I am a strong believer in open-source and data-sharing for the advancement of science. Whenever possible, I am releasing datasets and code that allow the recreation of results.
Sewer Terrain Inspection Knowledge (STINK)
![](https://hendrik.kolveng.com/wp-content/uploads/2020/02/deployed_in_sewer-1-649x1024.jpg)
This dataset contains 625 inspection points taken by the legged robot ANYmal on the floor of a concrete sewer. The floor was probed by a haptic, ‘scratching’, motion and 18 sensor signals (Force/Torque and two Inertial Measurement Units) were recorded. Together with ground-truth labels for the state of concrete degradation, supplied by professional inspectors, the data was used to train a support vector machine. The Matlab scripts are included in this package.
The full dataset can be downloaded here (Stink.zip)
Planetary Soil Impact Dataset (PALPATE)
![](https://hendrik.kolveng.com/wp-content/uploads/2020/04/image.png)
This dataset is recorded to demonstrate the applicability of machine learning methods to classify the complex feet-soil interaction on fine-grained soils. The dataset consists of 2600 feet-soil impacts, which were automatically executed and recorded with a specially designed testbed. Impacts were performed on a variety of Martian soil simulants, such as ES-1, ES-2, ES-3 and other. The dataset includes the sensor signals acquired by two different feet and sensors on the testbed. Additionally, a small dataset with 240 impacts created by the quadruped robot ANYmal is included.
The full dataset can be downloaded here (Palpate.zip)