Abstract:Objective To propose a low-cost and algorithm solution for dynamic monitoring of plantar pressure by measuring the vertical ground reaction force (VGRF) during walking with a capacitive insole sensor. Methods Subjects wore the capacitive insole sensor to collect walking data (fast walking and slow walking) on a Kistler 3D dynamometer, and the data collected by the capacitive insole sensor were pre-processed and fitted by the residual neural network ResNet18 to obtain the vertical ground reaction force with high accuracy. Results Compared with the analysis of the data collected from the Kister force measurment platform, the normalized root mean square error (NRMSE) for fast walking and slow walking were 8.16% and 6.49%, respectively, and the Pearman correlation coefficient were greater than 0.96. Conclusion This study provides an economical and highly accurate measurement method for the dynamic measurement of vertical ground reaction forces.