Flexible Strain Sensor Based on Conductive Hydrogel/KC@PDMS for Neck Motion Control Wheelchair Using EMD-LSTM Algorithm

Many disabled people are not suitable for the control of the current mainstream manual joystick-controlled wheelchair, so the study of new control methods can provide convenience for these people, including bioelectrical signal control, voice control, and visual control, but these control methods have peripheral equipment complex or susceptible to environmental noise interference, in contrast. These problems can be solved effectively by using a wearable flexible sensor to recognize human movement signal. To collect and identify the strain signal characteristics of the moving neck muscles, based on the principle of flexible piezoresistive sensor, we designed and produced a K-carrageenan at polydimethylsiloxane (PDMS) flexible sensor, designed a series of experiments to explore the conductive properties, mechanical properties, and air permeability of the sensor, and also designed the peripheral acquisition circuit and signal processing algorithm. Finally, the empirical mode decomposition (EMD)-long short-term memory (LSTM) algorithm is used to classify the signal and explore the feasibility of fitting the signal to the human neck to control the wheelchair. The results show that the maximum tensile rate of the sensor designed by us is as high as 180%, and the air permeability is also improved slightly. After 1000 cycles of stretching, the sensor still has good performance and can effectively collect human neck action signals. Through the field test, the accuracy of this system is up to 95%, which provides a new idea for wheelchair control.