SVM and RGB-D Sensor Based Gesture Recognition for UAV Control

Abstract

This research has the purpose of allowing anyone, with or without experience handling micro aerial vehicles, to operate unmanned aerial vehicles (UAV) in a natural and intuitive way, unlike typical interfaces that need experience and knowledge in piloting to be used. To achieve this, our approach uses gesture recognition, based on machine learning with Support Vector Machine (SVM) for classification and a RGB-D sensor for the feature extraction. Tests for recognition with different Kernel-SVM and for the RGB-D sensor with different levels of light were carried out.