Multi-Input Fuzzy Rules Emulated Networks with a Hertzian Contact Force Sensor

Abstract

In this paper, the implementation of an adaptive neuro-fuzzy network called Multi-Input Fuzzy Rules Emulated Network (MIFREN) is presented. This network has a simple structure to describe the human knowledge about the robotic manipulator and the sensor in the form of fuzzy IF-THEN rules. MIFREN is implemented as a force feedback controller which used a force sensor based on Hertzian contact and ultrasound. The contact probe exhibits high sensitivity to initial instantaneous contact and provides an estimation of the reaction force. Based on the force signal provided by this probe, the MIFREN's structure can be constructed and integrated directly in the controller without any mathematical model of the system. The experimental results with synthetic elastomer and tomato samples demonstrate the superior performance of MIFREN compared to a well tuned PI controller.