DLR Wheelchair Shared Control is a dataset developed by the German Aerospace Center (DLR) for human-robot shared control in wheelchair navigation. The dataset contains 570 episodes of a wheelchair robot navigating complex environments, including RGB images, depth data, and human input commands. It supports research in human-intervention guided policy learning and real-time adaptation to dynamic environments. The dataset emphasizes the integration of human feedback during robot operation, making it a valuable resource for advancing interactive and adaptive robotics research. While the dataset's license is not explicitly stated, it is primarily intended for academic use and is accompanied by evaluation scripts and pre-trained models for wheelchair navigation tasks.
The robot grasps a set of different objects in a table top and a shelf.
| Field | Value |
|---|---|
| Action Space | EEF Position |
| Control Frequency | 5 |
| Depth Cams | 0 |
| Episodes | 92233 |
| File Size | 1670.00 GB |
| Gripper | CLASH hand (https://www.dlr.de/rm/en/desktopdefault.aspx/tabid-8178/#gallery/35438) |
| Has Camera Calibration | False |
| Has Proprioception | True |
| Has Suboptimal | True |
| Language Annotations | Templated |
| Rgb Cams | 1 |
| Robot | Franka |
| Robot Morphology | Single Arm |
| Scene Type | Table Top, shelf |
| Wrist Cams | 0 |