You are on page 1of 2

Design of an Asynchronous Brain-Computer

Interface for Control of a Virtual Avatar

Hye-Soo An, Jeong-Woo Kim, and Seong-Whan Lee


Department of Brain and Cognitive Engineering, Korea University, Seoul, Korea
{hs_an, jw_kim, sw.lee}@korea.ac.kr

Abstract—Brain-Computer Interface (BCI) enables the human to


control external devices by measuring brain activities. Among the
various BCI paradigms, motor imagery (MI) is the natural one to
accomplish the objectives of BCI. The asynchronous mode
enables the user to perform the MI in a self-paced manner. In
this study, we propose a design of the asynchronous BCI based on
MI for control of a virtual avatar in BCI game. Filter bank
common spatial pattern (FBCSP) is applied in proposed system
to discriminate correctly and detect rapidly the user’s different
intention based on EEG analysis in real-time. In conclusion, we (A) (B)
expect that our system would improve the performance of MI- Figure 1. BrainRunners. (A) Screenshot. (B) Setup.
based asynchronous BCI system.
performance. To solve the problem, the Filter Bank Common
Keywords-Asynchronous system; Motor Imagery; Cybathlon; Spatial Pattern (FBCSP) algorithm [4], which enables the
Brain Computer Interface (BCI); Electroencephalogram (EEG); system to select autonomously the discriminative subject-
Filter bank common spatial pattern (FBCSP); specific frequency range, has proposed to be as effective in
synchronous but yet to be explored in asynchronous mode.
I. INTRODUCTION In this paper, we propose a novel design of the multiclass
Brain-Computer Interface (BCI) allows a user to control asynchronous MI-BCI with the advanced signal processing
computer applications by measuring brain signal such as methods used to detect and discriminate different motor
electroencephalography (EEG) [1]. Several EEG-based BCI imagery classes on real-time EEG analysis, which is applicable
systems rely on voluntary modulations of sensory motor to virtual running environment.
rhythms (SMRs) (e.g., imagination of motor movement) [2].
Also there is the evidence that patients diagnosed with II. MATERIALS AND METHODS
amyotrophic lateral sclerosis (ALS) can accomplish SMR
modulations [1]. A distinct effect is that both movement and A. Application
motor imagery (MI) of a limb are accompanied by a decrease BrainRunners presented in Fig. 1 is a multi-play “running”
of power in mu and beta frequency bands known as event- game developed for the BCI race of the Cybathlon [5]. Each
related desynchronization (ERD) and followed rebound of avatar continuously runs to the finish line of the race even
power in the beta band known as event-related synchronization though there is no command input. Running on a specific
(ERS). The spatial patterns of ERD and ERS are specific for colored pad, the player who controls the avatar has to send the
each limb. For these reasons, MI is considered as one of the matched command using the BCI. There are three kinds of the
natural paradigms for building a BCI [3]. colored pads including SPEED, JUMP, and ROLL action. The
There are two kinds of modes for MI-based BCI system standard user datagram protocol (UDP) as a network protocol
(MI-BCI): synchronous (cue-paced) and asynchronous (self- is used to send the commands for avatar to the game and the
paced) [3]. The former is easier to implement but less input value of command to each pad is predetermined in the
convenient to reflect the user’s intent. In contrast, the latter game manual. If the player sends the correct command in the
called asynchronous MI-BCI controls the interface respective action pad, then it gives the advantage time (8.0 s)
continuously without cues or temporal constraints in the real and lasts boost until the end of the pad. However, the erroneous
world. However, there were some limitations in the systematic command (e.g. ROLL command on SPEED pad) brings about
approach. First, the continuous signal processing over the time the delayed time (2.5 s). The player reaching the finish line as
was followed but did not consider the overlapped time. Second, the first wins the race.
the extension to multiclass classification for various commands
was not taken into account well. In addition, features extracted B. Asynchronous MI-BCI
from the broadly estimated frequency band had been Our BCI system runs in two phases: the calibration phase
commonly used and led to the degradation of system and the feedback phase. We designed an asynchronous MI-BCI

