DSP-based audio processing for controlling a mobile robot using a spherical microphone array

Abstract

Service robots are designed to assist humans in a wide variety of tasks. In order to do this, it may be particularly useful to provide the robot with spatial hearing - the ability to receive commands and understand where they are coming from. This study suggests using a spherical microphone array as the basis for a modular system for spatial audio processing which can be fitted on mobile (or static) service robots. The study relates to a recently developed prototype robot fitted with a low-order spherical microphone array. The robot is a small-scale table-top robot based around an Arduino uno micro-controller. Planar movement is achieved using a two-wheel drive and steering mechanism. Elevation is achieved using a robotic arm. The goal of the study is to reach full spatial control of the robot through spatial audio commands. Localization includes direction and distance using near-field and far-field algorithms. Several approaches to system design are discussed, and a DSP-based stand-alone system is presented. A first-order spherical microphone array was built and various source types were tested. Issues such as reverberation and multiple source tracking are also addressed.