Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 
 

135 
 

 

 

 

 

 

DESIGN AND IMPLEMENTATION OF WIRELESS VOICE 

CONTROLLED MOBILE ROBOT 

Dr. Ali Ahmed Abed 

College of Engineering- University of Basrah 

aaad_bah@yahoo.com 

Dr. Abbas A. Jasim 

College of Engineering- University of Basrah 

abbas.a.jasim@ieee.org 

Received 26 August 2015        Accepted 21 January 2016 

 

ABSTRACT 

This paper presents a technique for a speech recognizer used to control the motion of an intelligent 

automated mobile robot. The aim is to interact with the mobile robot using natural and direct 

communication techniques. The voice is processed to get proper and safe movement of a mobile robot 

and satisfying high recognition rate. Features are extracted from speech signal using Mel Frequency 

Cepstral Coefficients (MFCC). To realize feature matching, an efficient  Dynamic Time Warping 

(DTW)-based speech recognition system is presented which is applicable for isolated words of Arabic 

language. The tested words are compared to a trained database using this DTW algorithm. On the other 

side, the mobile robot is designed with two servo motors as driving actuators. These actuators are 

controlled by L298 motor driver circuit. The control algorithm is programmed and downloaded into a 

PIC18F45K22 microcontroller which is interfaced to a USB port of a 10" notebook computer. The 

robot proves a capability of understanding the full meaning of the five Arabic speech commands that 

steer it forward, backward, right, left, or stop. 

 

Keywords: Arabic speech recognizer, Mel Frequency Cepstral Coefficients, dynamic time warping,  
Pattern Recognition , Mobile robot. 

 

 تصميم وتنفيذ روبوت متحرك السلكي مسيطر عليه بالصوت

علي احمد عبدد.   

جامعة البصرة-كلية الهندسة  

عباس عبد االمير جاسمد.   

جامعة البصرة-كلية الهندسة  

 

 الخالصة

البحث طريقة لبناء مميز كالم يستخدم للسيطرة على حركة روبوت متحرك آلي وذكي يستطيع التفاعل وفهم لغة الكالم الطبيعية  يقدم 

آمنة  بصورة مباشرة. يقدم البحث الخطوات التفصيلية الالزمة لمعالجة االشارة الصوتية بما يضمن نسبة تمييز عالية تؤدي الى حركة

تستخدم  DTWوخوارزمية معتمدة على  MFCCوطبيعية للروبوت. الخوارزميات المستخدمة للمعالجة الصوتية هي: خوارزمية 

دة لتمييز الكلمات العربية المنفصلة. تعتمد عملية التمييز على مقارنة الكلمات االختبارية مع الكلمات المدربة مسبقاً والمخزونة في قاع

وذلك  L298تساق بواسطة مسيطر نوع  servoجانب آخر، فقد تم بناء روبوت متحرك ثنائي المحركات من نوع  بيانات مسبقة. من

لتحقيق خوارزمية السيطرة الصوتية. تمت موائمة الروبوت مع مميز الكالم من خالل معالج )مايكروكونترولر( نوع 

mailto:aaad_bah@yahoo.com
mailto:abbas.a.jasim@ieee.org


Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 

136 
 

PIC18F45K22  ًبأكملها من خالل توجيه الروبوت  تم اختبار المنظومة كدائرة بينية صممت بشكل كامل لهذا الغرض. واخيرا

 .على المسار المطلوبوجيه الروبوت باستخدام خمس كلمات عربية هي: امام، خلف، يمين، يسار، قف والتي بواسطتها يمكن ت

 روبوت المتحرك، الاالنماط مميز كالم عربي، معامالت تردد ميل، معامالت الوقت الديناميكي، تمييزالكلمات المفتاحية: 

 

NOMENCLATURE 

ANN 

Dist(x,y) 

DTW 

F 

FFT  

F mel  

GD 

HMM 

K 

L 

LD 

LPC 

M 

MFCC 

RR 

V 

X 

Y 

Artificial Neural- Network  

Euclidean distance between two points 

Dynamic Time Warping 

Tone frequency in Hz 

Fast Fourier Transform  

Mel frequency in Hz 

Global Distance 

Hidden Markov models 

Number of frames 

Number of samples in each frame 

Local Distance 

Linear Predictive Coding  

Number of samples that separated frames 

Mel Frequency Cepstral Coefficients  

Recognition Rate 

Voice Activity Detection 

Sequence feature vector in n dimensional space 

Another sequence feature vector in n dimensional space 

 

