Human activity recognition (HAR) using FMCW radar

Abstract

Human activity recognition (HAR) is an approach that is used to identify and classify human activities based on sensor data such as frequency modulated continuous wave (FMCW) radar, camera, accelerometers etc. HAR has gained a significant application in healthcare, smart home automation, search and surveillance, and monitoring of suspicious activity. Traditional approaches for HAR are based on the vision-based approach and wearable sensors. Vision-based systems fail due to privacy concerns, light illumination, and a limited field of view angle. The wearable sensor-based approach suffers due to inconvenience to the user and requires continuous use of the HAR sensor. Among the prominent sensing modalities, radar-based approach has gained significant attention due to their ability to operate in non-intrusive, privacy-preserving ways, ability to work in low lighting conditions or darkness, ability to penetrate through walls. The robustness of the radar-based approach against changing environment conditions making them suitable for HAR.