Drone audition: implementation of an indoor multi-drone system for sound source tracking

Abstract

In this study, we introduce an indoor multi-drone system for tracking audible sound sources. The system utilises a number of Crazyflie drones, a mini-sized, open-source drone development platform that allows rapid prototyping and development using the Robotics Operating System 2 (ROS 2). Equipped with a microphone array, sound source tracking is realised by triangulating the spectral response from the Multiple Signal and Classification (MUSIC) algorithm combined with particle filtering. We evaluate the system using real-life demonstrations, and results show the system successfully tracks a single moving sound source in real-time.