Letters in High Energy Physics (Feb 2024)
Development and Implementation of an Autonomous Solar-Powered Drone for Disaster Relief
Abstract
Introduction: Disaster relief operations face significant challenges in assessing and responding to emergencies in hazardous environments. Traditional methods often struggle due to limited access, power constraints, and operational risks. Autonomous drones offer a promising solution, but their reliance on battery power limits their operational duration. Objectives: This study focuses on developing a novel Solar Sentinel-Disaster Drone X (SS-DDX) equipped with a solar power system and an intelligent deep learning model for efficient human detection in disaster areas. Methods: The SS-DDX utilizes an ArduPilot Mega flight controller for autonomous navigation, a 3500mAh Lithium-Ion battery for power, and a 50W flexible solar panel for extended flight endurance. The drone employs a GoPro Hero 10 camera to capture high-resolution images of disaster zones. A Relief Vision Deep Convolutional Neural Network (RVDCNN) model integrated with the Intelligent Zebra Optimization (IZO) algorithm (IZ-RVDCNN) is developed to detect humans in the captured images. Data preprocessing techniques, including Min-Max normalization and Histogram of Oriented Gradients (HOG) feature extraction, enhance model performance. Results: The IZ-RVDCNN model demonstrates high accuracy (91%), precision (91.70%), and recall (98%) in human detection, significantly outperforming existing methods. Conclusions: The SS-DDX represents a significant advancement in disaster relief operations, offering increased autonomy, extended flight duration, and enhanced human detection capabilities. The study highlights the potential of solar-powered drones and intelligent deep learning algorithms in improving response efficiency and saving lives in disaster situations.
