Precision landing systems for unmanned aerial vehicles (UAVs) present a significant challenge that must be addressed to prevent damage and accidents. Traditional object detection systems for precision landing often require complex computations and substantial processing power, which can be difficult to implement on the limited resources of a typical flight computer. This study proposes the use of computer vision technology, specifically ArUco markers, to detect and track a drone's landing pad. In this research, a precision landing system utilizing ArUco markers is implemented on a quadcopter drone. The system employs a webcam module to detect the ArUco marker on the landing pad and a computer vision algorithm to track it. A flight controller estimates the relative position and orientation of the quadcopter, enabling precise landing on the pad. The proposed methodology includes camera modeling and calibration, marker detection, marker pose estimation, and quadcopter control. Experimental results demonstrate a significant improvement in landing accuracy when compared to traditional GPS-based landing. The system achieved an average landing accuracy of 34.1 cm, compared to 121.1 cm with GPS alone, resulting in an 87% accuracy improvement. These findings highlight the effectiveness of the proposed ArUco marker-based precision landing system in enhancing the landing accuracy of UAVs.
Computer Vision-Driven Precision Landing in Quadcopter Drones Using ArUco Markers
2024-08-07
729240 byte
Conference paper
Electronic Resource
English