1. INTRODUCTION 

The Arabic language is the fifth widely used language world-wide since there are at least 200 million 

people speak Arabic, (Khalid, 2013). There are little researches in speech recognition field that deal 

with Arabic as compared to English or Japanese. The Arabic language has monosyllabic and 

polysyllabic words with two categories of phonemes: pharyngeal and emphatic, which found in all 

Semitic languages, (Al-Zabibi, 1990) and (Alkhouli , 1990). The automatic speech recognition, which 

got a good attention for many decades, allows a computer to recognize spoken words inputted by a 

mike. Speech recognizers are used in many applications such as: interacting with deaf people, 

healthcare, home automation, robotics, etc. There are a large number of approaches for speech 

recognition such as: Dynamic time warping (DTW), Artificial Neural- Network (ANN), Hidden 

Markov models (HMM), etc.  In this work, an efficient DTW-based speech recognition system for 

isolated Arabic words is given as a feature matching algorithm and a Mel Frequency Cepstral 



Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 
 

137 
 

Coefficient (MFCC) approach is used as a feature extraction approach because of its robustness and 

effectiveness compared to other well-known methods like Linear Predictive Coding (LPC), 

(Lindasalwa, 2010). After that a mobile robot is designed, as will be explained in the subsequent 

sections, and controlled by the designed speech recognizer to get a complete speech controlled system 

suitable for different applications. It is desired to command the mobile robot by voice via special 

interface that plays a significant role as a master control circuit for the servo motors of the robot. In 

voice control system, a difficulty may appear in the control circuit leading to a recognition error, which 

means that the recognized command is interpreted as opposite command. For example "Left" is 

interpreted as "Right" especially in languages with very high acoustic similarity like Polish. This 

problem is not significant in Arabic when using the direction words because they differ completely in 

pronunciation. 

Unlike other languages, Arabic language is characterized by having tremendous dialectical variety, 

diacritic text material, morphological complexity which may lead to some challenges against having a 

highly accurate Arabic recognizer. In the work of (Jean-Marc , 2007), the voice of the speaker depends 

on the distance and azimuth. The work satisfied a distance of 2m and azimuth range of 10
o
 to 90

o
.  In 

2010, two voice recognition algorithms which are MFCC and DTW is built, evaluated and compared 

with other techniques to prove their effectiveness, (Lindasalwa, 2010). In 2011, (Ahmed,2011) had 

proposed a technique called (multiredgilet transform) with neural network to control the motion of a 

wheelchair dedicated for handicapped people. The work presented by (Rachna, 2011) is to build a 

microcontroller-based mobile robot controlled with speech. He studied various factors such as noise and 

distance factor for his speech recognition system. (Khalid, 2013) suggested DTW, MFCC and voice 

activity detection (VAD) for isolated words of Arabic language but with unsufficient recognition rates. 

In our work, we built a speech recognizer using MFCC, DTW, and VAD for five Arabic words and a 

high recognition rates are satisfied without depending on azimuth and the distance is limited by the 

wireless transmission distance. This speech recognition system is used to control the motion planning of 

an autonomous mobile robot designed completely to get a voice controlled robotic system. 

The rest of the paper is organized as follows: Section 2 is concerned with the explanation of our speech 

recognition system with all its stages. In section 3, the complete design of the mobile robot with the 

used components is explained. Section 4 provides the software structure of the overall system. In 

section 5, the obtained results and verification are given with some required discussion. Section 6 

summarizes the main conclusions. 

 

2. THE VOICE RECOGNITION SYSTEM 

The presented Arabic speech recognition system consists of the following stages: 

A. Preprocessing 

This stage is important to enhance the recorded speech signal characteristics by removing noise leading 

to obtain a high quality recorded speech. The high frequency contents of the input signal are 

emphasized by a first order FIR filter (implemented in software) to flatten the signal spectrum. Also, 

this stage should overcome the problem of using different types of microphones and different speaking 

loudness. This stage is hidden in the first stage "Read the voice input" of Figure1. 

B. Voice Activity Detection (VAD) 

Another problem that affects the performance of the speech recognizer is detecting the start and end 

points of the voice signal, (Khalid, 2013). The speech signal is segmented into spaced frames of 10ms 



Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 

138 
 

width. After that, short-term power and zero-crossing rate are used to detect the speech/non-speech 

regions. It is clear that short-term power is increased in speech regions while zero-crossing rate is 

increased in non-speech regions. Hence, these two techniques give a good indication of speech 

appearance. 

