Control of robot using a brain computer interface

Summary form only given. In the video, experiments demonstrating the control of robot using a brain computer interface are shown. This is done using “actemes”. Actemes are fundamental units of action. The actemes can be combined to perform complex tasks. The user who performs the tasks imagines the complex actions as an ordered combination of actemes. The EEG of the user during the process is collected and classified in real-time using independent component analysis (ICA)[1, 2] and a neural network classifier[3, 4] to identify the actemes. This ordered set of actemes can be considered to be meaningful sentence in a language. The order in which the complex action is constructed is called the grammar. Two such tasks are demonstrated using these experiments. The first task involves insertion of round peg into a hole and rotating the peg inorder to simulate the insertion of a screw into threaded hole. The actemes involved are “insert” and “rotate”. The second task is a point-to-point movement of a peg in which the peg is to be moved from a circular zone at the bottom right to the circular zone centered at the cross hair at the top left of the screen. The representative image is shown in Fig 1.

[1]  Erkki Oja,et al.  Independent component analysis: algorithms and applications , 2000, Neural Networks.

[2]  Tzyy-Ping Jung,et al.  Recursive independent component analysis for online blind source separation , 2012, 2012 IEEE International Symposium on Circuits and Systems.

[3]  J. White Neural engineering , 2007, Annals of Biomedical Engineering.