Obstacle Detection in Path Planning for Unmanned Aerial Vehicles based on YOLO

Unmanned Aerial Vehicles (UAVs) are increasingly used in a range of applications, including surveillance and delivery, disaster management, and precision agriculture. Real-time and precise object recognition is critical in these environments because it allows UAVs to perceive and interact with their surroundings independently. The You Only Look Once (YOLO) algorithm family has emerged as a promising solution for efficient object identification applications due to its ability to achieve short inference times while maintaining competitive detection accuracy. This work thoroughly investigates the use of YOLO variations in object recognition for UAVs. We analyze the architectural innovations and algorithmic optimizations introduced in several versions of YOLO and explain how they affect UAV-based object detecting tasks. Furthermore, we assess the constraints and opportunities associated with deploying YOLO models on UAV platforms, taking into account parameters such as computational efficiency, model size, and environmental robustness. Through this we hope to provide insights into cutting-edge methodologies, highlight new trends, and recommend areas for future research at the intersection of UAVs and YOLO-based object recognition.