C. Feature Extraction 

 Framing: The speech signal is segmented into K frames of L samples for each one. The adjacent 
frames are separated by M samples (M<L). In this work, values of L=256 and M=87 samples 

are chosen. 

 

 Hamming Window: It is applied to the above framed signal to reduce the discontinuity at both 
ends of each frame. 

 

 Fast Fourier Transform (FFT): It is used to convert the above windowed signal from time 
domain to frequency domain to prepare it for the next stage. 

 

 Mel Filter Bank: Since human hearing is less sensitive to frequencies higher than 1KHz, a Mel-
scale is used so that for each tone with a frequency F (in Hz), a subjective pitch is measured on a 

Mel-scale according to the following equation, (Plannerer, 2005): 

 

                                                                                                        (1) 

 

After calculating of the magnitude of the resulting FFT signal and using the Mel-scale filter 

bank, which consists of 24 triangular band pass filters having an equal spacing before 1 KHz and 

logarithmic scale after 1 KHz, the Mel spectrum coefficients are found as the summation of the 

filtered results. 

 

 Inverse Discrete Cosine Transform (IDCT): It is used to return back to time domain. But before 
that, the logarithm of the magnitude of the output of Mel-filter bank is computed since logarithm 

compresses the dynamic range of values.  

 

 Liftering: is a filtering in the spectrum domain that used to extract the vocal tract cepstrum. 
Some cepstrum coefficients at the end can be dropped. One can use the first 12 coefficients for 

each frame and neglect the others. 

D. Database Collection 

A feature database of spoken Arabic words for pattern matching process is created and stored in 

computer memory as a collection of template vectors. Feature database is built by recording several 

utterances for each one of the five target words collected from different speakers (male and female) 

during the training process. After each target word is recorded, preprocessing, VAD, and feature 

extraction processes are performed and the final feature vector pattern is obtained and stored in the 

database. For each of the adopted Arabic control target words, fifteen patterns, template vectors were 

created in this manner. Thus the training database of feature vectors contains 75 patterns arranged as 

five cluster one for each control word. All recording sessions are done under normal environment with 

8 kHz sampling frequency and 8 bit per sample mono channel. 



Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 
 

139 
 

E. Feature (Pattern) Matching  

Pattern matching process is performed in the runtime phase of the proposed system to find the template 

pattern that best fit to the spoken word. Pattern matching is done in online manner. The test spoken 

word is recorded in the same environment used for training phase. The test feature vector is then found 

after applying preprocessing, VAD, and feature extraction steps. Test feature vector is compared with 

all template vectors in the training database vector by vector. The best fit template vector is selected. 

The target word for which the best fit template belongs in is said to be recognized.  

The time alignment of different utterances is the core problem for distance measurement in the speech 

recognizer presented in this paper. A small shift leads to incorrect identification. Dynamic Time 

Warping (DTW) is an efficient method to solve this problem, (Shivanker, 2013). DTW algorithm aims 

to align two sequences of feature vectors by warping the time axis repetitively until an optimal match is 

found. In other words, it performs a piece wise linear mapping of the time axis to align both signals. If 

two sequences of feature vector in n-dimensional space is considered: 

 

                                                                           (2) 

 
Then they are aligned on the sides of a grid. The distance between two points is obtained by the 

Euclidean distance: 
 

                                      (3) 

 
The best match or alignment between these two sequences is the path through the grid, which 

minimizes the total distance between them. The overall distance (global distance) is calculated by 

finding and going through all the possible routes through the grid, each one compute the overall 

distance. Then global distance is obtained by the following recursive equation: 

 

                                               (4) 
 

GD=Global (Overall) distance, LD=Local (Euclidean) Distance. 

 

3. THE MOBILE ROBOT SYSTEM 

A. The Robot Vehicle 

The used vehicle is a new breed of tracked robot chassis designed specifically for mobile robot projects, 

(Rover 5, 2013) as shown in Figure 2. It has two actuated wheels and two passive wheels used for 

balance. The clearance can be adjusted by rotating the gearboxes in 5-degree to maintain tension as 

required. An optical quadrature encoder that gives 1000 pulses over 3 revolutions of the output shaft is 

embedded. It has two motors and encoders and 4 noise suppression coils. 

B. The Vehicle Motors and Motors' Driver 

The two motors used in this car are servo motors with the following specifications: motor rated voltage 

is 7.2V, stall current is 2.5A, output shaft stall torque is 10Kg/cm, and speed is 1Km/hr. The Motor 

driver is based on the L298, which is a dual H-bridge driver designed to drive inductive loads such as 



Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 

140 
 

DC and stepping motors, (User's Guide, 2012). It lets you drive two DC motors, controlling the speed 

and direction of each one independently. The connection diagram of the driver with the two motors and 

a real photo of it are shown in Figure 3. 

