International Journal of Engineering & Technology, 7 (2) (2018) 453-457

International Journal of Engineering & Technology

Website: www.sciencepubco.com/index.php/IJET
doi: 10.14419/ijet.v7i2.9190
Research paper

Smart wheelchair using android smartphone for

physically disabled people
Siti Nur Suhaila Mirin 1,*, Khalil Azha Mohd Annuar 1, Chai Pui Yook 1
1 Department of Electrical Engineering Technology, Faculty of Engineering Technology, Universiti Teknikal Malaysia Melaka,
Hang Tuah Jaya, 76100, Durian Tunggal, Melaka, Malaysia
*Corresponding author E-mail: nursuhaila@utem.edu.my

Abstract

This paper describes the development of a smart wheelchair system with voice recognition and touch controlled using an embedded system.
An android application is developed and installed on the android smartphone. The system is divided into two main modes: voice recognition
mode and touch mode. For the voice recognition mode, elderlies or physically disabled people (users) can provide the voice input, for
example, “go”, “reverse”, “turn to the left”, “turn to the right” and “stop”. The wheelchair will move according to the command given. For
the touch mode, the user can select the specified direction displayed within the four quadrants on the screen of the android smartphone to
control the wheelchair. An Arduino Uno is used to execute all commands. The MD30C motor driver and HC05 Bluetooth module are used
in this system. This system is designed to save time and energy of the user.

Keywords: Smart Wheelchair; Bluetooth Module; Touch Mode; Voice Recognition Mode; Android.

command given. When the user selects the “Go” arrow, the wheel-
1. Introduction chair will move in a forward direction, “Back” arrow prompts the
wheelchair to move backward, and “Left” arrow causes the wheel-
Due to the increased percentage of elderlies and physically disabled chair to turn left, and “Right” arrow makes the wheelchair turn right
people, wheelchairs are the best assistive devices to help them en- [9]. An elderly or physically challenged person can direct the direc-
hance their personal mobility. The traditional wheelchairs have tion and movement of the wheelchair with the help of the android
some limitations such as flexibility, bulkiness and limited functions smartphone in four different directions, left, right, forward, reverse
[1-2]. There are existing technologies which allow the users to use and stop. The wheelchair will move according to the command
human gestures such as the movements of hands, movements of leg given by the user.
[3-4], tongue [5] and head [6-7], and synchronize them with the In [10] presented an idea of an eye controlled system which enables
movements of the wheelchair for a better wheelchair controls for the movement of wheelchair depends on the movements of eye-
example smart wheelchair. balls. A camera is mounted on the wheelchair; the wheelchair can
A smart wheelchair is developed to help an elderly or physically move in a certain direction when the user looks at that direction by
disabled person (user) to move from one place to another inde- making eye movements. Based on the eye-detected location, the di-
pendently. An android application is developed and installed in the rection of the possible motion is found, the command is transmitted
android smartphone. It consists of two controlled modes, the first to the motor control device via Arduino.
mode is the touch mode and the second mode is the voice recogni- Pajkanovic and Dokic proposed a microcontroller system that ena-
tion mode [8]. In the first mode, the user can give the voice input bles an electric wheelchair to be controlled by the head motion. The
using an android smartphone. The android smartphone will convert system comprises electronic and mechanical components [11]. The
the voice commands into a string of data and this string of data will accelerometer is used to collect the head motion data. The output of
be sent to the Bluetooth module and lastly delivered to Arduino the digital system is connected to a mechanical actuator and it is
Uno. After that, Arduino will decodes and process it. The motor used to position the wheelchair’s joystick based on the user’s com-
driver will direct the wheelchair according to the command given. mand. The sensor data is processed by a novel algorithm; it is im-
When the user says “go”, the wheelchair will move forward, the plemented within the microcontroller. The user’s head motion is
word “reverse” causes the wheelchair to move backward, the word translated into the wheelchair’s joystick position.
“turn to the left” causes the wheelchair to turn left, and the word Nishimori et al. implemented a voice controlled wheelchair [12]
“turn to the right” causes the wheelchair to turn right. which uses the voice command as the interface. A grammar-based
For the second mode, the user can determine the wheelchair’s recognition parser named “Julian” is used in this system. This is
movement by selecting the desired direction on the android open-source software which is developed by Kyoto University and
smartphone phone screen. The command given by the user will be Nara Institute of Science and Technology. The voice commands
forwarded to the Arduino Uno via Bluetooth. The Bluetooth will consist of nine reaction commands and five verification commands.
convert the commands given by the user in a binary format and send The reaction commands consist of five basic reaction commands
them to the Arduino Uno. Arduino Uno will read and execute the and four short moving reaction commands. This system is based on
command and lastly send the digital values to the motor driver de- the commercial electronic wheelchair Nissin Medical Industries co.
vice. The motor driver will direct the wheelchair according to the NEO-PI. In this system, it consists of a headset microphone and a
Copyright © 2018 Siti Nur Suhaila Mirin et al. This is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
 454 International Journal of Engineering & Technology

