Neural Radiated-Noise Fields for Unmanned Underwater Vehicle Noise Spectrum Prediction in Three-Dimensional Scenes
Researchers have developed a novel method called Neural Radiated-Noise Fields (NRNF) to predict the acoustic signatures of unmanned underwater vehicles (UUVs). This approach models the noise spectrum as a continuous function of spatial position, UUV orientation, and frequency, overcoming limitations of traditional physics-based methods. The NRNF model achieved an average prediction error of 3.5 dB in the 50 to 5000 Hz band, demonstrating improved stability and generalization through the use of a learnable three-dimensional scene feature grid. AI
IMPACT This AI-driven approach could enhance UUV performance evaluation and acoustic signature characterization.