Development of a Real-Time Acoustic Detection and Tracking System for a Compact Autonomous Underwater Vehicle

Abstract

Passive acoustic detection and tracking of surface vessels from an autonomous underwater vehicle (AUV) has applications ranging from harbor security and domain awareness to scientific monitoring of vessel traffic. Prior work demonstrated the feasibility of detecting and estimating the bearing of surface vessels using a small hydrophone array and functional beamforming. This study extends that system to include data with AUV self-noise, robust multi-target bearing estimation, and closed-loop acoustic-based AUV motion control. A particle filter operating on beamformer output is developed, incorporating amplitude-based weighting, systematic resampling, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) clustering to produce bearing estimates of one or more acoustic sources at each time step. The filter is evaluated on field data collected in Norfolk, Virginia using a stationary hydrophone array, and data collected at Claytor Lake State Park with the Dragon AUV operating under its own propulsion. A rear suppression weighting tool is shown to be necessary for reliable tracking when AUV self-noise contaminates the rear quadrant of the beamformer output. A simulation environment is developed to evaluate filter performance quantitatively against known ground truths. The full closed-loop system is evaluated in three simulation scenarios of increasing complexity, demonstrating stable AUV pursuit of a single acoustic source from both ideal and non-optimal initial headings, and tracking in presence of two acoustic sources. These results confirm the feasibility of acoustically-guided AUV motion.