laptop. The user can select either to control by voice or button. The
control signal is sent to PIC from the laptop, and the PIC generates
the motor control signal to drive the wheelchair. Meanwhile, in [13]
introduce the voice recognition module with smart wheelchair and
have same control movement with this project but in this paper rec-
ognize the commands through microphone. While in [14] smart
wheelchair with voice recognition upgrade its capability with navi-
gation maps and person present location with GPRS and GSM sys-
tem.
In this paper will discuss, a smart wheelchair is developed to help
an elderly or physically disabled person (user) to move from one
place to another independently. An android application is devel-
oped and installed in the android smartphone. It consists of two con-
trolled modes, the first mode is the touch mode and the second mode
is the voice recognition mode. An elderly or physically challenged
person can direct the direction and movement of the wheelchair
with the help of the android smartphone in four different directions,
left, right, forward, reverse and stop. The wheelchair will move ac-
cording to the command given by the user.

2. Methodology
For this part will be divided into two main part, software and hard- Fig. 2: The Design Block Diagram for the Touch Mode.
ware development:
2.2. Hardware Implementation
2.1. Development of android application (MIT app inven-
tor) The system consists of six main components, two motor drivers,
two scooter DC motor, a Bluetooth module, a microcontroller, a
MIT App Inventor is an application to transform a complex lan- power supply and software. Both motors were mounted to both
guage of text-based coding into a visual and drag-and-drop building wheels and the control box is placed between them a 24V power
block. A command is a block that specifies an action to be per- supply is placed in front of the control box shows in Fig. 3.
formed on the phone. Some commands require one or more input
values to completely specify their action. In Fig. 1 and Fig. 2 repre- Table 1: The Table of Components
sent voice recognition mode and touch mode block diagram using No. Components Description
Android application. The best part for using android is an open- 1. Motor driver MD30C
source electronics platform and it is able to read an input and con- 2. Scooter DC motor 350W,24V,2600rpm
3. Bluetooth module HC-05
vert it into an output. Arduino is cheaper compared to other micro-
4. Microcontroller AVR(Arduino Uno)
controllers and it can run on windows, Macintosh OSX and Linux 5. Power supply 24V/7.0Ah lead acid and 9V battery
operating system. 6. Software used Arduino and MIT app inventor

(A) (B)

Fig. 1: The Design Block Diagram for the Voice Recognition Mode.

Fig. 3: (A) Front View and (B) Rear View of the Wheelchair.

The control box shows in Fig. 4 consists of Arduino Uno, an

MD30C motor driver and a Bluetooth module. Arduino Uno is used
to control the motor driver. Its only digital output pin is used.
 International Journal of Engineering & Technology 455

The command given by the user is transmitted to Arduino Uno via

the Bluetooth module. The Bluetooth will first convert the com-
mands given by the user in a binary format and send to Arduino
Uno. Next, the Arduino Uno will read and execute the command
and lastly send the digital values to the motor driver device. The
motor driver will direct the wheelchair according to the command
given.

Fig. 4: The Control Box.

Arduino Uno is a microcontroller board based on the ATmega328.

The architecture of Arduino microcontroller is the AVR and the op-
erating voltage is 5V. The HC-05 Bluetooth module is designed for
the transparent wireless serial connection setup. It has only 4 pins
which are 5V, GND, TX and RX. The 5V and GND pin are used
for power purpose and the TX and RX pin are used for a serial in-
terface. The pin configurations of 6 to 9 are used to control the mo-
tor driver. Pin 6 and 7 are connected to motor driver 1 and pin 8 and
9 are connected to motor driver 2. The pin configurations of 10 and
11 are used for serial interfacing with the TX and RX. Arduino Uno
will decode and process the data delivered from the Bluetooth. Then,
it will pass the signal to the motor driver. The motor driver will
direct the wheelchair according to the command given.

