Moving sound source localization and tracking for an autonomous robot equipped with a self-rotating bi-microphone...

Area of Science:

• Acoustics
• Signal Processing
• Robotics

Background:

• Accurate localization and tracking of sound sources in three-dimensional (3D) space are crucial for various applications, including robotics and audio analysis.
• Traditional methods often face challenges with moving sources and complex environments.

Purpose of the Study:

• To develop and evaluate novel methods for localizing and tracking a sound source moving in 3D space.
• To address the complexities introduced by a rotating bi-microphone system and a moving sound source.

Main Methods:

• Two distinct approaches were developed: one utilizing extended Kalman filters (EKFs) based on four state-space models, and another employing a Hilbert-transform based method.
• The bi-microphone system captures sound, generating a sinusoidal inter-channel distance difference (ICDD) signal influenced by source and array motion.
• EKFs identify the amplitude and phase of the ICDD signal, while the Hilbert method compares the analytic ICDD signal to a reference signal for localization.

Main Results:

• Observability analyses were performed on the state-space models to understand EKF performance limitations.
• The Hilbert-transform method, combined with a moving average filter, effectively reduces noise and artifacts in sound source localization.
• Simulation studies demonstrated the effectiveness of both proposed methods in comparison studies.

Conclusions:

• The presented extended Kalman filter and Hilbert-transform based methods offer viable solutions for 3D sound source localization and tracking.
• The study highlights the importance of analyzing state-space model observability for filter design.
• These techniques show promise for enhancing audio-based perception systems.