Computational intelligence and mechatronics solutions for robotic tactile object recognition

Abstract

The paper discusses signal processing and mechatronics solutions for tactile object recognition using robotic hands. The two important aspects analyzed are the recognition of objects from tactile displacement profiles obtained by force sensing transducers and the recognition of textures using a rubbing motion executed by a robotic finger equipped with a dynamic tactile fingertip. Neural network solutions are used as an intelligent approach to recognize symbols, such as embossed numbers and letters, as well as for the recognition of different texture profiles. The conception issues, the challenges as well as the experimental results obtained are discussed from both a biological and a technological perspective.