Abstract
Smart garments offer good potential to monitor and assess dynamic upper body movements, particularly from the shoulder joint with its 3 Degrees of Freedom (DOF). In this paper, a customizable lightweight smart shirt is proposed, which is designed and manufactured using Computer Numerical Control (CNC) knitting technology. This automated textile manufacturing method enables the seamless fabrication of multiple piezoresistive sensors at designated positions and at a high resolution within the garment. The paper presents a proof-of-concept system that can record both shoulder and elbow joint motions in a non-obtrusive and non-invasive manner. This system consists of: i) a fully CNC knitted smart crop top with multi-material stretchable strain sensors and interconnects; ii) a regression model, trained on several full Range of Motion (ROM) upper body actions, to convert these sensor signals into shoulder and elbow joint angles; iii) an assessment using an ergonomics software to translate these converted joint angles into muscle fatigue markers. To validate the workflow, the subject performed a pick-and-hold activity outside of the regression model's training dataset. Subsequently, the joint angles of the subject performing this pick-and-hold activity, converted from the smart garment, are compared with joint angle data from the Vicon motion capture system. Comparing the ergonomics software outputs from both datasets shows good agreement with overall low root mean square errors, the highest being 0.787% for maximum voluntary contraction (%MVC) and 1.896% for duty cycle limit (%DC).