A Two-Axis Robotic Arm Controller Design Using Exact Radial Bases And General Regression Neural Networks With FPGA Technique

Abstract

Although neural networks applicable to the solution of robotics control problems are, in fact, neurocontrollers, their function is specialized mainly to provide solutions to robot arm motion problems. In this paper, two types of neural network models are applied for solving a number of robot kinematics problems these types are: Exact Radial Bases Neural Network (RBENN) and General Regression Neural Networks (GRNN). Learning the robot arms motions to achieve the desired final position are the main kinematics tasks covered here. Hardware realization of a Neural Network, to a large extent depends on the efficient implementation of a single neuron. Field Programmable Gate Array FPGA-based reconfigurable computing architectures are suitable for hardware implementation of neural networks. FPGA realization of NNs with a large number of neurons is still a challenging task. This paper also discusses the issues involved in implementation of a multi-input neuron with linear/nonlinear excitation functions using VHDL programming language for FPGA.