Towards Real-time Underwater Object Detection and Identification: Integrating Acoustic Sensing with Edge Computing

In this work, we present a method to streamline underwater object detection, environmental monitoring, and surveillance. The proposed system employs an underwater acoustic sensor (UAS) network to detect and analyses objects within a three-dimensional underwater space, considering constraints imposed by acoustic path loss. The approach combines three key techniques–Delaunay’s convex hull-based reconstruction, the laws of magnetic equilibrium, and the Doppler effect–to improve the identification of object type, shape, location, and motion. Sensors are arranged in an optimized grid, with each grid containing eight sensors and a central insulated magnetometer to measure net magnetic field intensity. Data collected by the sensors is transmitted to a surface anchor or sink node for further processing and analysis. A prototypical deployment in an artificial pond further validates our approach, using a waterproof ultrasonic sensor, an electromagnetic coil, and a magnetometer synchronized via Arduino Uno. Real-time measurements confirm accurate detection and tracking of objects, demonstrating the effectiveness and feasibility of the proposed edge-enabled underwater monitoring framework.