C. The Robot Controller Circuit 

The brain of the robot that is considered as a decision making element and controls the motion planning 

of the robot is the microcontroller. PIC18F45k22, (Microchip Technology Inc., 2012), is one of the 

most popular microcontrollers that could be adopted in robotic systems. In this project, we used the 

EasyPIC7 development board, (User's Guide Mikroelektronika, 2013), which will save effort and 

time for building a complete controller and interface circuit. It has an RS232 port for UART serial 

communication with the computer as shown in Figure 4. Since there is no RS232 serial port in our 

notebook computer, a USB-to-RS232 converter is used. To facilitate the serial communication between 

the speech recognizer designed with MATLAB and the microcontroller-based interface circuit of the 

mobile robot, the following code is embedded within the speech recognizer: 

command=?; 

s=serial('com1'); 

fopen(s); 

fwrite(s,command); 

fclose(s); 

delete (s); 

Where: "command" is the value that corresponds to the desired direction given acoustically by a person. 

"com1" is the serial port number available, which may take other numbers depending on the computer.  

D. The Microphone 

 A quality microphone is the key when utilizing automatic speech recognition (ASR). In most cases, a 

desktop microphone is not sufficient because they may pick up more ambient noise that gives ASR 

programs a hard time. Hand held microphones are also not the best choice as they can be cumbersome 

to pick up all the time. While they do limit the amount of ambient noise, they are most useful in 

applications that require changing speakers frequently. The best choice and the most common is the 

headset style. It allows the ambient noise to be minimized, while allowing us to have the microphone at 

the tip of our tongue all the time. Logitech Wireless Headset H600 (shown in Figure 4) is found 

experimentally to be the best choice, so it is used to control the robot while it is moving. The headset 

will send the voice orders to the robot as a response to a low sound tone that tell us when one can start 

speaking. This headset consists of the following parts: Nano USB receiver plugin in the notebook 

computer, Noise-canceling microphone, 2.4 GHz wireless transmitter, Simple on-ear control, and six-

hour rechargeable battery. 

E. The NoteBook Computer 

It is a 10 inches laptop computer with USB ports for serial communication with the microcontroller of 

the robot and for the nano receiver of the headset. A MATLAB R2010 is installed to be the 

environment of the speech recognizer that takes the input voice from our wireless mike and produces 

the corresponding commands outputted via the USB port to the interface circuit reaching to the 

microcontroller which in turn interprets these commands and produces the suitable action for the motors 

to begin the planned motion. The block diagram of the overall computer control hardware system is 

shown in Figure 5. Of course, this laptop has a relatively large size the matter that may restrict the 

robot usage to some applications but it is desirable to other applications that need a computer controlled 

robots such as moving robots in space (Que, 2011), and industrial robots (Yoshioka , 2014). 

http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Yoshioka,%20T..QT.&newsearch=true


Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 
 

141 
 

4. THE OVERALL SOFTWARE STRUCTURE 

Voice recognition works based on the premise that a person voice exhibits characteristics that are 

unique among different speakers. The signal during training and testing sessions can be greatly different 

due to many factors such as: people voice change with time, health condition, speaking rate and also 

acoustical noise and variation recording microphone. To explain the software stages starting from voice 

input to commands output, a flowchart is shown in Figure 1. The microcontroller should be 

programmed in order to receive the commands from the serial port and use them to actuate (drive) 

motors accordingly. MicroC is a powerful, feature rich development tool for PIC micros designed to 

provide an easy solution for developing applications in embedded systems and control, (User's Manual 

of Mikroelektronika, 2012). The C program downloaded inside our PIC is given below: 

char uart_rd; 

void main() { 

ANSELC = 0; 

UART1_Init(9600); 

Delay_ms(100); 

TRISB = 0x00; 

LATB = 0x00; 

while (1) { 

if (UART1_Data_Ready()) { 

uart_rd = UART1_Read(); 

 LATB =uart_rd; }}} 

 

5. RESULTS AND DISCUSSION 

In order to evaluate the performance of the presented speech recognizer, recorded samples are separated 

into training and testing sets. The recognition rate of each Arabic word can be calculated by the 

following equation, (Khalid, 2013): 
 

 *100%                             (5) 

Each word of the five Arabic words is trained fifteen times to provide sufficient number of template 

vectors for four different cases of experimental work. Hence, the number of stored templates is 75. In 

the test phase, each of the five words is tested at run time of the mobile robot that operates 30 times. 

