Chadi
Fouad Riman
College
of Engineering and Technology, American University of the Middle East, Kuwait
E-mail: chadi_riman@yahoo.com
Submission: 5/14/2019
Revision: 9/18/2019
Accept: 9/25/2019
ABSTRACT
The aim of this project is to ease mobility for
people with upper and lower disability in order to live independently. This
paper presents the design steps and specification to a low cost hands free
eye-blink controller to control and electric wheelchair. Nowadays, people are
using joystick to control motorized wheelchair. The eye-blink controller
technology gives an alternative solution to mobility problem; especially for
the people who are quadriplegics. By interfacing eye-blink controller, the
directions of the wheelchair are controlled. This report will provide the
designing step, related solutions, and component details and specifications.
Keywords: mobility; smart controller interface; disability; assistive technology; embedded systems
1.
INTRODUCTION
Nowadays, a very important goal for
motor disabled people is to be autonomous in their mobility. They want to be
able to depend on themselves on their daily life tasks. Unfortunately, around
15% of the population worldwide lives with disability according to World Health
Organization and disability means losing the mobility (WHO, 2017). But nowadays
there are many choices for electric power wheelchair and the common one is are
joystick control, but this kind of electric wheelchair cannot be used for all
disabled people who suffer from the disability of the upper part of the body.
However; many hand-free controllers
have been developed for disabled people to make their mobility easier such as:
voice recognition technique, head control, eye movement, and chin control. But
each one of those different designs has at least one disadvantage for example
the voice recognition technique is not appropriate in noisy places. Moreover;
in the eye movement technique the user cannot see his surrounding freely, he
must concentrate on controlling the wheelchair.
For this, the propose of this paper
is to design a new hand-free method to control the wheelchair depending on eye blinking , so the user can
move freely with blink his eyes and the wheelchair will move to a specific
direction forward, right, left and stop. The controller will be applied on a
GoPiGo robot instead of electric wheelchair for demonstration purposes. Next,
the literature review will be mentioned, followed by the solution selection,
design schematic, experimental results, discussions, and conclusion.
2.
LITERATURE REVIEW
The electric wheelchair is very
popular and has been around since the mid of the last century. Early electric
wheelchairs simply used the frame of a manual wheelchair and added an electric
motor to it. Unlike manual wheelchairs, which require a great deal of upper
body strength to use, electric wheelchairs require virtually no effort on the
part of the user. They are also often referred to as power chairs or electric
power chairs.
Most electric wheelchairs use a
joystick control that is mounted to the armrest of the chair. This design is
the same used on the first electric wheelchairs and can be configured for use
on the left or right side of the power chair. There are also a number of
alternate controls available for those who are not able to operate the
joystick. (Arshak; Buckley; Kaneswaran, 2006)
Perhaps the most common alternate
wheelchair control allows the wheelchair to be controlled by the user’s breath.
Blowing into the wheelchair moves it forward and breathing in moves it
backwards. The direction of the wheelchair can also be controlled. Different
types of wheelchairs were designed with different types of controllers such as
head, chin, tongue, eye gaze and sip-and-puff.
There are many types of wheelchair's
controllers that serve several types of upper body disabilities in order to
meet the challenges and the needs of the patients. Among these, several
effective ideas of controllers that help mobility will be reviewed as the
following.
Standard
joystick controller
This type of controllers helps
alders and people with legs disability to move the electric wheelchair. The
joystick controller controls the electric wheelchair manually by moving the
joystick in different direction. After specifying the command from the user,
specific signal will be sent to the microcontroller where it identify the
command and execute it. Thus, the command will be send to the motor as a
digital signal. (HOVEROUND,
2012)
Sip-and-puff
controller
The sip-and-Puff controller is an
assistive technology that sends signals using air pressure by sipping
(inhaling) or puffing (exhaling) through a tube in order to move the electric
wheelchair. The idea of this design is based on a pressure sensor (absolute air
pressure) connected to a microcontroller circuit. The sensor measures the
pressure and sends it to the microcontroller. Then the microcontroller converts
the analog signal into a digital signal and sends it to the wheelchair
controllerto perform the wanted movement action. (Mougharbel et al., 2013)
Eye-blink
controller
This type of controller controls the
electric wheelchair by how many times the eye blinks. The eye blinking
mechanism is designed to produce commands forward, backward, right, left and
stop. This system involves three stages: image detection, image processing and
sending signals to the wheelchair controller. The eye blinks are detected using
a camera and sensor that are placed in front of the user. The sensor will send
the data to microcontroller which will process the information in an embedded
computer and then send the corresponding output signals to the wheelchair
controller to start moving the wheelchair. (Purwanto; Mardiyanto;
Arai, 2017)
Head
motion controlled wheelchair
This type of electric wheelchair has
a tilt communicator system that responds to head movements. It could be used by
disable persons who cannot move their hands and legs but they can move their
head. In addition, it works by using
tilt sensors. In addition, it’s a plan to fit the disabled person setting on it
and have a weight up to 100 kg. (Nehru, 2012)
Voice
controlled wheelchair
This system is designed to control
the wheelchair through the voice recognition. The components of this system are
microcontroller with microphone sensor, motors to move the wheelchair and
ultrasonic sensor to detect if there is any obstacle in front of the wheelchair
to stop it. (Pires; Nunes, 2002)
The next Table 1 shows a comparison
of the five above mentioned techniques.
