NOVATEUR PUBLICATIONS

INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING RESEARCH AND TECHNOLOGY

[IJIERT] ISSN: 2394-3696 Website: ijiert.org
VOLUME 8, ISSUE 9, Sep. -2021

SMART BLIND STICK FOR DETECTION AND AVOIDANCE OF

OBSTACLE
Amith S Bharadwaj
Software Engineer, Sakha Global, Bangalore, India
[email protected]

Ganavi J.
Software Engineer, Bangalore, India
[email protected]

Dr. R. Rajkumar
Assistant Professor, Dept of ISE, RNSIT, Bengaluru, India
[email protected]

ABSTRACT
Eye sight plays a major role in collecting most of the information from the real world and that information
will be processed by brain. Blindness or visual impairment is a condition that affects many people around
the world. Visually impaired people suffer inconveniences in their daily and social life with respect to
challenges during commuting. This condition leads to the loss of the valuable sense of vision. Worldwide,
there are millions of people who are visually impaired, where many of them are blind. The need for
assistive devices was and will be continuous. There is a wide range of navigation systems and tools
existing for visually impaired individuals. The blind person truly requires an aid in identifying objects.
Smart Blind Stick is an interactive device which mainly aims at helping the blind to navigate easily and in
a safer manner. In a normal day to day situation a blind person waves the blind stick ahead of them in
order to check for any objects or obstacles. The smart stick helps them in this by detecting if any obstacle
is blocking the path being taken by the subject. The device detects the obstacle with the help of a camera
attached to the front of the stick. On detection of the obstacle, it is identified and appropriate instructions
are provided to the user. The instructions to the blind person are sent over earphones. Thus, using the
various technologies, the stick provides a safer and a better navigation experience for the visually
challenged.

Keywords: Image Processing, Machine Learning, IoT, Computer Vision.

INTRODUCTION
In computer science, digital image processing is the use of computer algorithms to perform image
processing on digital images. As a subcategory or field of digital signal processing, digital image
processing has many advantages over analog image processing. It allows a much wider range of
algorithms to be applied to the input data and can avoid problems such as the build-up of noise and signal
distortion during processing. Since images are defined over two dimensions (perhaps more) digital image
processing may be modeled in the form of multidimensional systems.
Blind stick or white cane is introduced to blind people after the First World War as a mobility tool to
detect the obstacles in the path of the user. This paper proposes a smart obstacle finding stick for visually-
impaired people, which helps a blind person by detecting the obstacles using Ultrasonic sensors, a camera
and a Raspberry pi. The main objective of this is to help a blind people to move more freely by informing
the blind person about the circumstances & present condition of the path where he/she is walking using a
reliable stick.

PROBLEM STATEMENT
Existing System: Majority of them are using a conventional white cane to aid in navigation. The limitation
in white cane is that the information’s are gained by touching the objects by the tip of the cane. The
traditional length of a white cane depends on the height of user and it extends from the floor to the person’s
sternum.
66 | P a g e
 NOVATEUR PUBLICATIONS
INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING RESEARCH AND TECHNOLOGY
[IJIERT] ISSN: 2394-3696 Website: ijiert.org
VOLUME 8, ISSUE 9, Sep. -2021
Limitations of Existing System
Existing systems have the following limitations.
 Very Expensive
 Difficulty to identify the object
 No security of the stick

Proposed System
The proposed system architecture (Fig 2.a) will utilize the features of microcontrollers, machine learning
and IoT for its core functionalities.

Figure 1: Proposed System Architecture.

Advantages of the Proposed System

The proposed system has the following advantages over the existing systems.
 Low Cost
 Very Handy
 Instruction is passed on through earphones
 Camera to detect the obstacle ahead
 Easy Navigation

SYSTEM DESIGN
The use case of the Smart Blind Stick is as shown in Figure 2. The stick is used by the blind person to
navigate around in an environment by avoiding the obstacles. The stick is mounted with a Raspberry Pi
setup. Once the obstacle is recognized, audio instruction over earphones is given to the user and asking him
to slow down the two ultrasonic sensors on either side of the stick are triggered on detection of an obstacle.
67 | P a g e
 NOVATEUR PUBLICATIONS
INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING RESEARCH AND TECHNOLOGY
[IJIERT] ISSN: 2394-3696 Website: ijiert.org
VOLUME 8, ISSUE 9, Sep. -2021
These are used mainly for re-directing the user. An obstacle detection approach based on Cascaded
Convolution Neural Network will be designed. The approach will achieve a better performance by
cascading three different Convolution Neural Networks with high accuracy.

Figure 2: Use Case of a Smart Blind Stick

REQUIREMENTS
Hardware Requirements
 Raspberry Pi
 Pi camera
 Three Ultrasonic sensors
 Bread board
 Power bank

Software Requirements
 OpenCV
 Tensorflow
 PuTTY
 VNC Viewer

IMPLEMENTATION
The camera is attached in the middle part of the stick and is connected to the microprocessor. It captures
the images at regular time intervals and converts the rgb (red, green, blue) frame into a grayscale frame.
The ultrasonic sensors judge the distance of the obstacle by the time taken by the ultrasonic signal to
strike the obstacle and return. An obstacle detection approach based on Cascaded Convolution Neural
68 | P a g e
 NOVATEUR PUBLICATIONS
INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING RESEARCH AND TECHNOLOGY
[IJIERT] ISSN: 2394-3696 Website: ijiert.org
VOLUME 8, ISSUE 9, Sep. -2021
Network will be designed. The approach will achieve a better performance by cascading three different
Convolution Neural Networks with high accuracy. The data flow is as shown in the Figure 3.

