Real-time neuro-inspired sound source localization and tracking architecture applied to a robotic platform

Abstract

This paper proposes a real-time sound source localization and tracking architecture based on the abilityof the mammalian auditory system using the interaural intensity difference. We used an innovative bin- aural Neuromorphic Auditory Sensor to obtain spike rates similar to those generated by the inner haircells of the human auditory system. The design of the component that obtains the interaural intensitydifference is inspired by the lateral superior olive. The spike stream that represents the IID is used toturn a robotic platform towards the sound source direction. The architecture was implemented on FPGAdevices using general purpose FPGA resources and was tested with pure tones (1-kHz, 2.5-kHz and 5-kHzsounds) with an average error of 2.32 °. Our architecture demonstrates a potential practical application of sound localization for robots, and can be used to test paradigms for sound localization in the mammalianbrain.