Acoustic Vector Sensor-Based UAV Sound Source Localization via Covariance Enhancement and Confidence Guidance...

Area of Science:

• Acoustics
• Signal Processing
• Robotics and Autonomous Systems

Background:

• Unauthorized unmanned aerial vehicle (UAV) intrusions necessitate accurate detection and localization.
• Conventional acoustic sensing methods using microphone arrays are limited by the need for numerous elements and large apertures.

Purpose of the Study:

• To propose a compact, passive, and low-cost acoustic sensing method for UAV detection and localization.
• To enhance the accuracy and robustness of acoustic-based UAV surveillance systems.

Main Methods:

• Development of the Covariance Enhancement and Confidence-guided Tracking for 3D Acoustic Localization (CECT-3DAL) method using a single acoustic vector sensor (AVS).
• Implementation of adaptive diagonal loading for covariance matrix robustness at low SNR.
• Application of exponential spectral enhancement and eigenvalue-ratio-based confidence for improved direction estimation and angle sequence smoothing.
• Utilizing a dual-sensor geometric model for a closed-form 3D solution.

Main Results:

• Simulations showed azimuth and elevation root-mean-square errors (RMSEs) below 1.5° for SNR above 4 dB.
• Anechoic chamber tests demonstrated confidence-weighted smoothing reduced standard deviations from 4.34°/2.63° to 1.46°/0.86°.
• Field experiments achieved an average horizontal RMSE of 0.209 m against GPS, with continuous and smooth trajectory tracking.

Conclusions:

• The proposed AVS-based CECT-3DAL method provides a compact, passive, and low-cost solution for counter-UAV acoustic surveillance.
• The method significantly improves the accuracy and robustness of UAV localization compared to conventional techniques.
• CECT-3DAL offers a viable alternative for enhancing security in sensitive areas against UAV threats.