Comparing Data Representation Techniques for Tactile Sensing in Classification Tasks

The present paper explores the importance of utilizing statistical and frequency-based data pre-processing techniques in tactile sensing for robotic applications. The study compares the performance of raw data-based classification with these techniques and highlights the limitations of previous studies that overlook their benefits. The paper also discusses multimodal tactile sensing and relevant literature on statistical and frequency-based data preprocessing techniques applied to time series data from sensors. The aim is to enhance tactile sensing systems' classification accuracy and robustness for more efficient and effective robotic applications.