A robot navigation method based on human-robot interaction for 3D environment mapping

Abstract

This paper proposed a navigation method based on human-robot intuitive interaction for unknown environments 3D mapping. Firstly, we define some gestures of robot control so that the older or children can easily use to guide the robot mapping. Based on SVM, gestures recognition is realized. Secondly, a human-robot interaction framework is designed that combines remote controller based on gestures with the robot autonomous obstacle-avoidance. The VR glass is used for the user to enhance the reality sense of the robot facing parts of environment, so people can interact with the robot more naturally. Then, the mobile robot can execute expected navigation path and use a RGB-D sensor to build 3D map of surrounded environment. Finally, interaction experiments is carried out to test algorithm valid based on the platform includes one host computer with a RealSense and a VR glasses, and a mobile robot-TurtleBot2 with Kinect and Zed sensors.