Intelligent control of robot arm using artificial neural networks

Abstract

An intelligent technique has been developed to track cancer automatically. First, a conventional ultrasonic diagnostic device is fixed on top of a robot arm to track the cancer as shown on a monitor. Second, an artificial neural network has been applied to learn the track trajectory. Using the proposed method, the cancer location can be identified and automatic tracking becomes possible. In this study, a neural network with recurrent connections has been used to predict the position of the cancer using the respiration waveform information and the diagnosis device's location. In addition, for fixing the joint angle of the robot arm in order to direct the diagnosis device to move to that position, an inverse kinematics model using the neural network has been proposed. Finally, by combining these two neural networks, an automatic control of the robot arm has been successfully carried out.