An Investigation of Gesture Recognition Technology Based on Artificial Intelligence and Wireless Sensing

Abstract

Now human-computer interaction technology has been closely integrated with the daily life, of which gesture recognition technology has received special attention, especially those using artificial intelligence wireless sensing solutions. This wireless gesture recognition technology mainly determines hand movements by analyzing changes in radio signals, and it has many advantages, such as better protecting user privacy, not affected by ambient light, and compatible with various devices. However, this technology also has many problems, such as unstable wireless signals, insufficient algorithm adaptability, and error due to different environments, devices or users. Models include LAGER, WiGNN, Wi-AM, and WiVi-GR. Each of these models has its own characteristics. LAGER solves the problem of labelless data through adversarial mapping, WiGNN improves cross-domain adaptability with dynamic topology, Wi-AM combines meta-learning to handle sample deficiencies, and WiVi-GR combines wireless signals with visual signals to break through the limitations of a single mode. The article concludes with some problems that the technology has current aspects, such as multiple users’ difficulty in distinguishing when making gestures at the same time.