Table 1: Comparison of 4 different wheelchair controllers.
Controller type |
·
Joystick control |
·
Voice control |
·
Tilt control |
||
Power consumption |
Low |
Average (IR) |
Avg. |
High |
Avg. |
Processor Speed |
Low |
High |
High |
High |
High |
Causes user’s
fatigue |
After heavy usage |
After heavy usage |
After light usage |
After light usage |
After light usage |
Used with upper
limb disability |
No |
Yes |
Yes |
Yes |
Yes |
Additional
controller cost |
No |
75 USD |
140 USD (MOUGHARBEK; El-HAJJ; GHAMLOUCH; MONACELLI, 2013) |
160 USD (Pires; Nunes, 2002) |
150 USD (Nehru, 2012) |
|
3.
SOLUTION SELECTION
The five different designs that were
mentioned in the previous section meet the user needs. Among them, only one
proves its quality and effectiveness based on some requirements and criteria. A
comparison of these designs will be mentioned to choose the appropriate one.
According to the table data Table 1,
the eye blink has the highest power consumption since it has camera screen,
unless it uses less power requiring equipment such as IR sensor. The
sip-and-puff controller uses average power consumption since it uses
microcontroller and microprocessor in its design unlike the joystick controller
that requires low power consumptions because they don't contain additional
components. Regarding to the fatigue caused on the user, the sip-and-puff,
tilt, and voice controllers cause the most fatigue, while the others cause less
fatigue.
This is due to the fact that the
lungs will be tired of inhaling and exhaling additional air for a long period,
the voice will be tired talking all the time, and the neck muscles will be
exhausted. Although the joystick controller has no additional charges like the
others, but it is not of practical use if the user has upper limb disability.
Therefore, the eye-blink controller is chosen to be implemented due to its
reasonable price, not causing fatigue, and its suitability for upper
extremities handicap. (Purwanto; Mardiyanto; Arai, 2017; Spd.org.sg, 2017; Rehabmart.com, 2019)
4.
DESIGN SCHEMATIC
The high-level design is distributed
into several small blocks. The wheelchair block is interchanged with GoPiGo
robot controller for the sake of experimentation.
Figure 1:
System’s high level design
IR Sensors
The first block after the user is
the two Infra-Red sensors, one for each eye. It is considered as a main part of
the design. The functionality of these sensors is to detect an eye blink from
the user’s left and right eyes. The sensor produces analogue signal as output,
which is sent to the microcontroller. (Agarwal et
al., 2015)
Microcontroller
Circuit
The second block is the
microcontroller that takes the analogue signals from the IR sensors and
converts them into digital signals which distinguish four states: no blink,
left eye blink, right eye blink and both eyes blink. These signals will be sent
to the embedded computer board for further action. The microcontroller used in
our prototype is a PIC16F877A microcontroller. (Milan Verle, 2008)
Computer
board (Raspberry Pi)
The third block is the Raspberry Pi
computer. It acts as the brain of the controller because it reads the data from
the microcontroller and converts it into commands that are sent to the robot’s
or wheelchair’s controller. (TechRepublic, 2017).
Wheelchair
or Robot controller
The fourth block is the controller
of the wheelchair or the robot. In our prototype, it is the controller for
GoPiGo robot. It receives the commands from the Raspberry Pi and sends the
actions to the robot motors in order to move the robot accordingly. (D. Industries, 2018)
Wheelchair
or Robot motors
The fifth and last block is the
motors of the wheelchair or the robot. In our prototype, it is the motors for
GoPiGo robot. They move according to desired directions sent by the robot’s
controller board.
