Published July 24, 2025
| Version 1.0.0
Dataset
Open
RH5 Dataset for Hybrid State Estimation
Authors/Creators
-
1.
University of Bremen
-
2.
German Research Center for Artificial Intelligence
-
3.
German Research Centre for Artificial Intelligence
Description
The dataset was used for training and benchmarking of state estimation algorithms. It consists of data from various motions of the RH5 humanoid robot (DFKI), such as static walking, squatting, balancing on a single leg, and waist movements along the sagittal and frontal axes. It has the following properties:
- 30 trajectories from the real RH5 robot
- 14 trajectories from the simulated RH5 robot
- 575258 samples
- 50 minutes
Details about how to load the dataset and the trajectories are given in a README.md file.