<![CDATA[This paper presents the comparison betweenthree artificial intelligence methods which are applied in brainwave to classify human conditions whether focus or relax inorder to operate robotic arm; these three methods are AdaptiveNeuro Fuzzy Inference System, Backpropagation NeuralNetwork, and Extreme Learning Machine. The system has fourinputs variable and single output variable. The inputs are RootMean Square (RMS) of four brain waves frequency bandmagnitude which is sensed by Neurosky Mindwave HeadsetEEG Sensor. These four waves are beta, alpha, theta and delta.The output variable is decision between 0 for relax and 1 forfocus. In this implementation, the sensor with one dry electrodeis placed properly in forehead in order to enable the accuratedelivery of RAW wave. This RAW data then transformed fromtime domain into frequency domain using Discrete FourierTransformation (DFT) in order to extract frequency band of thesignal. The RMS of each frequency band signal is then processedto classify user conditions whether focus or relax and sent tomicrocontroller to control robotic arm mechanism, grip (focus)and un-grip (relax). In this comparison, three methods aretreated equally with the same training and testing data. Theresult is Neural Network has higher accuracy to classify focusand relax data with average error 0.00009, followed by AdaptiveNeuro Fuzzy Inference System with 0.00563, and the last isExtreme Learning Machine with average error 0.10994]]>
Copyrights © 2019