2.3. Work flow

The power supply is used to supply electrical energy to the Blue-

tooth module, Arduino Uno, motor driver and motor. The user can
Fig. 6: Flow-Chart of Voice Recognition Mode.
control the movement of the wheelchair by giving command via the
android smartphone. The user can select one of the modes to give
command.

Fig. 5: Smart Wheelchair Work-Flow Process.

For the voice recognition mode in Fig 6, the user must turn on the
Bluetooth on the android smartphone, and then select the correct
paired device to connect. When the Bluetooth is connected, the user
can start to give command. For the touch mode in Fig 7, in the same Fig. 7: Flow-Chart for Touch Mode.
way as the voice recognition mode, the user needs to connect to
Bluetooth and start giving command by selecting the direction spec-
ified within the four quadrants on the screen.
 456 International Journal of Engineering & Technology

1. Results and discussion From above results shows there is a maximum of 3 wrong words in
every five repeated words. There are 25 over 45 commands recog-
nised by the voice recognition mode. The percentage of consistency
For the voice recognition mode and touch mode, the wheelchair will
of the circuit in the noisy area is 55.6%.
move according to the command given. There are four possible di-
From both results it can see that the consistency of the voice recog-
rections of movements: forward, backward, left, right.
nition mode in the silent area is higher than the noisy area. This
means that the voice recognition mode is less efficient in the noisy
3.1. The voice recognition mode area.
In order to test the effectiveness of this system, a few experiments 3.1.2. Voice recognition mode tested by ten random speakers in
were carried out to analyse the consistency of the voice recognition two different environments
mode in different conditions and voice inputs. The first experiment
was carried out in two different areas, silent area and noisy area. For this testing 10 random speakers were asked to give the voice
The second experiment was tested by 10 different speakers in dif- command. Each person is required to give 9 commands; “go”,
ferent environments. Each speaker was asked to give 9 commands “back”, “forward”, “reverse”, “left”, “turn to the left”, “right”, “turn
and the number of times the system response correctly will give the to the right” and “stop”.
result. For each of the tests, was repeated 10 times and results are Condition 1: Silent area
averaged.
Table 4: The Test Results of Ten Random Speakers in the Silent Area
3.1.1. Voice recognition mode testing in two different environ- F F F F F M M M M M Total
ments Go 0 1 0 1 1 1 0 0 1 1 6
Back 1 1 1 1 1 0 1 1 1 1 9
Two experiments were carried out in different areas, in a silent area Forward 1 1 0 1 1 1 1 0 0 1 7
and noisy area. The purposes of these experiments are to identify Reverse 1 1 1 0 1 1 0 1 0 0 6
the consistency of the voice recognition mode in different areas that Left 0 0 1 1 0 1 1 1 0 1 6
are embedded in an android phone. The percentage of consistency Turn to
1 0 1 0 1 1 0 1 1 0 6
the left
each command when spoken by the user is calculated by the fol-
Right 0 1 1 1 1 0 1 1 0 1 7
lowing equation: Turn to
1 1 0 1 1 1 1 1 1 1 9
the right
Number of Recognised Words
percentage of consistancy = X100% Stop 1 1 1 1 1 1 1 0 1 0 8
Number of Tested Words
(1) Total 6 7 6 7 8 7 6 6 5 6 90