Authorized licensed use limited to: Escuela Superior de Ingeneria Mecanica. Downloaded on April 29,2021 at 18:11:10 UTC from IEEE Xplore. Restrictions apply.
Figure. 2 System architecture of the asynchronous MI-BCI
applicable to BrainRunners online BCI game. The system classification result of the asynchronous BCI system is not
architecture is shown in Fig. 2. transmitted to the remote application.
1) Calibration phase: The subjects are instructed to a) Online interface control: Connecting the BCI system
perform motor imagination tasks indicated on the gray screen to BrainRunners, the avatar is synchronized to each subject.
for every eight seconds. One trial is composed of 2 s fixation While the avatars are standing at the starting line, every
cross, 4 s motor imagery cued by arrows, and 2 s blank screen. command is invalid. After starting the race, the subject
There are the arrow-shaped visual cues pointing left, right, or controls the avatar using the BCI input.
down and each cue corresponds to imaginary movement of the
III. DISCUSSION AND CONCLUSION
left hand (LH), right hand (RH), or foot (F). Also the blank
screen refers to resting state (REST). In one run, 25 trials of In the real world, the key is how does BCI system rapidly
each motor imagination condition are recorded. Above all, we detect and correctly discriminate the user’s different intention.
This paper has described a novel design of the multiclass
record the raw EEG data in one run of performed movements asynchronous MI-BCI for control of a virtual avatar. In
and then three runs of imagined movements [1]. The recorded particular, finding the subject-specific parameters in the
the EEG signals are used for the training of the classifier and multiclass extension of FBCSP and applying those in the
assessment of generalization error by cross-validation [2]. asynchronous MI-BCI would improve the performance of
a) Filter Bank Common Spatial Pattern: In the FBCSP, classification compared to the previous MI-BCI studies. In
the EEG signals are first bandpass-filtered into multiple addition, the introduction of the overlapped sliding window for
frequency bands, namely, 4-8, 8-12, …, 36-40 Hz [4]. The real-time operation is expected to make the system more
signals are segmented in the regular time interval of 4s. After efficient and the user more convenient. Therefore, our future
that, spatial filtering using the Common Spatial Pattern (CSP) work is to implement this designed BCI and demonstrate the
for multiclass MI is performed for nine bands. Then, Mutual systematic efficiency in some experiments.
Information based Best Individual Feature (MIBIF) algorithm
is used to select the discriminative CSP features from the filter ACKNOWLEDGMENT
bank [4]. This work was supported by ICT R&D program of
b) Regularized Linear Discriminant Analysis: The MSIP/IITP. [R0126-15-1107, Development of Intelligent
shrinkage-Regularized Linear Discriminant Analysis (RLDA) Pattern Recognition Softwares for Ambulatory Brain-Computer
is used to classify the selected features. In particular, the Interface].
RLDA classifier in One-Versus-Rest (OVR) approach is
employed for classification of four-class of MI (e.g., LH, RH, REFERENCES
F and REST). [1] B. Blankertz, C. Sannelli, S. Halder, E.M. Hammer, A. Kübler, K.-R.
Müller, G. Curio, and T. Dickhaus, “Neurophysiological Predictor of
2) Feedback Phase: After the calibration phase, the SMR-based BCI Performance,” NeuroImage, Vol. 51, No. 4, 2010, pp.
subjects are instructed to control the avatar in BrainRunners 1303-1309.
by performing the motor imagination tasks. In this phase, the [2] B. Blankertz, R. Tomioka, S. Lemm, M. Kawanabe, and K.-R. Müller,
EEG signals are recorded with 250 Hz of sampling rates. For “Optimizing Spatial Filters for Robust EEG Single-Trial Analysis,”
IEEE Signal Processing Magazine, Vol. 25, No. 1, 2008, pp. 41-56.
real-time operation, features are calculated every 40 ms with a
[3] R. Kus, D. Valbuena, J. Zygierewicz, T. Malechka, A. Graeser, and P.
sliding 3 s window [1]. The output of the classification is Durka, “Asynchronous BCI Based on Motor Imagery With Automated
processed through applying the CSP filters, band-pass Calibration and Neurofeedback Training,” IEEE Trans. Neural Syst.
filtering, calculating log-variance and applying the OVR Rehabil. Eng., Vol. 20, No. 6, 2012, pp. 823-835.
RLDA classifier in 1b). To send the subject’s commands to [4] K. K. Ang, Z. Y. Chin, H. Zhang, and C. Guan, “Filter Bank Common
Spatial Pattern (FBCSP) in Brain-Computer Interface,” Proceedings of
the application, we construct UDP object associated with the the IEEE International Joint Conference on Neural Networks, 2008, pp.
remote host. Until binding to the local socket, the 2391-2398.
[5] Cybathlon: BCI Race – Races & Rules (Ver. 2015-06-12),
http://www.cybathlon.ethz.ch/the-disciplines/bci-race.html

Authorized licensed use limited to: Escuela Superior de Ingeneria Mecanica. Downloaded on April 29,2021 at 18:11:10 UTC from IEEE Xplore. Restrictions apply.

You might also like