Article Preview
Top1. Introduction
Robots and humans can interact in a variety of ways, including vision, hearing, touch, and smell. Among them, humans obtain about 80% to 90% of information through vision. Therefore, although human-computer interactive decoration technology has now undergone five development stages of keyboard, mouse, touch, multimedia and virtual reality, there are also many interactive methods suitable for robotics, such as face recognition, expression recognition, voice recognition, Human motion recognition, gesture recognition, and gaze tracking, etc., but vision-based human-computer interaction is still the mainstream interaction method (Camille, 2020). Among them, the three human-machine interaction methods of human faces, human movements, and gestures have been initially applied in the virtual environment, and have a very broad application prospect. For this reason, the gesture interactive decoration design of robots has become increasingly important.
Most of the earliest gesture interactive decoration techniques used a special electromechanical or magnetic induction device to directly measure the posture of the hand and the angle of the joints. Aiguo developed a glove-type sensor system (SayreGlove), which is the predecessor of data gloves (Song, 2015). Ding has obtained the patent of “data glove”. Through “data glove”, the computer can obtain rich information such as the position of the hand and the extension of the finger (Qi, 2017). Ross Mead has also conducted research on gesture recognition, and can recognize 46 gesture symbols through similar devices (Ross, 2016). Judith Tiferes used a CyberGlove model data glove with 18 sensors in a domestic sign language recognition system, and for the first time in the world, realized a continuous Chinese sign language recognition system capable of recognizing more than 5000 words (Judith, 2018). Although this gesture interaction method can accurately and real-timely display 3D models of human hands, it is necessary to wear heavy data gloves when performing gesture interaction, which is not convenient for natural movement, and the equipment cost is high, and it is not easy to popularize. Therefore, the way of acquiring low-precision human movements through simple external devices is gradually emerging in the field of gesture interaction. Most of the earliest gesture interaction technologies used a dedicated electromechanical or magnetic induction device to directly measure the posture of the hand and the angle of the joints (Yiannis, 2016). Such as the glove sensor system (SayreGlove) developed by ToraDefanti and Daniel Sandin in 1977, which is the predecessor of data gloves. In 1983, Grimes first patented the “data glove”. Through the “data glove”, the computer can obtain rich information such as the position of the hand and the extension of the finger. In 1991, Fujitsu also carried out research work on gesture recognition, which can recognize 46 gesture symbols through similar devices. In 2000, Professor Gao Wen (now a professor at Peking University) of the Institute of Computing Technology of the Chinese Academy of Sciences and Professor Wu Jiangqin (now a professor at Zhejiang University) of Harbin Institute of Technology used CyberGlove model data gloves with 18 sensors in the Chinese Sign Language Recognition System. For the first time in the world, a continuous Chinese sign language recognition system capable of identifying more than 5,000 words has been implemented. Although this gesture interaction method can accurately and real-timely display 3D models of human hands, it is necessary to wear heavy data gloves when performing gesture interaction, which is not convenient for natural movement, and the equipment cost is high, and it is not easy to popularize. Therefore, the way of acquiring low-precision human movements through simple external devices is gradually emerging in the field of gesture interaction.