Condition 1: Silent area There is a maximum of 3 wrong words in every 10 speakers’ re-
This experiment was carried out in a silent area. Five tests were peated words. The percentage of the consistency could be higher if
performed on the circuit based on the commands listed on Table 2 the test is done with good pronunciations and the words are spoken
below. at a moderate pace. This test proved that there is no major difference
in the consistency and efficiency of the system whether the speaker
Table 2: The Voice Commands in the Silent Area is male or female.
No. Testing Condition 2: Noisy area
1 2 3 4 5 Total
Command Table 5: The Test Results of Ten Random Speakers in the Noisy Area
Go 1 1 1 1 1 5 To-
F F F F F M M M M M
Back 1 1 1 0 1 4 tal
Forward 0 1 0 1 1 3 Go 1 1 0 1 1 1 0 0 1 0 6
Reverse 1 1 1 0 1 4 Back 0 1 0 1 0 1 0 0 0 1 4
Left 0 1 0 1 1 3 Forward 0 0 1 0 0 0 1 1 0 0 3
Turn to the left 1 0 1 1 1 4 Reverse 0 0 1 0 0 0 1 1 0 1 4
Right 1 1 1 1 0 4 Left 0 0 1 0 0 0 1 0 1 0 3
Turn to the right 1 0 1 1 1 4 Turn to the
0 1 0 0 0 1 0 1 0 1 4
Stop 1 1 1 1 1 5 left
Right 1 0 0 0 0 0 1 0 1 0 3
Based on the results above, there is a maximum of 2 wrong words Turn to the
0 1 0 0 0 0 0 0 0 1 2
in every five repeated words. There are 36 over 45 commands rec- right
Stop 1 0 0 1 1 1 0 0 1 1 6
ognised by the voice recognition mode. The percentage of con-
Total 3 4 3 3 2 4 4 3 4 5 90
sistency of the circuit in the silent area is 80%.
Condition 2: Noisy area
There is a maximum of 7 wrong words in every 10 speaker’s re-
This experiment was carried out in a noisy area. Five tests were
peated words. The percentage of the consistency could be higher if
performed on the circuit based on the commands listed in Table 3
the test is conducted in a silent environment with good pronuncia-
below.
tions and the words are spoken at a moderate pace. This test proved
Table 3: The Voice Commands in the Silent Area
that there is an accuracy and efficiency of the system in a noisy area,
No. Testing however it is lower than the silent area.
1 2 3 4 5 Total
command 3.2. The touch mode
Go 1 0 1 1 0 3
Back 0 1 0 0 1 2 After implementation of the smart wheelchair and its functionality
Forward 1 1 0 0 1 3 was tested, it is found that the movement of the wheelchair using
Reverse 0 0 1 1 0 2 the touch mode revealed an excellent functionality in all directions.
Left 1 0 1 0 1 3
The android application is perfectly designed in order for the user
Turn to the left 1 0 1 0 1 3
Turn to the right 1 1 0 0 0 2 to control the wheelchair.
Right 0 1 1 0 1 3
Stop 1 1 0 1 1 4
 International Journal of Engineering & Technology 457

