RT Conference Proceedings
T1 Transfer learning or design a custom CNN for tactile object recognition
A1 Gandarias, Juan Manuel
A1 Pastor-Martín, Francisco
A1 Muñoz-Ramírez, Antonio José
A1 García-Cerezo, Alfonso José
A1 Gómez-de-Gabriel, Jesús Manuel
K1 Sensores
AB Novel tactile sensors allow treating pressure lectures as standard images due to its highresolution. Therefore, computer vision algorithms such as Convolutional Neural Networks(CNNs) can be used to identify objects in contact. In this work, a high-resolution tactilesensor has been attached to a robotic end-effector to identify objects in contact. Moreover,two CNNs-based approaches have been tested in an experiment of classification of pressureimages. These methods include a transfer learning approach using a pre-trained CNN on anRGB images dataset and a custom-made CNN trained from scratch with tactile information.A comparative study of performance between them has been carried out.
YR 2018
FD 2018-10-29
LK https://hdl.handle.net/10630/16729
UL https://hdl.handle.net/10630/16729
LA eng
NO International Workshop on Robotac: New Progress in Tactile Perception and Learning in Robotics
NO Universidad de Málaga. Campus de Excelencia Internacional Andalucía Tech. Spanish project DPI2015-65186-R, the European Commission under grant agreement BES-2016-078237, the educational project PIE-118 of theUniversity of Malaga
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 23 ene 2026