Figure 3: Data Flow Diagram of Smart Blind Stick

The Convolution Neural Network will have a certain robustness to the obstacle and illumination. At the
backend, a Python IDE will be used to run the Machine Learning applications. The Keras API, using the
TensorFlow library as a backend, will be used to design the Convolution Neural Network (CNN) that will
be trained in obstacle recognition.

Pseudocode 1
prot= "MobileNetSSD_deploy.prototxt.txt"
mod= "MobileNetSSD_deploy.caffemodel"
os.system("--prototxt "+ prot +" --model " + mod)

The captured image is passed to the MobileNetSSD model for detection of obstacle.

Pseudocode 2
GPIO.setmode(GPIO.BOARD)
GPIO.setup(PIN_TRIGGER, GPIO.OUT)
GPIO.setup(PIN_ECHO, GPIO.IN)
Once an obstacle is detected in the captured image, the ultrasonic sensors are triggered to find the distance
to the detected obstacle. This information is utilized to then provide information to the user to avoid the
obstacle.
69 | P a g e
 NOVATEUR PUBLICATIONS
INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING RESEARCH AND TECHNOLOGY
[IJIERT] ISSN: 2394-3696 Website: ijiert.org
VOLUME 8, ISSUE 9, Sep. -2021
Pseudocode 3
text='Checking for obstacle'
tts=gTTS( text , lang='en')

The Google Text To Speech(gTTS) package is used to convert the text msg to audio which is conveyed to
the blind person over earphones.

RESULT/ TESTING ANALYSIS

Unit Testing
This test case checks if the three ultrasonic sensors being used are working accurately, as shown in Table 1.

Table 1: Unit testing results for ultrasonic sensors

Sl.No Name of Item being Sample Expected output Actual output Remarks
test tested Input
1 Proximity Ultrasonic Place the Respective 16.2,21.8,35.6 Pass
sensing sensor obstacle distances
test ahead (16.2,21.8,35.6)

These below test cases check as to whether the object detection module is working accurately as shown in
Table 2.
Table 2: Unit testing results to check Object Detection
Sl.No Name of Feature Sample Expected Actual Remarks
test being tested Input output output
1 Object Image Person as Person Person Pass
detection processing obstacle ahead ahead
2 Object Image Chair as Chair Chair Pass
detection processing obstacle ahead ahead
3 Object Image Bottle as Bottle Bottle Pass
detection processing obstacle ahead ahead

Validation Testing
These below test cases are for the various possibilities possible on detection of an obstacle shown in Table
3.
Table 3: Validation Test Cases
Sl.No Name of test Feature Sample Expected Actual Remarks
being tested Input output output
1 Validation Re-routing Obstacle to Instruction Instruction Pass
test directions the left of to re-route to to re-route to
the blind the right the right
person
2 Validation Re-routing Obstacle to Instruction Instruction Pass
test directions the right of to re-route to to re-route to
the blind the left the left
person
3 Validation Re-routing Obstacle to Instruction Instruction Pass
test directions the either to re-route to re-route
side of the by taking by taking
blind person two steps to two steps to
the back the back

CONCLUSION
This project is designed to create a system using ultrasonic sensors and providing voice command through
headphone to the blind people. It would help a visually impaired person navigate through a public place
independently. The proposed system tries to eliminate the faults in the previous system. It aims to solve the
70 | P a g e
 NOVATEUR PUBLICATIONS
INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING RESEARCH AND TECHNOLOGY
[IJIERT] ISSN: 2394-3696 Website: ijiert.org
VOLUME 8, ISSUE 9, Sep. -2021
problems faced by the blind people in their daily life. The system also takes measures to ensure their safety.
The design Smart Blind Stick using ultrasonic sensors and with voice output is of great benefit to blind
people when it comes to independent mobility. The advantage of the system lies in the fact that it can prove
to be very low-cost solution to millions of blind persons worldwide. Text-to-Speech conversion is used to
provide voice command as output. The blind person can easily navigate from one place to another as we
are providing voice message.

REFERENCES
1) Divyank Yarravarapu, Tina Manghnani, Artificial Vision for the Blind using ICane Electronic Aid for
Blind People, IOSR Journal of Electronics and Communication Engineering (IOSR-JECE), Nov.-Dec
.2016.
2) Giva Andriana Mutiar1, Gita Indah Hapsari, Ramanta Rijalu, Smart Guide Extension for Blind Cane,
2016 Fourth International Conference on Information and Communication Technologies (ICoICT).
3) Radhika R, Payal G Pai, Rakshitha S, Rampur Srinath, Implementation of Smart Stick for Obstacle
Detection and Navigation, International Journal of Latest Research in Engineering and Technology
(IJLRET), May 2016.
4) Akhilesh Krishnan, Deepakraj G, Nishanth N, Dr. K.M. Anandkum, Autonomous Walking Stick for the
Blind using Echolocation and Image Processing, International Conference on Contemporary
Computing and Informatics (ic3i), 2016.
5) Rishabh Gulati, GPS Based Voice Alert System for the Blind, International Journal of Scientific &
Engineering Research, Vol. 2, Issue 1, January 2011.

71 | P a g e