Implementation and Calibration of Force Sensing Resistors in Insoles

_INTRODUCTION_ Many researchers have built pedobarography measurement systems [1]. Pedobarography, the study of forces interacting between the plantar surface of the feet and a supporting surface, can be used in many applications. Examples are analysis of gait and posture in for example orthotics design and monitoring of rehabilitation. Force sensing resistors (FSR) measure approximately the ground reaction force which is the force acting on the foot from the insole. Low sensor height, high linearity and good durability are three important sensor properties. FSRs are down to 0.2mm thick but can be non-linear and they can break easily if implemented wrongly. How can the FSRs be implemented in insoles and calibrated to overcome these sensor type shortcomings?_IMPLEMENTATION_ Discrete FSRs are often placed at the heel, inside and outside of the metatarsal pad and the big toe pad. It is vital to protect the boundary of the active sensor area. If the boundary is exposed to too much mechanical strain it will rapidly break down. This is because the upper and lower layers of the sensor get short-circuited. First the sensor will show intermittent maximum values and later stop working at all. One solution is to remove insole material right under the sensor boundary to relieve it from mechanical strain [2]._CALIBRATION_ The FSR manufacturers recommend using at least four different loads for the calibration and be aware that static load makes the sensor drift over time. It is important to choose a similar calibration procedure as the application it will be used for, e.g. dynamic calibration if the application is measurement while walking. A load cell should be used for calibrating the FSR. Linear or exponential regression are not recommended. Three or fourth power polynomials have shown to produce acceptable errors [3]._DISCUSSION_ FSRs have two less than ideal properties regarding linearity and durability. The lifetime of the sensor can be increased by taking special care of not exposing the active sensor area boundary to mechnical stress. Calibration should be done dynamically if the application is e.g. measurment while walking. Three power, or higher, polynomials are recomended for the regression._REFERENCES_ _[1]_ N. Hegde, et al., A Comparative Review of Footwear-Based Wearable Systems. Electronics 5(3), (2016)._[2]_ P. A. R. Hellström, Wireless Wearable Measurement System Based on Pedobarography for Monitoring of Health. Lic. Thes., (2016)._[3]_ J. M. Brimacombe, et al., Effect of calibration method on Tekscan sensor accuracy. J. Biomech. Eng. 131(3), (2009).