A Radar Measurement Setup with a Ground Truth System for Micro-Doppler Human Movements

Abstract

Human gestures and movements recognition is becoming an indispensable task in today’s world for many ranges of applications. For example, radar sensors are of big interest in the automotive industry due to its many advantages such as the ability to work in various weather conditions beside performing during night as well as during daytime. A ground truth system would be advantageous in labelling and obtaining more details of the measured scenarios. In this paper, a measurement setup with a ground truth system using Kinect is going to be used. Processing of the data obtained from Kinect proves to be helpful in getting more information about the performed gestures. The synchronization between Kinect and radar is also an important issue, which is performed here on a hardware level to minimize the delays from the software level. In a world where deep learning is increasingly used in various aspects of life, having more synchronized data points from Kinect beside the ones from the radar is also needed.