Mobile robot motion with respect to the corresponding voice command is monitored and recognition 

accuracy (recognition rate) is computed. Table 1 shows the RR for a sample of tested words in the 

database which have been already recorded. While Table 2 shows the elapsed time required to 

perform the matching process for four different cases use 3, 5, 10, or 15 template vectors for each 

word. From Table 1, three template feature vectors per target word give poor recognition rate. RR is 

significantly increased with 5 template vectors. Using 10 template vectors give very good RR. 

Increasing the number of template vectors to 15 led to little increasing in the recognition rate as 

shown in Figure 6. On the other side, the time needed for DTW to perform the matching process for 

the four defined cases seems to vary in approximately linear form as shown in Figure 7. From the 

held experiment with results given in Table 1 and Table 2, using ten template feature vectors per 

word gives very good RR with moderate time. The input voice waveforms of the five spoken words 

are shown in Figures 8-12. 

Now, the presented speech system is applied to the motion of the designed mobile robot which is 

tested with the five Arabic words in Table 1. It is possible with a series of voice commands to steer 



Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 

142 
 

the robot along a desired trajectory to its target. Also, controlling the voice commands leads to control 

motion and avoid obstacles in the way of the robot. To satisfy a linear path for the robot, only the 

isolated words "امام" or "خلف" is used (both wheels are rotated clockwise or anti-clockwise) as shown 

in Figure 13. Circular path of the robot is obtained by a single isolated voice command "يسار" or 

 according to the desired direction. Figure 14 shows circular motion of the robot in clockwise "يمين"

direction using voice command "يمين" in which right motor is "off" and left motor is "on". 

 

6. CONCLUSIONS 

Using a simple and efficient automatic speech recognition technique for isolated Arabic words to satisfy 

the motion planning of mobile robots is the interest of this paper. The extracted features are stored in a 

.mat file using MFCC algorithm. The experiment results are analyzed with the aid of MATLAB and 

proved good efficiency especially when applied with the designed mobile robot. This process can be 

extended for a number of speakers. Also, the work proved that the DTW is a sufficient nonlinear feature 

matching technique in speech identification with minimal error rates and fast computing speed the 

matters that are very important in robotics. Generating database stores feature vectors to be matched 

with run time test words rather than storing the whole speech signal saved memory space and made 

matching process very fast. The processing units (the notebook and the microcontroller) are directly 

attached to the mobile robot in one package that made the design representing a complete autonomous 

robot. Voice commands are given using wireless microphone results in flexible movement of the robot. 

The range of movement of the implemented mobile robot is only restricted by the range of the wireless 

microphone. The physical range of movement can be extended simply by speaker movement because 

there is no fixed station is used for any processing purpose. 

 

REFERENCES 

 

Ahmed Q. A., 2011, Controlled of mobile robots by using speech recognition, Journal of Babylon 

University, Applied Science, Vol. 3, issue 19. 

 

Alkhouli M., 1990, Alaswaat Alaghawaiyah, Dar Alfalah, Amman, Jordan, (Arabic reference). 

 

Al-Zabibi M., 1990, An acoustic-phonetic approach in automatic Arabic speech recognition, The 

British Library in Association with UMI, UK. 

 

Jean-Marc V., Shun’ichi Y., Jean R., Francois M., Kazuhiro N., and Hiroshi G., 2007, Robust 

recognition of simultaneous speech by a mobile robot, IEEE Transactions on Robotics, Vol.23, No.4. P. 

742-752. 

 

Khalid A.D., Ala F., Iyad F., Baraa A., and Saed W., 2013, Efficient DTW-Based speech recognition 

system for isolated words of Arabic language, World Academy of Science, Engineering and 

Technology, Vol. 7, P. 106-113, Iraq. 

 

Lindasalwa M., Mumtaj Begam and I. Elamvazuthi, 2010, Voice recognition algorithms using Mel 
frequency cepstral coefficient (MFCC) and dynamic time warping (DTW) techniques, Journal of 

Computing, Vol.2, issue 3, ISSN 2151-9617. 



Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 
 

143 
 

Microchip Technology Inc., 2012, Low power, high performance microcontrollers. 

 

Plannerer B., 2005, An Introduction to speech recognition, Bernd Plannerer, Munich, Germany. 

