background image
ORIGINAL ARTICLE
D. Purwanto · R. Mardiyanto
Department of Electrical Engineering, Institute of Technology 
“Sepuluh Nopember” (ITS), Surabaya, Indonesia
K. Arai (
)
Department of Information Science, Saga University, 1 Honjo, Saga 
840-8502, Japan
e-mail: arai@is.saga-u.ac.jp
This work was presented in part at the 14th International Symposium 
on Arti? cial Life and Robotics, Oita, Japan, February 5–7, 2009
Artif Life Robotics (2009) 14:397–400 
© ISAROB 2009
DOI 10.1007/s10015-009-0694-x
Djoko Purwanto · Ronny Mardiyanto · Kohei Arai
Electric wheelchair control with gaze direction and eye blinking
1 Introduction
The ability to move freely is highly valued by all people. 
However, it is sometimes dif? cult for a person with a physi-
cal disability. Nowadays, an electric wheelchair is commer-
cially available for disabled people. It generally requires 
considerable skill to operate. Moreover, some disabled 
people cannot drive an electric wheelchair manually, even 
with a joystick, because they lack the physical ability to 
control the movement. To enable a disabled person to drive 
a wheelchair safely and easily so that they can enjoy a 
higher quality of life, researchers have proposed several 
electric wheelchair systems. The use of voice commands to 
control an electric wheelchair is one research result. A small 
number of command words and high-performance voice 
recognition are employed in this system. An electric wheel-
chair control with electro-oculography (EOG) techniques 
has also been proposed. In this case, the different com-
mands for the wheelchair are derived from the electro-ocu-
lography (EOG) potential signals of eye movements.
A system for electric wheelchair control using the eyes 
   
In this research, a method of electric wheelchair control 
by gaze direction and eye blinking properties is proposed. 
A camera is set up in front of the wheelchair user to capture 
Abstract  We propose an electric wheelchair controlled by 
gaze direction and eye blinking. A camera is set up in front 
of a wheelchair user to capture image information. The 
sequential captured image is interpreted to obtain the gaze 
direction and eye blinking properties. The gaze direction is 
expressed by the horizontal angle of the gaze, and this is 
derived from the triangle formed by the centers of the eyes 
and the nose. The gaze direction and eye blinking are used 
to provide direction and timing commands, respectively. 
The direction command relates to the direction of move-
ment of the electric wheelchair, and the timing command 
relates to the time when the wheelchair should move. The 
timing command with an eye blinking mechanism is designed 
to generate ready, backward movement, and stop com-
mands for the electric wheelchair. Furthermore, to move at 
a certain velocity, the electric wheelchair also receives a 
velocity command as well as the direction and timing com-
mands. The disturbance observer-based control system is 
used to control the direction and velocity. For safety pur-
poses, an emergency stop is generated when the electric 
wheelchair user does not focus their gaze consistently in any 
direction for a speci? ed time. A number of simulations and 
experiments were conducted with the electric wheelchair in 
a laboratory environment.
Key words  Electric wheelchair control · Gaze estimation · 
Blinking measurement
Received and accepted: May 18, 2009
background image
398 
image information. The camera direction is focused onto 
the face area of the wheelchair user. The camera is con-
nected to a computer with vision processing and electric 
wheelchair motion control capabilities. With vision process-
ing, the sequential captured image is interpreted to obtain 
the gaze direction and eye blinking properties. The gaze 
direction is expressed as the horizontal angle of the user’s 
gaze, and is derived from the triangle formed by the centers 
of the eyes and the nose. The eye-blinking properties are 
obtained from the timing of blinks. The gaze direction and 
eye blinking are used to provide the direction and timing 
commands, respectively, for the electric wheelchair. The 
direction command relates to the direction of movement of 
the electric wheelchair, and the timing command relates to 
the time when the electric wheelchair should move. The 
eye-blinking mechanism with a blink duration of at least 
400 ms is designed to generate a timing command for when 
to move. Furthermore, the electric wheelchair also receives 
a velocity command to move at a certain velocity, as well 
as the direction and timing commands. The disturbance 
observer-based control system is used to control the direc-
tion and velocity. For safety purposes, an emergency stop 
is generated when the electric wheelchair user does not 
focus their gaze consistently in one direction for a speci? ed 
time.
2 Proposed system
2.1 System overview
 
processing and control purposes is placed in the bottom of 
the wheelchair. The electric wheelchair uses a differential 
steering mechanism. The wheelchair design considers the 
following four important factors: safety, ease of operation, 
low price, and convenience.
Four commands are use to control the electric wheel-
 
