Summary: | 碩士 === 國立臺灣海洋大學 === 通訊與導航工程學系 === 107 === This study applies drones in the cage farming industry and sets two goals that drones need to complete. One is environment detection and another is cage recognition. Considering computing power of the on-board processor and computation of the image processing, we use different image recognition theory to recognize different target. In environment detection, Hough Transform is used to recognize simple target (single color circle), and get the horizontal distance of the drone and target from image to adjust the drone and drop the sensor. In cage recognition, Support Vector Machine (SVM) and neural network are used to recognize the cage, and comparison of these recognition theories is given. This study integrates the drone, camera, Raspberry Pi and servo motor, and uses basic sensors and cruising function with GPS. The drone can fly to the destination, recognize the target and do the mission which is set previously.
|