Classification of materials by acoustic signal processing in real time for NAO robots

Abstract

In recent years, humanoid–biped type robots are being adopted by academic and research institutions to create an ecosystem that promotes the development of innovation and social appropriation of technology. Robot-taken decision projects on NAO robots in the field of artificial vision are more common than those ones related to auditory events for the same purpose. The main motivation for this work is to demonstrate the convenience of using an acoustic signal stage, for the classification of common wastes produced at the university. This project establishes an ideal scenario which allows taking it as reference for future projects in the acoustics field, by a methodology that enables audio signals acquisition and its real-time processing in NAO robots. This methodology consists in the robot’s programming for autonomous identification of sounds, which are sampled and saved in raw data. Then, a mathematical treatment is made in order to establish the ranges of frequency for an appropriate classification of materials. In the end, NAO robot identified and classified main wastes produced on the campus: cardboard, plastic, glass, and metal.