Abstract
Our human joint mobility is vital to perform many activities of daily (ADL) needs, and any reduction in the range of motion (ROM) on the joint could have hinted at some form of lesions within the limb. Hence gathering reliable and precise data of the joint function over an extended period is imperative for clin ical assessment. However, most of these measurement approaches utilized ei ther non-wearable systems (NWS) or wearable systems (WS) that are too rigid and interfere with limb movement. Therefore there is a need to look at a new breed of WS measurement devices that the user can wear and track their kine matic data without interfering with the limb motion. Over the past decade, con ductive fabrics (CF), sometimes called e-textiles or smart textiles, has evolved as an emerging trend in medical application to measure kinematic parameters due to their comfort & soft property. In this thesis, we focus on the design methodology of wearable device using conductive fabric (CF) to perform human joint motion sensing. The central idea is to explore how CF can be utilized as a strain sensor to correlate the electrical signal to angular kinematic data using an appropriate mechanical model. We begin by characterizing the electromechanical property of Electrolycra & SUTD knitted sensor to understand key parameters affecting the performance of these two types of CF-based strain sensors. A sensor architecture was proposed us ing a constant current to reduce its measurement fluctuation. We have shown that such architecture improved the standard deviation (SD) of a commercial CF from 0.2 to 0.034. The results of this study yield an appropriate electrome chanical model for CF-based sensor design. Next, we introduce a new wearable device concept comprising a commercial CF strain sensor (Electrolycra) embedded as part of an inverted slider-crank (ISC) mechanism for elbow joint flexion-extension sensing. The use of the ISC vi mechanism has the benefits of not requiring anthropometric information from the user to relate the joint parameters to the CF strain sensor readings, which is a limitation of existing designs. We found that with this type of joint sensing device, we can track the elbow extension motion over a range of 140? with a maximum error of 7.66%. For the human trial, it shows that it can track the subject elbow flexion-extension with an error ranges of between 8.24? to 12.86? , and an acceptable average Spearman Coefficient (??) value of 0.95. Finally, we explored the concept of embedded sensors as part of the fabric material. Here we proposed an orthogonal sensor arrangement so that the an thropometric information can be measured as part of the donning process. Two knitted knee brace (single & H-shape sensor) was manufactured and tested on a human subject. We found that for this type of joint sensing device, we can track the knee motion of 3 types of activities of daily living (ADL) with an error range of between 7.89? to 16.46? and a mean Spearman Coefficient (??) value of 0.87 on a single embedded sensor design. For the H-shaped sensor design, it is between 21.4? to 30.3? with Spearman Coefficient (??) of between 0.167 to 0.525.