Application to Pointing Movement System for Human-Robot Interface using Foreground Segmentation with Gaussian Mixture Model

Yihsin Ho, Yoshihiro Yamashita, Eri sato-Shimokawara, Toru Yamaguchi, and Norio Tagawa


Service robot, Vision, Human-Robot Interface, Gaussian Mixture Model


Human-robot interface has become one of main research areas on robotic systems. The authors propose an application to Pointing movement system for human-robot interface, and focuses on developing an intuitive operation using a new manner. Our application utilized a stereo camera to capture human’s images, and utilize them as primary sources of information retrieval. We apply Foreground segmentation with Gaussian Mixture Model (GMM) for fingertip recognition to be the main image processing method to propose system. The system is mainly executed in the following steps: a stereo camera captures the user’s hand motion, the system recognizes the user’s instruction from the hand motion and the situation (as well as environmental information), and then, control a robot motion. In this paper, we describe our applied image processing method, which employs foreground segmentation with GMM for fingertip recognition, and overall concepts of the proposed application is then also defined. We present a robot demonstration to specify proposed system. Finally, the demonstration experiment is presented to show the effectiveness and research potential of our application.

Important Links:

Go Back