SENSOR FUSION TECHNIQUES FOR DRONES IN IOT-BASED SURVEILLANCE SYSTEMS

This work introduces a novel system incorporating Internet of Things (IoT) empowered drones, robotics, and deep learning methods, specifically Convolutional Neural Networks (CNN), to identify areas affected by landslides. The system uses modern sensors including cameras, LiDAR scanners, and environmental sensors, offering detailed information from distant and dangerous environments. The CNN model analyses this information to classify areas as damaged or undamaged with very high precision. Robustness of the system is emphasized using multi-sensor data acquisition, efficient preprocessing, and application of transfer learning to deliver optimal performance. Results reflect a high classification accuracy of over 99%, validating the system's capability in real-time landslide detection and disaster management. The current work seeks to contribute to more scalable and efficient disaster management systems using IoT and AI technologies.