Novel Contactless Sensing Technique for Real-time Human Activity Detection

Recent research has looked to implement real-time contactless sensing within a healthcare application which can pro-vide monitoring of vulnerable people living at home. Currently systems make use of wearable devices to achieve this but this requires users to always be wearing devices. This paper presents a real-time contactless system which makes use of radio frequency signal propagation to determine if a person is sitting or standing. This is achieved by observing incoming channel state information to detect movements and passing detected movements to an AI model that can predict sitting or standing motions from changes the signal amplitude described in the channel state information. The system is able to make accurate real-time classifications in multiple environments.

近期研究致力于在医疗保健应用中实现实时非接触式传感，以便对居家的弱势群体进行监测。目前，相关系统利用可穿戴设备来实现这一目标，但这要求用户始终佩戴设备。本文介绍了一种实时非接触式系统，该系统利用射频信号传播来确定一个人是坐着还是站着。这是通过观察传入的信道状态信息以检测移动，并将检测到的移动传递给一个人工智能模型来实现的，该模型能够根据信道状态信息中描述的信号幅度变化来预测坐立动作。该系统能够在多种环境中进行准确的实时分类。