AI Based Drone for Surveillance

Abstract

Unmanned Aerial Vehicles (UAV also known as drone) are rising as part of surveillance, security and many other applications and so as the intelligence in drone. Here, an autonomous drone path planning framework to reach a dynamic target is proposed. The objective is to generate an intelligence algorithm for drone, accessing the location of target through developed mobile app for user using built-in GPS in real-time and navigating the drone to reach dynamic target. Live location coordinates synchronization with cloud database on requirement of user accessing through application developed for mobile platform to communicate with ground station, intelligence algorithm for drone initiation with coordinate data real-time synchronization, and autonomous flight to reach dynamic target. Object detection and tracking using YOLOv3 framework trained on visDrone-2019 dataset on TensorFlow module in Python is developed for scene surveillance and possible applications on aerial images captured through feed of camera onboard drone. Robustness and flexibility of framework is configured according to the noise and physical disturbances in feed during flight.