Design of an Amphibious Drone based on Monocular Vision for Lake Monitoring and Rescue
Qiancheng Ma, Jundong Qi · Advances in computer science research · 2024
The freshwater available for direct human use is extremely scarce compared to the total amount of water on Earth.Most of this freshwater is located in freshwater lakes.As human civilization progresses, the ecological destruction of freshwater lakes has become increasingly common.Drones have become ideal tools for lake monitoring due to their high mobility and wide coverage.However, these drones mostly do not directly contact the water, which introduces some limitations.To overcome these limitations and improve lake monitoring efficiency, this paper proposes an amphibious drone.This drone directly contacts the water body.It is equipped with monocular vision sensors and water quality sensors.The study validates the feasibility of the floating structure by comparing the displacement to the displacement mass ratio.The feasibility of the vision and water quality modules is verified through practical testing.The study combines the YOLO algorithm with optimized neural network algorithms and the inverse concentration gradient algorithm.The monocular vision sensor and water quality sensor achieve precise positioning and dynamic tracking of pollutants.This study validates the superiority of amphibious drones in monitoring freshwater lakes.It provides a new solution for future protection and monitoring of freshwater lakes.