Contactless human activity and sport action identification based on UHF RFID tag wall and AI

Abstract

The existing approaches to activity assisted living(AAL) are complex, expensive, and intrusive, which reduce theirpracticality and end-user acceptance. However, advancements inconcomitant technology, such as artificial intelligence and wirelesscommunications, can be leveraged to improvise advanced AALsystems that can reduce healthcare costs and hospitalisations. Such systems can identify, monitor, and localize hazardousactivities to respond swiftly to unforeseen events. To addressthese issues, we present a Transparent RFID Tag Wall (TRTWall) that utilizes a passive ultra-high frequency (UHF) arrayof radio frequency identification (RFID) tags and deep learningbasedpredictive analytics for contactless human physical activitymonitoring. Specifically, we consider five different activities thatinclude sitting, standing, walking (in both directions), and noactivity. Our experimental results demonstrate that TRT-Wallcan accurately differentiate among five distinct activities withan average accuracy is 95.6%. This suggests that our proposedcontactless AAL system possesses the significant potential toenhance elderly patient-assisted living.