Underwater Surveillance Robot

Abstract

An underwater robot is a specialized device designed to collect and monitor data in underwater environments. These robots have various applications, including oceanography research, search and rescue missions, underwater archaeology, monitoring marine life and habitats. This paper aims to develop a robotic fish which detects the objects using YOLOv8 model. It swims and floats with the help of propellers and peristaltic pump present inside the model. The robotic fish monitors the underwater conditions like temperature, and pressure. Using YOLOv8 model, the robot detects seven marine creatures like fishes, penguins, star fish, shark, puffin, stingray, and jelly fish. The precision for all fishes lies between 0.1 to 0.4 except for the jelly fish and penguin. The maximum value of recall of all fishes lies in range of 0.6 to 0.8. Puffin fish has the lowest precision. The proposed model detects marine creatures with precision of 0.78, recall of 0.65 and map value of 0.72.