Use of Unmanned Aerial Vehicle for Autonomous Person Detection and Tracking

Unmanned Aerial Vehicle (UAV) technology has rapidly advanced, enabling the execution of increasingly complex autonomous tasks. This work presents a cognitive surveillance approach utilizing the low-cost UAV, integrating perception and decision-making for indoor monitoring. The system combines communication protocols, ROS-based control, ORB-SLAM3 visual mapping, OctoMap-based occupancy modeling, and OpenCV for real-time face detection and tracking. A lightweight web interface was implemented for operator interaction, enabling autonomous navigation commands, real-time video streaming, and automatic email alerts triggered by detected faces. Experiments in both simulated and real environments demonstrated that the UAV successfully mapped unknown spaces, avoided obstacles, and navigated autonomously without collisions, relying solely on monocular odometry. The face detection and tracking module consistently maintained the target in the field of view, capturing and transmitting images via email. Despite hardware limitations, results highlighted the effectiveness of learning from environmental cues, adapting navigation, and maintaining continuous target tracking. This confirms the feasibility of low-cost UAVs for autonomous cognitive surveillance applications in security-critical scenarios.