NeuroFuzzy Controller Design and FPGA-Based Embedded System for ACROBOT Model

Abstract

This work presents the modeling and control system design for a robot that rides a bicycle using the well-known Acrobot model for slow speeds, which was finally implemented in an FPGA-based embedded system. In this work the implementation of the controller was achieved for the Acrobot option following two ways. The first one implementation was developed based on modern control theories, involving:(a) the states feedback controller issues based on the appropriated poles allocation, guarantying stability and (b) the designs of a LQR (Linear Quadratic Regulator) type controller that minimizes the criterion of quadratic cost. The second one implementation was in turn achieved by means of some intelligent control methods, involving: (a) artificial neural networks, (c) tuning a neurofuzzy system, which joints the capacity of learning of the neural networks with the power of linguistic interpretation of the fuzzy inference systems (neurofuzzy system ANFIS, Adaptive NeuroFuzzy Systems) and (c) fuzzy logic. For testing the designed controller, a simulator has been also developed and connected to the controller embedded in FPGAs.