| Wearable gesture recognition devices are intelligent devices that combine sensors and corresponding algorithms to identify human body movements.They achieve gesture recognition by monitoring the movements and postures of the wearer’s limbs.Currently available wearable gesture recognition devices on the market mainly utilize traditional rigid sensor components paired with microelectromechanical systems(MEMS)to achieve sensing functionality.However,rigid sensors are expensive to manufacture,offer poor wearer comfort,have limited stretching range,and lack flexibility in usage.Therefore,flexible sensors with high sensitivity,conformability to the human body,wide stretching range,and low cost have become a new approach for designing intelligent wearable gesture recognition devices.However,current flexible sensors suffer from poor durability,data fluctuations,and these issues become more prominent with the increase in data dimensions.These problems have become the main limiting factors for the application of flexible sensors.This study focuses on the common problems of poor durability and stability in flexible sensors and conducts research around MXene/PUbased flexible sensors.An integrated sensor design solution is proposed,and based on this,a data glove,a data wristband,and a data elbow pad using flexible sensors are designed and fabricated.An artificial neural network is employed to construct a gesture recognition model,and ultimately,gesture recognition is visualized.The main work and achievements of this paper are as follows:(1)The mechanism of MXene/PU composite-based flexible sensors is studied,and integrated flexible sensing units based on MXene/PU composite material are designed and fabricated.Tests show that the sensing units have high sensitivity,fast response speed,and high stability.(2)A data glove,a smart wristband,and a smart elbow pad based on the above integrated flexible sensing units are designed and fabricated.A gesture signal acquisition circuit based on the voltage division method is designed,and data processing is achieved through index-weighted filtering,cutoff filtering,and normalization algorithm.(3)A gesture recognition model based on a multilayer perceptron algorithm is designed.The model is trained using a collected dataset of the gestures,achieving recognition of 10 static numeral gestures and 9 tactical gestures with an accuracy of 100%.(4)A human-machine interaction system for gesture recognition is designed and implemented,enabling real-time recognition of the gestures and real-time display of recognition results. |
