Dynamic modeling and compensation based on FLANN for wrist force sensor

The wrist force sensor is one of the most important sensors in the intelligent robot, and its dynamic characteristic needs to be studied and improved. This paper proposes a new method of sensor dynamic modeling and compensation based on a functional link artificial neural network (FLANN). The sensor models are described in form of differential equation. In the dynamic modeling, adopting a FLANN trained by a simple neural algorithm, the unknown coefficients of the equation are estimated accurately. A dynamic compensating digital filter is cascaded with the sensor so as to improve sensor dynamic response. In the design, using the inverse model of sensor as the model of dynamic compensating filter, its coefficients are learned by FLANN. The effect of this method is better than that of the pole-zero offset method.