FEATS: Finite Element Analysis for Tactile Sensing

Project Page | Paper | Code

FEATS (Finite Element Analysis for Tactile Sensing) is a framework and dataset designed to estimate contact force distributions directly from raw images of a GelSight Mini optical tactile sensor. By using a U-net architecture trained on labels generated through Finite Element Analysis (FEA), the model can predict normal and shear force distributions from gel deformations.

Dataset Description

The dataset consists of GelSight Mini images paired with shear and normal force distribution labels inferred from FEA simulations.

• Data format: The dataset is stored as .npy files.
• Requirements: It is recommended to use numpy version 2.X or higher to load the dataset.
• Contents: The data includes training, validation, and test sets featuring raw sensor images and corresponding force maps.

For detailed instructions on data preparation, training, and evaluation, please refer to the official GitHub repository.

Citation

If you use this dataset or the associated code in your research, please cite:

@misc{helmut_dziarski2025feats,
      title={Learning Force Distribution Estimation for the GelSight Mini Optical Tactile Sensor Based on Finite Element Analysis}, 
      author={Erik Helmut and Luca Dziarski and Niklas Funk and Boris Belousov and Jan Peters},
      year={2025},
      eprint={2411.03315},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2411.03315}, 
}