Application of the artificial neural networks on visually guided robot

Abstract

In this chapter, the carrying of an object at a workspace, which was perceived by vision, to another location was realized by a robot arm with five axes. Basic image process techniques were used for object recognition and position determination. If the desired object was inside the workspace, the inverse kinematics solution was realized, and then after coordinates of the object's location was sent to the robot arm. The inverse kinematics solution of the robot arm was performed with Artificial Neural Networks (ANN) model (Multi Layer Perceptron-MLP and Radial Basis Function (RBF) Neural Network) based on the forward kinematics solution. For an inverse kinematics solution of the robot, the training data set was created in the ANN method by using the robot's forward kinematics values first and then, ANN modeling was realized. After the robot's inverse kinematics solution was realized, the determined joint angle values were directed to the robot arm and moving the object to the desired location was realized successfully. Experimental results presented in this chapter indicate that RBF is more efficient solution than MLP for inverse kinematic solution of visually guided robot. © 2011 by Nova Science Publishers, Inc. All rights reserved.