Large-area flexible MWCNT/PDMS pressure sensor for ergonomic design with aid of deep learning learning

The achievement of well-performing pressure sensors with low pressure detection, high sensitivity, large-scale integration, and effective analysis of the subsequent data remains a major challenge in the development of flexible piezoresistive sensors. In this study, a simple and extendable sensor preparation strategy was proposed to fabricate flexible sensors on the basis of multiwalled carbon nanotube/polydimethylsiloxane (MWCNT/PDMS) composites. A dispersant of tetrahydrofuran (THF) was added to solve the agglomeration of MWCNTs in PDMS, and the resistance of the obtained MWCNT/PDMS conductive unit with 7.5 wt.% MWCNTs were as low as 180 omega/hemisphere. Sensitivity (0.004 kPa(-1)), excellent response stability, fast response time (36 ms), and excellent electromechanical properties were demonstrated within the pressure range from 0 to 100 kPa. A large-area flexible sensor with 8 x 10 pixels was successfully adopted to detect the pressure distribution on the human back and to verify its applicability. Combining the sensor array with deep learning, inclination of human sitting was easily recognized with high accuracy, indicating that the combined technology can be used to guide ergonomic design.