An online brain–computer interface using non-flashing visual evoked potentials

Abstract

Not until recently have motion-onset visual evoked potentials (mVEPs) been explored as a modality for brain-computer interface (BCI) applications. In this study, the first online BCI system based on mVEPs is presented, in which selection is discerned by subjects' focused attention to the moving cursor at a target virtual button. An adaptive approach was used to adjust the number of trial presentations according to the participants' online performance. With the EEG signal acquired from only a single channel, an acceptable information transfer rate of 42.1 bits min(-1) was achieved, averaged by 12 subjects. Furthermore, an online application for the Google search system was developed based on this paradigm. The promising results, that all of 12 participants were able to operate the system freely, validate the feasibility of a practical motion-onset VEP-based BCI which could be embedded into computer screen elements, such as menu, button and icon, for various applications.

Performance optimization of ERP-based BCIs using dynamic stopping

“…A very simple approach was taken in [9], where the average distance from the SVM hyperplane was calculated for target subtrials from the calibration data. Online, the trial is stopped as soon as the sum of classification scores up to that iteration is larger than N times the threshold for one of the classes.…”

“…As done in the original implementation [9], we averaged three trials in the time domain for Liu during the online phase. For all other methods, single subtrial classification was performed.…”

“…The simplest way to do this is by learning the optimal number of iterations on the calibration data and use this online. As performance of BCI is variable over time, more dynamic methods have been proposed [7], [8], [9], [10]. They stop the stimulation in a trial, based on the classification scores it has seen so far.…”

An Event-Related Potential-Based Adaptive Model for Telepresence Control of Humanoid Robot Motion in an Environment Cluttered With Obstacles

“…It includes N200 potential that is a negative valley appearing at poststimulus 180-325 ms and P300 potential that is a positive peak appearing at poststimulus 250-800 ms [6]. Since the first P300 speller based on the "oddball" paradigm was developed [7], the ERP-based brain-computer interface (BCI) systems have emerged, e.g., spelling sentences [8], controlling electrical applications in a virtual [9] or a lab environment [10], browsing Internet websites [11], controlling a wheelchair [12], [13], a hospital-bed nursing system [14], a robotic arm [15], and a humanoid robot [16]- [19]. Nevertheless, the telepresence control of a humanoid robot [20]- [24] via brain signals to perform complex operational tasks is not only helpful for the disabilities addressed in biomedical engineering but also very important for operators controlling humanoid robots in military, astronautic, and industrial applications.…”

Making brain-computer interfaces better : improving usability through post-processing

“…Verschore et al [63] decreased the necessary number of repetitions from 12 to only 2.69 on average. Liu et al [64] managed to increase their information transfer rate from 16 to 26 bits per minute in offline evaluation, and even to 42 bits per minute online. 3 d i s c u s s i o n a n d c o n c l u s i o n s 7 7 …”