计算机科学
手势
带宽(计算)
人工智能
手势识别
可穿戴计算机
无线电频率
计算机视觉
电子工程
特征提取
传输(电信)
支持向量机
天线(收音机)
声学
功率(物理)
动力传输
相(物质)
特征(语言学)
钥匙(锁)
工程类
无线
语音识别
作者
J B P Stephenson,Manoj Kumar,Johannes Schneider,Shanmugam Kumar,Mahmoud Wagih
标识
DOI:10.1109/wisnet69500.2026.11408682
摘要
This paper presents a novel over-the-air (OTA) gesture recognition system using a single antenna, 3D-printed using liquid metals infills, and utilizing magnitude-only RF transmission data, unlike previous approaches that rely on phase or complex-valued data. The antenna is fabricated via stereolithographic additive manufacturing with liquid metal microchannels, enabling a wearable and deformable sensing platform. Gestureinduced variations in antenna gain are detected in the farfield through received power measurements, eliminating the need for multiple antennas or phase calibration. The antenna's $\left\vert S_{21}\right\vert$ response was measured at -45 dB at 3.64 m and a frequency bandwidth in the MHz range, with gesture capture performed using a vector network analyzer (VNA) over a bandwidth of $150-300 \text{MHz}$. A dataset of 10 British Sign Language gestures was collected at a distance exceeding 3.5 m and classified using machine learning (ML) techniques. The results show a high accuracy of 92%, validating magnitude-based OTA sensing relying on uncalibrated S-parameter magnitudes.
科研通智能强力驱动
Strongly Powered by AbleSci AI