The figure below shows the detailed connections. The green board on the
right is the Raspberry PI. The red board is the GoPiGo controller. The black
chip is the PIC microcontroller. The two similar components down are the IR
sensors.
Figure 2: System’s low level connection
Figure 3: GoPiGo robot used in the prototype
Figure 4: System’s eye glasses with sensors
The next figure shows the system
flowchart that illustrates the full process. At first the IR sensor will detect
the eye blink. If the right eye blinked, then the motor will move to the right
direction. If the left eye blinked, then it will move to the left direction. If
both eyes blinked, and the motor is stopped then the motor will go to forward
direction. Finally, if both eyes blinked and the motor is not stopped then the
motor will stop.
Figure 5: System’s flowchart
5.
EXPERIMENTAL RESULTS
In order to validate the prototype,
we need to test it on different scenarios and compare the results to the ones
without the eye-blink, in the latter case using the wheelchair’s joystick with
the hands. In the case of GoPiGo robot, we will compare the eye-blink
controller with the normal GoPiGo touch control panel.
In particular, three different path
scenarios were used for testing the eye-blink controller:
•
Straight
path
•
Curved
path
•
Maze
path
Straight path
The idea is to move straight between
two points A and B with the eye-blink controller and without it, as shown in
the following figure. In this test, the starting point is the orange box, and
the target point is the blue box. This test took 4.3 seconds to perform with
conventional control, and 7.3 seconds with eye-blink control. This is the
easiest test that includes moving in just one direction then stopping. The
incurred delay was just for the user to blink his eye and the microcontroller
getting the sensor’s reading then sends it to the Raspberry PI computer.
Figure 6: Straight path test scenario
Curved path
In this experiment, the user will
move the robot on a drawn path as in the figure 7. In this test, the starting
point and end point is the same. This test took 11.5 seconds to perform with
conventional control, and 19.5 seconds with eye-blink control. This scenario is
more challenging than the first one because it includes moving straight and
turning right and left as required in the path. The user needs to stay close to
the path and not deviate by more than 5 cm. At each turn there is more delay
for microcontroller response time, which was fixed at 0.5 seconds.
Figure 7: Curved path test scenario
Maze path
In this experiment, the user will
try to move the robot from the center of a maze to outside it as shown in
figure 8. In this test, the starting point is the center of the maze, and the
end point is the exit of the maze. This test took 25 seconds to perform with
conventional control, and 35 seconds with eye-blink control. This scenario is
the most challenging among the three scenarios because it involves moving in
all directions while avoiding hitting any of the maze walls. Again the delay
here was for microcontroller’s response time to the sensor’s readings.
Figure 8: Maze test scenario
6.
DISCUSSION
The prototype’s results are
acceptable when compared to conventional joystick control, although not so
good. The main reason behind it is probably the 0.5 second delay incurred for
the microprocessor to read the data from the sensor. Furthermore, the user
needs a proper training time to get used to the eye-blink system. The table
below shows the time comparison between regular control and eye-blink control
performance for the three different scenarios. On the average, the user needs
50% more time to execute the same task. This result is not optimal, but it is
acceptable taking into consideration the user’s disability in his upper
extremities.
Table 2: Time comparison between standard and
eye-blink control.
Scenario |
Standard |
Eye-Blink |
Difference |
Straight path |
4.3s |
7.3s |
3s |
Curved path |
11.5s |
19.5s |
8s |
Maze path |
25s |
35s |
10s |
Total |
40.8s |
61.8s |
21s |
Enhancements
can be made on the prototype to produce better results. The initial delay used
by the PIC microcontroller to take the sensor’s measurement is set to 0.5
second. Another delay used by the Raspberry PI to read the values sent by the
PIC microcontroller is set at 1 second. This makes the total delay of 1.5
second for each movement, which is certainly not acceptable.
A valid idea is to connect the
sensor directly to the Raspberry PI computer, thus eliminating the initial 0.5
second delay, and then using an interrupt to detect the change in the sensor
reading by the Raspberry PI computer. This will remove the other 1 second
delay. Therefore, all the 1.5 seconds useless delays can be removed per action.
If we review the first scenario, it needs only two actions: move forward then
stop.
The difference of 3 seconds shown in
table 2 reflects the 1.5 seconds delay per move, since we have only two moves
(forwards and stop). Ideally, we will reach the same timing as in standard
joystick control. This will happen after proper training on the new eye-blink
system.
