A Micro-Doppler Based Human Pose Estimation Framework for Single-channel Ultra-Wideband Radar

Human pose estimation based on optical sensors is difficult to solve the situation under harsh environment and shielding. In this paper, a Micro-Doppler (MD) based human pose estimation for the single-channel ultra-wideband (UWB) radar, called MD-Pose, is proposed. It is found that the MD signature contains the time-varying motion of the human torso, arms, legs, hands, and feet, which can be used in the pose estimation task. The MD spectrum of human echo is obtained by the STFT transform, which is the input of proposed MD-Pose. An asymmetric U-Net network is trained with UWB radar MD spectrum, which can estimate the human keypoints. The experimental results of a single-channel UWB radar validate the effectiveness of the proposed method.