Neuro-fuzzy control of bilateral teleoperation system using FPGA

This paper presents an adaptive neuro-fuzzy controller ANFIS (Adaptive Neuro-Fuzzy Inference System) for a bilateral teleoperation system based on FPGA (Field Programmable Gate Array). The proposed controller combines the learning capabilities of neural networks with the inference capabilities of fuzzy logic, to adapt with dynamic variations in master and slave robots and to guarantee good practical robustness against the disturbances, by adjusting neuro-fuzzy network output parameters in a short time, thanks to the computing power of FPGA and its high sampling frequency. The design methodology adopted to design the control algorithm aims to minimize the hardware resources used by the FPGA in order to optimize the execution and the design times, and this by using the Fixed-Point Tool and HDL Coder features of MATLAB-Simulink. The proposed controllers were experimentally validated on a teleoperation system comprising a pair of one degree of freedom. The experimental results clearly show that the proposed ANFIS control algorithm significantly outperformed the conventional control methods (PID).