19-11-2012, 06:13 PM
Use of Forehead Bio-signals for Controlling an Intelligent
Wheelchair
[attachment=40266]
INTRODUCTION
Facial movement plays an important role in human
computer interaction for rehabilitation of disabled people.
Wheelchair users who have limitation in controlling an
electric-powered wheelchair with their limbs, for example,
people who suffer from spinal cord injury, quadriplegia,
hemiplegia or amputation, and on the other hand have
potential ability in generating facial movement, can adopt
face movements as an alternative control method for HMI
purpose. Compared with limb or hand, face region contains
complicated muscle groups that can generate delicate
movements especially in forms of facial expressions and
specific face motions from which plenty of information can
be retrieved by attaching sensors to designated face area.
Various face movement based input interfaces such as
sip and puff controller, infrared head controller [1], head
gesture controller [2], voice controller [3] and eye-gaze
controller [4] are developed for manipulability, safety, and
comfortableness of the user. Besides, signals such as
Electromyography (EMG), Electrooculography (EOG) and
Electroencephalography (EEG) can also be received from
specific muscle groups and organs on the face.
SYSTEM ARCHITECTURE
The proposed experimental system contains three parts. The
first part is the data acquisition device CyberlinkTM [8],
which is composed of 1) a data processing box and 2) a
wearable headband. The second part of the system is an
Intelligent Wheelchair platform which is used for
evaluating real-world performance of the control system in
the future. The third part is the human machine interface
which is responsible for extracting features and classifying
selected movement patterns and mapping the classified
patterns into wheelchair control commands; Details of this
part will be discussed further in Section III and IV.
CONTROL MOVEMENT AND FEATURE SUBSET
SELECTION
Control Movement Selection
Five face movements namely forehead single click (FSC),
forehead double click (FDC), left eye closing (LEC), right
eye closing (REC) and rhythmic jaw movement (RJM) are
selected as control movements due to a joint consideration
of uniqueness and distinguishability of movement patterns
reside in their EMG and EOG waveforms and the easiness
for the user to exert and learn.
FSC is a short time eyebrow lifting movement and FDC
is produced by two consecutive FSC. The process of LEC
and REC start from closing movement of left or right eye
and last by keeping the eye closed. The finishing of the
movement is marked by the opening of the closed eye. RJM
is a repetitive jaw movement that can be produced by the
user imitating a jaw clenching chewing-like motion. These
five control movements are easy for the user to perform and
learn as they are derived from the natural facial movement
in daily life.
CONCLUSION AND FUTURE WORK
This paper provides a solution for a new face movement
based control method that combines features from both
EMG and EOG signals. Different from previous research in
[8][9][10], this control method allows the user to control the
wheelchair without sharing extra attention to monitor the
state of control, so that the user can operate the wheelchair
and observing the surrounding environments at the same
time. Moreover, facial motion noises such as eye blinks and
random facial expression movements can be effectively
removed by employing both transient and steady features
Our future work will include bringing more EMG
channels from muscle group of different face regions and
getting more accurate EOG signals. EMG classification
methods such as support vector machine (SVM) and other
neural network based algorithms can also be applied to
improve the classification speed and accuracy.
Wheelchair
[attachment=40266]
INTRODUCTION
Facial movement plays an important role in human
computer interaction for rehabilitation of disabled people.
Wheelchair users who have limitation in controlling an
electric-powered wheelchair with their limbs, for example,
people who suffer from spinal cord injury, quadriplegia,
hemiplegia or amputation, and on the other hand have
potential ability in generating facial movement, can adopt
face movements as an alternative control method for HMI
purpose. Compared with limb or hand, face region contains
complicated muscle groups that can generate delicate
movements especially in forms of facial expressions and
specific face motions from which plenty of information can
be retrieved by attaching sensors to designated face area.
Various face movement based input interfaces such as
sip and puff controller, infrared head controller [1], head
gesture controller [2], voice controller [3] and eye-gaze
controller [4] are developed for manipulability, safety, and
comfortableness of the user. Besides, signals such as
Electromyography (EMG), Electrooculography (EOG) and
Electroencephalography (EEG) can also be received from
specific muscle groups and organs on the face.
SYSTEM ARCHITECTURE
The proposed experimental system contains three parts. The
first part is the data acquisition device CyberlinkTM [8],
which is composed of 1) a data processing box and 2) a
wearable headband. The second part of the system is an
Intelligent Wheelchair platform which is used for
evaluating real-world performance of the control system in
the future. The third part is the human machine interface
which is responsible for extracting features and classifying
selected movement patterns and mapping the classified
patterns into wheelchair control commands; Details of this
part will be discussed further in Section III and IV.
CONTROL MOVEMENT AND FEATURE SUBSET
SELECTION
Control Movement Selection
Five face movements namely forehead single click (FSC),
forehead double click (FDC), left eye closing (LEC), right
eye closing (REC) and rhythmic jaw movement (RJM) are
selected as control movements due to a joint consideration
of uniqueness and distinguishability of movement patterns
reside in their EMG and EOG waveforms and the easiness
for the user to exert and learn.
FSC is a short time eyebrow lifting movement and FDC
is produced by two consecutive FSC. The process of LEC
and REC start from closing movement of left or right eye
and last by keeping the eye closed. The finishing of the
movement is marked by the opening of the closed eye. RJM
is a repetitive jaw movement that can be produced by the
user imitating a jaw clenching chewing-like motion. These
five control movements are easy for the user to perform and
learn as they are derived from the natural facial movement
in daily life.
CONCLUSION AND FUTURE WORK
This paper provides a solution for a new face movement
based control method that combines features from both
EMG and EOG signals. Different from previous research in
[8][9][10], this control method allows the user to control the
wheelchair without sharing extra attention to monitor the
state of control, so that the user can operate the wheelchair
and observing the surrounding environments at the same
time. Moreover, facial motion noises such as eye blinks and
random facial expression movements can be effectively
removed by employing both transient and steady features
Our future work will include bringing more EMG
channels from muscle group of different face regions and
getting more accurate EOG signals. EMG classification
methods such as support vector machine (SVM) and other
neural network based algorithms can also be applied to
improve the classification speed and accuracy.