Smart Sensor Using Function Approximation

This paper reports a neural network (NN) implementation of function approximation. Function approximation (aka, nonlinear regression) identifies input-output relationship from given input-output data set. Concept of function approximation is applied to develop a smart position sensor. Smart position sensor consists of three standard sensors that are coupled with a neural network to produce an estimate of the location of an object in one dimension. MATLAB is used to construct and train the multi layer feed forward neural network. Hardware implementation of trained neural network is done on Arduino Uno microcontroller board.