7.
CONCLUSION
In conclusion, helping physically
disabled people to increase their mobility and depend on themselves is the main
challenge. This challenge is increased when even the upper extremities are not
functioning. In this case, an ordinary joystick controlled electric wheelchair
is not sufficient to provide autonomous user movements.
In our work, we designed an
electrical wheelchair controlled by eye blinking. This is a successful idea to
solve the mentioned problem. The user only needs to blink his right, left, or
both eyes to drive the wheelchair without using any other part of his body. Our
design depends on two Infrared Radiation (IR) sensors to detect left or right
eye blinking in order to decide on moving forward, right, left, or stop. This
system was built in a prototype that controls a GoPiGo robot, and then tested
in different scenarios to produce outputs compared to standard control.
As a future work, the design needs
several enhancements, starting by installing the system on a real electric
wheelchair instead of a robot.
Furthermore, a hardware redesign can
omit the microcontroller’s part to reduce delay and cost. A software redesign
can add software interrupts that will cause further delay reduction and make
the system on same level as a standard joystick control.
Another important enhancement is
adding distance sensors that will detect obstacles on the wheelchair’s way.
This detection can be included in the software’s design in order to avoid
obstacles.
A third enhancement involves adding
a wireless connection (instead of a wired connection) between the glass sensors
and the controller.
Further enhancements can include
solar energy instead of a regular chemical battery. Also make the controller
waterproof so that the chair can be used in outdoor environment. An emergency
special code (in blinking) can be added. This code will cause a GSM modem to
send SMS message to an emergency number for help.
8.
ACKNOWLEDGMENT
The author wishes to thank the AUM
computer engineering students who performed the preliminary testing on the
prototype, and reported the results mentioned in this work.
REFERENCES
Agarwal, T.; et al. (2015) Different Types of Sensors with their Applications - Edgefxkits,
Buy Electronics & Electrical Projects in India, 07-Dec-2015. Available:
https://www.edgefx.in/different-types-of-sensors-with-applications.
Arshak, K.; Buckley, D.;
Kaneswaran, K. (2006) Review of Assistive Devices for
Electric Powered Wheelchairs Navigation, The
ITB Journal, v. 7, n. 1, Article 3.
D. INDUSTRIES. (2018). GoPiGo, Available: http://www. dexterindustries.com/gopigo/.
HOVEROUND. (2012) Electronic Joystick Controllers
For Power Wheelchair, April 16, 2012. Retrieved from https:// www.hoveround .com/help/learn-more/power-wheelchairs-101/power-chair-parts-electronic-joystick-controller
Milan Verle (2008) PIC
Microcontrollers. 1st EDITION (2008), mikroElektronika.
Mougharbel, I.; El-Hajj,
A.; Ghamlouch, H.; Monacelli, E. (2013) Comparative
study on different adaptation approaches concerning a sip and puff controller
for a powered wheelchair, Science and
Information Conference (SAI), p. 597-603.
Nehru, J. (2012), “Head Controlled Wheelchair with Patient
Monitoring System”. Retrieved from
http://aac-rerc.psu.edu/wordpressmu/RESNA-SDC/2012/06/13/head-controlled-wheelchair-with-patient-monitoring-system/
Pires, G.; Nunes, U.
(2002) A wheelchair steered through voice commands and
assisted by a reactive fuzzy-logic controller, Journal of Intelligent and Robotic Systems, v. 34, n. 3, p.
301-314, 2002.
Purwanto, D.; Mardiyanto,
R.; Arai, K. (2017) Electric wheelchair control with gaze
direction and eye blinking. Springer,
2017. Available: https://link. springer.com/article/10.1007/s10015-009-0694-x.
REHABMART.COM (2019) Joystick Assistive Technology Switch. Available:
https://www.rehabmart.com/product/
joystick-assistive-technology-switch-32072.html.
SPD.ORG.SG, (2017) Types of Controllers for Powered Wheelchairs. Available: http://www.spd.org.sg/updates/
detail/types-of-controllers-for-powered-wheelchairs-313.htm
TECHREPUBLIC. (2017) Raspberry Pi 3: The inside story from the new $35 computer's
creator. Available: ttps://www.techrepublic.com/article/raspberry-pi-3-the-inside-story-from-the-new-35-computers-creator/.
WHO - WORLD REPORT ON DISABILITY. (2017) Who.int. Retrieved from http://www.who.int/disabilities/world_report/2011/report/en/