2.2 Electric wheelchair model
 
?
 and linear 
velocity  v. The wheelchair motion depends on the electric 
signal applied to the motor of each wheel. The angular 
velocities 
?
l
 and 
?
r
 represent the direct input signals to 
move the wheels of the electric wheelchair to the left or to 
the right, respectively. The relation between direction and 
velocity (
?
,  v) and the angular velocities of the wheels 
Fig. 1.  Electric wheelchair hardware
Fig. 2.  Commands for the electric wheelchair
Fig. 3.  Electric wheelchair model
background image
 
399
(
?
l
, 
?
r
) is given by Eq. 1, where R
w
 is radius of a wheel, 
?
 
is the gear ratio, and l is the distance between the wheels.
v
R
R
R
l
R
l
W
W
W
W
l
r
?
?
?
?
?
?
?
?
?
?
?
?
? =
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
2
2
 
(1)
2.3  Electric wheelchair system
 
2.4  Vision interpretation system
2.4.1  Gaze direction estimation
 
ˆ
? =
?
?
?
?
?
f
a b
L
1
 (2)
Fig. 4.  Electric wheelchair system
Fig. 5.  Relation between the triangular shape and the gaze
Fig. 6. Triangular shape from the 
eyes and the nose
The direction command 
?
cmd
 is derived from the gaze 
direction estimation 
?
ˆ  with a smoothing function formula-
tion to yield natural changes in commands and to suppress 
measurement noise.
?
?
cmd
=
( )
f
2
ˆ
 (3)
background image
400 
2.4.2 Eye-blinking interpretation
An interpretation of eye blinking is useful to generate a 
timing command which can be applied directly to the con-
troller. Eye blinking with a duration of at least 400 ms is 
used as the timing command. The algorithm of the timing 
command is as follows:
1.  If <1 blink>then <disable controller>
2.  If <2 blinks>then <enable controller>
3.  If <long blinking>then <set to move backward>
2.4.3 Safety detection
The safety detection control monitors the behavior of the 
electric wheelchair user and the performance of the vision 
system. “Bad” behavior is de? ned as when the user does 
not focus during wheelchair control. Failure of the vision 
system occurs when the system fails to detect the user’s face. 
The safety command algorithm is de? ned as
If <bad user behavior>or <face detect failed>then <disable 
controller>
2.5  Electric wheelchair control system
2.5.1  Electric wheelchair motors and controller
In both of the wheelchair motors, the disturbance observer 
technique is applied. The disturbance observer is a tech-
nique to estimate the disturbance existing in a plant, and to 
make the motion controller be an acceleration controller 
also.
 
 With an acceleration controller, the wheelchair motors 
can be modeled as an ideal integrator, and the block diagram 
of the controller is realized by constant gain K.
v
K
K
v
v
?
?
?
?
?
?
?
?
? =
?
?
?
?
?
?
?
?
?
?
?
?
?
?
0
0
cmd
cmd
 
(4)
2.5.2  Angular to linear velocity conversion and accelera-
tion reference generator
The conversion from angular to linear velocity is formu-
lated in Eq. 1. Furthermore, the acceleration reference can 
be derived from Eq. 1 as follows:
?
?
?
?
?
?
?
?
l
r
W
W
W
W
R
R
R
l
R
l
?
?
?
?
?
? =
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
2
2
1
 
(5)
3 Results
 
4 Conclusion
The feature points in the triangle made by the positions of 
the eyes and the nose can be formulated to generate a direc-
tion command for the electric wheelchair. Combining the 
direction, velocity, timing, and safety commands yields an 
effective method of controlling an electric wheelchair. With 
the disturbance observer technique applied as a control 
strategy, the design criteria of the wheelchair control system 
can easily be satis? ed.
References
1. Arai K, Purwanto D (2007) Electric wheelchair control with the 
human eye. 7th International Conference on Optimization: Tech-
niques and Applications, December 2007, Kobe, Japan
2.  Arai K, Uwataki H (2007) Computer input system based on viewing 
vector estimation with iris center detection from the image of users 
acquired with relatively cheap web camera allowing user move-
ments. J Inst Elec Eng Jpn C-127(7):1107–1114
3. Ohnishi K, Shibata M, Murakami T (1996) Motion control for 
advance mechatronics. IEEE/ASME Trans Mechatronics 1(1):
56–67
0
5
10
15
20
0
0.2
0.4
time (sec)
(m/sec)
Linear velocity response
 
command
actual
0
5
10
15
20
0
0.1
0.2
time (sec)
(rad/sec)
Rate of direction response
 
 
command
actual
0
5
10
15
20
0
10
20
30
time (sec)
(rad/sec)
Actual response of wheelchair motors
 
 
left
right
Fig. 7. Simulation results