Blind Stick

SMART BLIND STICK WITH OBJECT DETECTION
AND ASSISTANCE
Project Reference No.: 281_45S_BE_4353
College : The Oxford College of Engineering, Bengaluru -560068

Branch : Department of Information Science and Engineering
Guide(s) : Prof. Mr. Yadhukrishna M R
Student(S) : Mr. Bhishek N
Mr. Ananda Theertha V
Ms. Anuhya Y
Ms. Nagavarshitha C H
Keywords:
Convolutional Neural Networks, Machine Learning, Tensor Flow, Open CV.
Introduction
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. Ultrasonic sensors are used for proximity detection of the
obstacle. The stick vibrates on approaching an obstacle. This adds to the safety of the
blind person. The appropriate instructions to the blind person is given over Bluetooth
earphones. Thus using the various technology, the stick provides a safer and a better
navigation experience for the visually challenged.
Objective
A blind stick is a tool that helps the blind people to navigate and move around in public
areas. The blind person is expected to move the stick in front of him to ensure no obstacle
is in his way. When the stick comes in contact with an obstacle, it is an indication for the
blind person that he needs to change the path he was moving in. The smart blind stick
aims at giving the blind person a better understanding of the path he is moving in. We
achieve this by using a camera and a few ultrasonic sensors. The stick, with the help of a
camera, detects the obstacle the blind person is approaching.
The smart blind stick aims at giving the blind person a better understanding of the path he
is moving in. We achieve this by using a camera and a few ultrasonic sensors. The stick,
with the help of a camera, detects the obstacle the blind person is approaching. This
obstacle is detected and analysed and this data is retrieved on an android device. The
data received, which is the obstacle detected, is converted to speech and appropriate
instructions are given over Bluetooth earphones to the blind person.
The ultrasonic sensors placed at either side of the stick are used to detect the obstacles
present in the sides. This helps us in giving a more efficient instruction to the blind person
as to how he can evade the obstacle in front of him in a safe manner. The data of the
ultrasonic sensors are manipulated using an Arduino UNO.
With the help of the data processed using these technologies, we aim at providing a better
and more convenient method of navigation for the blind person. This stick helps them
achieve a higher rate of self-dependency. The system gives a priority to safety and comfort
and is also cost effective.
Methodology
1. Obstacle detection:
Object detection is the main module of the project. The module mainly aims at finding the
right models for detection of the obstacle. There are multiple algorithms that can be used
to detect the presence of the obstacle. These models are loaded on to the Raspberry Pi.
The obstacle detection actually takes place on the Pi.
2. Obstacle Recognition:
This is the second step once the detection of the obstacle is completed. The models used
for the recognition can be trained to recognize custom objects that are required by us. The
training is done by feeding the network of the model with enough images of the required
object, which allows the m0del to learn from it. Once the model has learnt from the training
data, it makes prediction on an image or a video fed provide by us. The model can also be
used to refer obstacle in real time as well.
3. Distance sensing:
Ultrasonic sensors (HCSR04) are used for distance sensing. The Ultrasonic sensors have
an echo and a trigger. The trigger Pin sends out an ultrasonic signal out on starting the
sensor. The signal on encounter of an obstacle, returns the signal. This signal is captured
by the echo terminal of the USs. Based on the time taken by the signal to travel from the
trigger and back after hitting the obstacle, is used to calculate the distance. This distance
is used to trigger the obstacle detec6 and obstacle recognize modules on the stick
4. Text to Speech:
This is a small but a vital module of the project. This is used to convert the text messages
to speech. These instructions are given over an earphone to the user. The google API
GTTS (Google Text-to-Speech) is used for this purpose.
5. Obstacle avoidance:
This is the final module of the project. This mainly deals with the instructions given to the
user based on the detected obstacle and also the distance to the closest obstacle on
either side to re-route. These are converted to speech using gtts. These instructions are
given to the blind person over earphones.
Figure: System Architecture
Results and Conclusions
The figures shows the Pi camera detecting obstacles that its detecting in the initial run with
the following confidence percentages. The outcomes of test results for a variety of user
interactions with the application are seen below:
Figure: Object detection

Figure: Car object detection
Figure: Potted plant detection under night light

The next figure shows the model being loaded into the Raspberry Pi for the first time as it
begins object detection with the action and reaction time in detection being calculated and
displayed. The front sensor is responsible for detection of an object.
Scope for future work:

The stick can be enabled with GPS which can help the blind for better navigation. The
smart blind stick can be trained for more number of objects which in turn would help the
blind person to move around in various neighborhoods with increased level of safety. In
the future, the stick can be used for face detection. This increases the safety of the blind
person in knowing the identity of the person in front of him.

Blind Stick

Uploaded by

Copyright:

Available Formats

Blind Stick

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Blind Stick

Uploaded by

Copyright:

Available Formats

SMART BLIND STICK WITH OBJECT DETECTION

College : The Oxford College of Engineering, Bengaluru -560068

Convolutional Neural Networks, Machine Learning, Tensor Flow, Open CV.

Figure: Object detection

Figure: Potted plant detection under night light

Scope for future work:

You might also like