Abstract: To develop a full-textile smart data glove for gesture recognition, intarsia plating technology and knitted full-forming technology were used to seamlessly introduce flexible strain sensors into the finger joints. Therefore, a flexible knitted strain sensor was successfully developed and applied to the preparation of fully fashioned smart gloves. 
Firstly, the flexible strain sensor was developed by using intarsia plating technology. The sensor was characterized in detail, including surface morphology, air permeability, sensitivity, response time, strain monitoring range, cyclic stability and water washing resistance. The sensitivity factor depends on the stretching direction. Results showed that the sensitivity of longitudinal (along the wales) stretching was much higher than that of transverse (along the courses) stretching. Therefore, the performance testing of the knitted sensor and the developed smart gloves imbedded with these sensors were all based on the longitudinal stretching mode. The measured sensitivity coefficient of the knitted strain sensor ranged from 13 to 90 within 30% strain, the response time was less than 50 ms, and it still maintained a relatively stable resistance after 8,000 cycles of stretching, showing good sensing performance. The testing results confirmed the ability of the sensor to capture strain signals in real time. In addition, the sensor also retained good breathability and wearing comfort of conventional fabric gloves. 
Then, the fully fashioned strain-sensing glove was developed by using knitted full-forming technology. Ten knitted sensors, which were located at 10 finger joints of five fingers, were incorporated. As for the preparation of smart gloves, advanced CAD design and a computerized flatbed knitting machine were adopted. The sensors reflect the bending state of the finger. Through the data acquisition and transmission system, the prepared knitted strain-sensing glove can accurately capture and distinguish hand movements in real time, so as to achieve dynamic gesture monitoring. In terms of hand function training for gesture recognition, the sensor glove shows excellent sensing ability to accurately capture small changes in hand movements. Therefore, by real-time monitoring of hand movement speed and finger curvature, this smart glove can serve as a rehabilitation training tool for patients with hand movement disorders. By collecting samples of different hand gestures and aided by the machine learning algorithm, an efficient gesture recognition model was built, which achieved the average recognition accuracy of 99.5% in the actual test.
The smart sensing gloves can effectively realize gesture recognition and be used in scenarios such as hand function training and human-computer interaction. It has broad application prospects in fields such as rehabilitation medicine and leisure and entertainment. In the future, with the flexibility of knitting technology and the diversity of knitting structures, fully fashioned knitted sensor devices such as smart knee pads and smart elbow pads can also be made, which have broad application prospects in rehabilitation medicine and sports.

Key words: knitted strain sensor, intarsia plating, fully fashioned sensing gloves, hand gesture recognition, hand function training

摘要： 为开发一种用于手势识别的全织物智能数据手套，利用嵌花添纱技术和针织全成形技术，将柔性电阻应变式传感器无缝引入到手套的手指关节部位。采用扫描电子显微镜、透气性分析仪和传感性能参数分析仪进行测试分析，结果表明：针织应变传感器在30%应变内的灵敏系数范围为13 ~ 90，响应时间低于50 ms，且在8000次循环拉伸后仍保持较为稳定的电阻，表现出良好的传感性能。制备的针织全成形应变传感手套能够准确实时捕捉手部运动数据，同时还保留了常规织物手套所具有的良好透气性和穿戴舒适性。辅以机器学习，该智能传感手套可有效实现手势识别，可用于手部功能训练和人机交互等场景，在康复医疗和休闲娱乐等领域具有广阔的应用前景。

关键词: 针织应变传感器, 嵌花添纱, 全成形传感手套, 手势识别, 手部功能训练