Artificial skin systems support human–robot interactions through touch. The interpretation of touch modalities indeed represents a crucial component for the future development of robots that can properly interact with humans. Independently of the specific employed transducer, one of the key issues is how to process the massively complex and high-dimensional tactile data. In this paper, machine learning technologies (namely, support vector machines and extreme learning machines) support a pattern-recognition framework that can fully exploit the tensor morphology of the tactile signal. Furthermore, a practical strategy is provided to address the intricacies of the training procedure. Experimental results show the effectiveness of the proposed approach.
A Tensor-Based Pattern-Recognition Framework for the Interpretation of Touch Modality in Artificial Skin Systems
GASTALDO, PAOLO;PINNA, LUIGI;SEMINARA, LUCIA;VALLE, MAURIZIO;ZUNINO, RODOLFO
2014-01-01
Abstract
Artificial skin systems support human–robot interactions through touch. The interpretation of touch modalities indeed represents a crucial component for the future development of robots that can properly interact with humans. Independently of the specific employed transducer, one of the key issues is how to process the massively complex and high-dimensional tactile data. In this paper, machine learning technologies (namely, support vector machines and extreme learning machines) support a pattern-recognition framework that can fully exploit the tensor morphology of the tactile signal. Furthermore, a practical strategy is provided to address the intricacies of the training procedure. Experimental results show the effectiveness of the proposed approach.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
