| Title |
Object Detection Using CNN for Automatic Landing of drones |
| Authors |
최지욱(Ji-Wook Choi) ; 황도경(Do-Kyung Hwang) ; 안종우(Jong-Woo An) ; 이장명(Jang-Myung Lee) |
| DOI |
https://doi.org/10.5573/ieie.2019.56.5.82 |
| Keywords |
쿼드콥터 ; 무인항공기 ; 비전 센서 ; 객체 인식 ; 제어 ; CNN |
| Abstract |
Unmanned aircraft must be in a stable position during takeoff, landing and flight to perform tasks such as monitoring. Existing drones use the Return To Home function, which automatically returns to the take-off point and landing when in flight. This method has the disadvantage that it is difficult to land at the correct point because the landing takes place using GPS. In this paper, we propose an algorithm that recognizes the landing point by attaching the vision sensor to the quadrotor type unmanned aerial vehicle and automatically landing. Using the CNN (Convolutional Neural Network) based algorithm, it is possible to reduce the occurrence of disturbance or object recognition error at the landing point, using the vision data to calculate the coordinates of the landing point and to control the landing precisely at the point. In order to verify the validity of the algorithm, the landing using the Return to Home function and the GPS data of the landing using the proposed algorithm were compared and verified by experiments. |