Hidden-Markov-Model-Based Hand Gesture Recognition Techniques Used for a Human-Robot Interaction System

摘要

In this paper, we present part of a human-robot interaction system that recognizes meaningful gestures composed of continuous hand motions in real time based on hidden Markov models. This system acting as an interface is used for humans making various kinds of hand gestures to issue specific commands for conducting robots. To accomplish this, we define four basic types of directive gestures made by a single hand, which are moving upward, downward, leftward, and rightward individually. They serve as fundamental conducting gestures. Thus, if another hand is incorporated to making gestures, there are at most twenty-four kinds of compound gestures by the combination of the directive gestures using both hands. At present, we prescribe eight kinds of compound gestures employed in our developed human-robot interaction system, each of which is assigned a motion or functional control command, including moving forward, moving backward, turning left, turning right, stop, robot following, robot waiting, and ready, so that users can easily operate an autonomous robot. Experimental results reveal that our system can achieve an average gesture recognition rate of 96% at least. It is very satisfactory and encouraged.