Experimental Evaluation of UAV Path Planning Through Integration of YOLO-Based Obstacle Detection with ROS Using Stereo Camera

This paper presents a comprehensive integration of existing technologies for autonomous robot path planning, with a specific focus on Unmanned Aerial Vehicles (UAVs). The integration involves leveraging the You Only Look Once (YOLO) algorithm for obstacle detection, the Robot Operating System (ROS) framework for seamless integration, a stereo camera for finding positions, and a classical path planning algorithm. The novelty lies in the experimental results demonstrating the successful operation of the integrated system, and the use of only an external single stereo camera for positioning and obstacle detection, eliminating the need for expensive sensors. The system demonstrates good accuracy, reliability, and real-world applicability. This integrated approach has practical value, advancing autonomous robot navigation research and enabling various applications in complex environments.