AI-Powered Sonar Detection and Marine Conservation System

Abstract

Underwater object detection plays a crucial role in applications such as environmental monitoring, marine debris tracking, and search-and-rescue missions. However, traditional sonar systems are often expensive and complex, limiting their use in academic and field-based deployments. In this project, a cost-effective and modular sonar-based detection system is developed using the waterproof JSN-SR04T ultrasonic sensor and the ESP32 microcontroller. The system is designed to identify and classify submerged objects such as debris, boats, and human remains using AI models like YOLOv9. The captured sonar data is processed and displayed in real time on an OLED screen, while also being transmitted wirelessly via the ESP32’s Wi-Fi capabilities. The design is modeled and analyzed in ANSYS to assess mechanical stability and performance in underwater environments. Two materials, FU 4270 and FU 2451, are compared with conventional aluminum housing to evaluate structural integrity and waterproof reliability under pressure and vibration. This project demonstrates a practical, low-cost approach to underwater detection using a sonar system called JSN-SR04T, connected to an ESP32 microcontroller. It uses YOLOv9 for object classification to identify marine debris. The design includes ANSYS simulation for testing and focuses on creating a cost-effective solution with IoT integration.