Frequency-aware decomposition learning for sensorless wrench estimation in vibration-rich robotic contact

Supplementary material for the paper 'Frequency-aware decomposition learning for sensorless wrench estimation in vibration-rich robotic contact'. The study addresses sensorless wrench estimation over a multi-step-ahead horizon in a vibration-rich grinding task. This archive includes (1) a preprocessed robot dataset collected from teleoperated grinding of gypsum blocks, (2) a preprocessed subset of the public RH20T dataset (Fang et al., 2024) used for the transfer learning study and subject to its original license, and (3) the trained model checkpoints that reproduce tables and figures in the paper. The preprocessing and evaluation code is available in the accompanying code repository (https://github.com/leehyeonbeen/FDN).