QUADCOPTER LANDING CONTROL BASED ON FUZZY LOGIC AND COMPUTER VISION

Abstract

The article addresses the problem of autonomous quadcopter landing in the event of loss of connection with the operator, which is a crucial task for modern unmanned aerial vehicles (UAVs). The main focus is on using computer vision and fuzzy logic technologies to improve the accuracy and reliability of landing under conditions of high uncertainty. The YOLO model is used for processing input images, ensuring efficient real-time object detection and enabling the identification of potentially safe landing zones for the drone. The integration of sensor data and computer vision allows for a comprehensive assessment of the landing area.The proposed approach includes fuzzy logic algorithms that adapt to dynamic conditions and enable decision-making in cases of incomplete or noisy data. This is particularly relevant for unpredictable situations such as changing weather conditions, the presence of moving objects, or reduced sensor signal quality.Based on the study, software of autonomous control system was developed, consisting of a flight controller and a data processing model. The controller activates the autonomous mode upon loss of connection with the operator, after which the system analyzes the terrain images and selects the optimal landing trajectory. To simulate the system’s operation, the LiftOff simulator was used, which allowed testing the algorithms in dynamic scenarios and optimizing their performance.The test results confirmed the system’s high accuracy and reliability under various conditions, which is important for applications in search and rescue operations, territory monitoring, and other tasks requiring autonomous UAV operation. The proposed approach demonstrates the effectiveness of combining modern computer vision methods with adaptive fuzzy logic to solve quadcopter autonomous control tasks.