A Natural Gesture-based Human-robot Interactive System

Current robot teleoperation technology has achieved great progress compared with the earlier robot teleoperation system; the efficiency of human-robot interaction has been significantly improved; and the interactive methods have been essentially innovated. Among various interactive technology, Gesture control has become one of focuses of the robot teleoperation area due to its intuitiveness and convenience.This paper combines the previous research achievements of our laboratory with professional knowledge of computer science learned in master course and raises a gesture-based natural human-robot interaction system. Compared to the existing human-robot interactive systems, this system exacts gesture information without wearing any data gloves or using various sensors. It decreases the difficulty of interaction and lessens fatigue during the teleoperation.The natural human-robot interaction system this paper mentions employs Client/Server structure and uses leap motion devices to input hand data. When operators use leap motion devices to interact with the remote robot, the input devices collect position and posture data. After collecting data, every client employs anti-shake technology to smooth the gesture data and then sends the data to the server according to the instruction from the server. The server collects the hand data from clients and picks up the best gesture data and then calculates the movement increment. Then the server transfers the hand data to robot joint angle by the inverse solution algorithm. Finally the server sends the movement instruction to the remote robot. During the teleoperation, operators control the remote robots by changing their own hand position and posture.Operators only need to use natural fluent gesture to control the remote robot. It raises efficiency of human operator and makes the interaction easier and more intuitive.