Performance of Conductive Rubber-based Sensor Array Calibration Using Cubic Spline

Conductive rubber is used for various applications in engineering fields due to its advantages such as a wide range of electrical conductivities, low cost and ease of shaping. The rubber needs to be calibrated before it can be used in response to the applied force. Due to the highly nonlinear property of rubber, researchers usually apply methods such as mathematical model and neural network to map the voltage-force response. These methods, however, require complex calculation and large quantity of training data. In this paper, an implementation of cubic spline for calibrating the voltage-force response is reported. This method not only requires a small number of data, but also provides good curve fitting for nonlinear function. Two types of cubic spline are explained and carried out in experiments. The result shows that the cubic spline technique was able to map the voltage-force up to 50 N. The calculated average measurement error was 3.44 +/- 1.58 N, mean squared error for border sensing elements was 7.78 N-2 and inside sensing elements was 3.47 N-2. Paired t-test for cubic spline and back propagation neural network shows significant difference (p < 0.001) where cubic spline performed much better than the latter method. The result indicated that the cubic spline method was able to estimate voltage-force response and feasible for force sensor application.