Sixth International Conference on Advances in Computing, Control and Networking - ACCN 2017
Author(s) : CHOONSUNG NAM, DONG RYEOL SHIN, EARL KIM, JANG YEOL LEE, JUN HEON KIM
In this paper, we propose a method to improve the recognition rate of gestures motion and scalability problem which can occur in gestures when operate drone using machine learning. For these purposes, the gestures data transmitted from the drone controller are used for machine learning on real time, and a new learning model is periodically created using the gestures data stored in the HDFS(Hadoop Distributed File System). The goal of our proposed system is to increase the recognition rate of the gestures motion when new learning model is created. In addition, it is possible to expect enhanced scalability through recognition of gestures motion, and that drone is able to recognize a new gestures motion which is not defined in the server.