2.0: A Large-Scale Dataset for Bimanual Mobile Manipulation
Published:Dec 31, 2025 05:59
•1 min read
•ArXiv
Analysis
This paper addresses the limitations of current robotic manipulation approaches by introducing a large, diverse, real-world dataset (RoboMIND 2.0) for bimanual and mobile manipulation tasks. The dataset's scale, variety of robot embodiments, and inclusion of tactile and mobile manipulation data are significant contributions. The accompanying simulated dataset and proposed MIND-2 system further enhance the paper's impact by facilitating sim-to-real transfer and providing a framework for utilizing the dataset.
Key Takeaways
- •Presents RoboMIND 2.0, a large-scale real-world dataset for bimanual and mobile manipulation.
- •Includes tactile-enhanced and mobile manipulation trajectories.
- •Provides a simulated dataset for sim-to-real transfer.
- •Proposes MIND-2 system, a hierarchical framework for utilizing the dataset.
Reference
“The dataset incorporates 12K tactile-enhanced episodes and 20K mobile manipulation trajectories.”