 

Que D., Jian Yang ; Bo Wei ; Hui Li ; Zhihong Jiang ; Danfeng Li ; Hongjie Li ; Qiang Huang, 

2011, A method on trajectory plan for humanoid space robot , Robotics and Biomimetic (ROBIO), 
IEEE International Conference, P. 281-286. Thailand. 

 

 

Rachna J., and Saxena S., 2011, Voice automated mobile robot, International Journal of Computer 

Applications (0975-8887), Vol. 16, No. 2, India. 

 

Rover 5, available at:   https://www.sparkfun.com/products/10336 

 

Shivanker D., Geeta N., and Poonam P., 2007, Isolated speech recognition using MFCC and DTW, 

International Journal of Advanced Research in Electrical, Electronics and Instrumentation Engineering, 

Vol.2, Issue 8, 2013, ISO 3297: certified organization, India. 

 

User's Guide, 2012, L298 dual H-bridge motor driver. 

 

User's Manual, 2012, MicroC: C compiler for Microchip PIC microcontrollers, Mikroelektronika, 

Belgrade. 

 

User's Guide, 2013, EasyPIC7", Mikroelectronika, Belgrade. 

 

Yoshioka, T. ; Shimada, N. ; Ohishi, K. ; Miyazaki, T. ; Yokokura, Y., 2014, Link-coupled 

vibration suppression control considering product of inertia for industrial robots, Advanced Motion 

Control (AMC), IEEE 13th International Workshop,  P. 675 – 680, Yokohama,  Japan. 

 

 

Table (1): Recognition rates for different feature sets 

Tested  

Arabic words 

Transcription English 

writing 

          RR 

3      5    10    15 

 امام
Amam 

Forward 43    67   90    93 

 Khalf Backward 40    63   80    80 خلف

 Yemeen Right 46    73    86   90 يمين

 Yesar Left 53     83    90  93 يسار

 Qif Stop 46     67    83  86 قف

 

http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Jian%20Yang.QT.&searchWithin=p_Author_Ids:37538613900&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Bo%20Wei.QT.&searchWithin=p_Author_Ids:38236767200&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Hui%20Li.QT.&searchWithin=p_Author_Ids:37407161800&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Zhihong%20Jiang.QT.&searchWithin=p_Author_Ids:37405774600&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Danfeng%20Li.QT.&searchWithin=p_Author_Ids:37535499900&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Hongjie%20Li.QT.&searchWithin=p_Author_Ids:38240925400&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Qiang%20Huang.QT.&searchWithin=p_Author_Ids:37279982900&newsearch=true
http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=6181299&queryText%3Dspace+robot
https://www.sparkfun.com/products/10336
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Yoshioka,%20T..QT.&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Shimada,%20N..QT.&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Ohishi,%20K..QT.&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Miyazaki,%20T..QT.&newsearch=true
http://ieeexplore.ieee.org/search/searchresult.jsp?searchWithin=p_Authors:.QT.Yokokura,%20Y..QT.&newsearch=true
http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=6815878
http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=6815878


Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 

144 
 

Table (2): Feature matching time for different cases 

No. templates Comparison Time (msec) 

3 5.8 

5 9.4 

10 15.2 

15 23.7 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 Figure (1): The overall software system 

 

 

 



Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 
 

145 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Power 
Control 

lines  

Figure (3): Motors driver connection diagram and its real photo 

 

 

 

 

 

 

 

 

 

Figure (2): The robot chassis 

 

 

 

 

 

Figure (4): The overall designed listening robot with its main parts 



Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 

146 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure (8): The input voice signal of "امام" word 

 

 

 

 

Figure (9). The input voice signal of "خلف" word 

 

 

 

 

 
Figure (5): The overall hardware system 

 

 

 

 

 

 

 

 

Left motor 

of robot 

Right motor 

of robot 

RS232 

Figure (6): Relationship between RR and  

no. of template vectors 

 

 

 

 

 

 

 

Figure (7): Relationship between time and 

no. of template vector 

 

 

 

 

 

 

 



Al-Qadisiyah Journal For Engineering Sciences,         Vol. 9……No. 2 ….2016 
 

 
 

147 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 1000 2000 3000 4000 5000 6000 7000 8000
-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

Figure (10): The input voice signal of "يسار" word 

 

Figure (11): The input voice signal of " ميني " word 

 

Figure (12): The input voice signal of "قف" word 

 

Figure (13): Linear path navigation of the mobile robot 

 

Figure (14): The circular path navigation of the mobile robot 

 

y 

X