TTBA

Abstract

People have been spending much time on free-weight exercises that can strengthen the muscles, connective tissues and tendons. To decrease the risk of injury and reap the benefit of free-weight exercises, a monitoring system which helps users exercise scientifically is necessary. Wearable sensors or changes of Radio Frequency (RF) signal have been exploited for activity sensing in prior work. However, wearable sensors methods are intrusive and may bother users, while the RF-based methods require training data and fail to extract fine-grained features of each action. Therefore, our goal is to design a system which is non-intrusive, privacy-insensitive and non-training. Tracking the free-weight equipment may meet the challenge of large accuracy loss, only by extending existing RF-based 2D tracking methods to 3D space. Instead, regarding the 3D moving as 1D moving simplifies the problem. We find that most actions in free-weight exercises can be divided into two kinds, circular and vertical motions. Therefore, this paper proposes a RFID-based system, TTBA, to track free-weight equipment instrumented with passive RFID tags. We implement a low-cost prototype of TTBA to recognize and track the two basic actions. Extensive experiments show that TTBA achieves high tracking accuracy for both motions, even mm-level accuracy for the vertical motion. And the potential TTBA has to achieve an assessment system is also showed in the practical evaluation.