References
Table 6: The Touch Mode by Button Command
Button com- Left Right Condition of the wheel-
mand wheel wheel chair [1] Karen Rispin and Joy Wee, “Comparison between Performances of
↑ Forward Forward Move forward Three Types of Manual Wheelchairs Often Distributed In Low-Re-
↓ Reverse Reverse Reverse source Settings”, Journal Disability and Rehabilitation: Assistive
← Stop Forward Turn to the left Technology, Volume 10, Issue 4 (2015) pp. 316-322.
→ Forward Stop Turn to the right [2] Rory A. Cooper and Arthur Jason De Luigi, “Adaptive Sports Tech-
O Stop Stop Stop nology and Biomechanics: Wheelchairs”, Original Research Para-
lympic Sports Medicine and Science, Volume 6, Issue 8 (2014) pp.
31-39. https://doi.org/10.1016/j.pmrj.2014.05.020.
In order to test the efficiency of this system, the experiments were [3] Sari Abdo Ali, Khalil Azha Mohd Annuar, Muhammad Fahmi Mis-
tested with an unload condition and 4 different weights of people kon, “Trajectory Planning For Exoskeleton Robot By Using Cubic
with the same travelled distance. And Quintic Polynomial Equation”, International Journal of Applied
Engineering Research, Volume 11, Issue 13, (2016) pp. 7943-7946.
Table 7: The Time Consumed for Different Load Conditions [4] S.A. Ali, K.A.M. Annuar, M.F. Miskon, M.H. Harun and M.F.M.A.
Condition Distance (m) Time consumed (s) Halim “Design and Control Leg-exo Robot for Rehabilitation Pur-
unload 6 6.43 pose”, Proceedings of Innovative Research and Industrial Dialogue
<40kg 6 7.00 2016 (2017) pp. 13-14.
40-45kg 6 7.20 [5] Liao Lu, Ping Yi Deng, Ying Wu, Jie Jun Bai, Yun Xiao Zhang, Yi
45-50kg 6 7.30 Xiang, Liang Jin Shi and Rusen Yang, “Control System of Powered
>50kg 6 7.45 Wheelchairs Based on Tongue Motion Detection”, International
Journal of Software Science and Computational Intelligence, Vol-
ume 8, Issue 4, (2016) pp. 60-76.
The time consumed for the wheelchair to move to the desired direc-
https://doi.org/10.4018/IJSSCI.2016100104.
tion depends on the weight of the load. Thus, when the weight of [6] M. F. Ruzaij, S. Neubert, N. Stoll and K. Thurow, "A speed compen-
the load increases, the time consumed for the wheelchair to move sation algorithm for a head tilts controller used for wheelchairs and
to the destination increases too. The load with more than 50kg has rehabilitation applications," 2017 IEEE 15th International Sympo-
the highest time consumed and the wheelchair with no load has the sium on Applied Machine Intelligence and Informatics (SAMI),
lowest time consumed. Herl'any, 2017, pp. 000497-000502.
https://doi.org/10.1109/SAMI.2017.7880360.
[7] Satish Kumar, Dheeraj, Neeraj, and Sandeep Kumar, “Design and
2. Conclusion Development of Head Motion Controlled Wheelchair”, International
Journal of Advances in Engineering & Technology, Vol. 8, Iss. 5,
This wheelchair system is combination of mechanical, electrical (Oct 2015) pp. 816-822.
[8] Rana Mohammad Yousef, Omar Adwan, and Murad Abu-Leil. “An
and communications system. The main objectives were to design
Enhanced Mobile Phone Dialler Application for Blind and Visually
an android application that can direct the movement of a wheelchair, Impaired People”, International Journal of Engineering & Technol-
to develop the voice recognition mode and touch mode to help the ogy, Volume 2, Issue 4, (2013) pp. 270-280.
elderlies and physically disabled people to move their wheelchairs https://doi.org/10.14419/ijet.v2i4.1101.
independently and to provide the elderlies and physically disabled [9] Khalil Azha Mohd Annuar, Md Zin, Muhammad Haikal, Mohamad
people with the ability to control the movement of the wheelchairs Haniff Harun, Mohd Ab Halim, Mohd Firdaus, Arman Hadi Azahar,
by using android smartphones. The system designed has undergone “Design and Development of Search and Rescue Robot” Interna-
a few tests and successfully completed the basic performance. The tional Journal of Mechanical & Mechatronics Engineering, Volume
16, No. 02 (2016) pp. 36-41.
objectives were achieved as the software and hardware implemen-
[10] Cerejo, R., Correia, V. & Pereira, N., “Eye Controlled Wheelchair
tation work well as expected. This system will helps the elderlies Based On Arduino Circuit”, 3(6), (2015) pp.94–98.
and physically disabled people to control wheelchairs with either a [11] Nishimori, M., Saitoh, T. & Konishi, R., “Voice Controlled Intelli-
touch mode or voice recognition mode, therefore this success is to gent Wheelchair” SICE Annual Conference 2007, (2007) pp.336–
serve many people with disabilities. From the conducted research, 340.
it can be seen clearly that a mobile controlled wheelchair will have [12] Pajkanovic, A. & Dokic, B., “Wheelchair Control by Head Motion”,
a bright future. It should be continued and developed in the future Serbian Journal of Electrical Engineering, 10(1), (2013) pp.135–151.
as it has a huge potential to improve its performance, reliability and https://doi.org/10.2298/SJEE1301135P.
[13] Srishti, Prateeksha Jain, Shalu, Swati Singh., " Design and Develop-
safety. For future work, it is suggested to use a more powerful and
ment of Smart Wheelchair using Voice Recognition and Head Ges-
lighter weight motor to support various weights of users. Besides, ture Control System”, International Journal of Advanced Research in
this system needs a lot of enhancements to improve its accuracy and Electrical, Electronics and Instrumentation Engineering, Vol. 4, Is-
functionality. This can be further improved by decreasing the time sue 5, (2015) pp.4790-4798.
delay in voice mode and an obstacle sensor can be attached to avoid [14] C. Swaroop, S. Sabarinath, Mohammed Akramali, “Embedded
the collision of wheelchairs. Based Smart Wheel Chair with Voice Recognition”, Journal of Ad-
vanced Engineering Research, Volume 2, Issue 2, 2015, pp.65-68.

Acknowledgement
The authors would like to thanks for the support given to this
research by Ministry of Higher Education Malaysia and Universiti
Teknikal Malaysia Melaka (UTeM) for support this under
RAGS/1/2015/TK0/FTK/03/B00118 research grant.