Approximate modelling based on genetic algorithm for a POF force sensor for Human-Robot Interaction in Robotic Walker

Abstract

This paper presents the development and analysis of a low-cost POF-based force sensor. The system consists of two POF segments with sensing zones, with light source and photodetector arranged on a handle printed on polymer materials using 3D technology. The proposed sensor was characterized and compared with a commercial MTA400 force sensor (FUTEK, USA). An approximate viscoelasticity model was used to reduce the errors, and a genetic algorithm is used to define the model constants. The root mean square error decreased from 0.8310KgF without using the model to 0.3714KgF with the proposed. The POF force sensor proved to be comparable to the reference system as well as low-cost and easy to fabricate.