D

ISO 9001:2008 Certified | MSME | FMSCI

Autonomous Vehicle with Smart Control

Ch . Mallikarjuna Rao
ID: 150040147
Branch : ECE

Company Guide : Ms. Kalpna Kumari

Designation : Training Department Head
University Guide :Mr.M.Ravi Kumar
Designation/Department :Assoc. Professor ECE
ABSTRACT
• Autonomous vehicles are very famous now a day’s. In this project we develop a
autonomous vehicle prototype. We use Machine learning concepts to achieve
autonomous driving. We initially train the vehicle manually through remote access
using internet. During this training we obtain data from sensors regarding object
distance around the vehicle at every instance of time and current direction of the
vehicle. Later we feed this data into machine learning algorithms and develop a
classifier which predicts the directions for new sensor data using previous
experience.The objective of this innovation is to develop a autonomous vehicle which
can be controlled through internet with on board diagnostic systems. This system has
the ability to solve the road accident, traffic issues and provides more comfortable
ride to its passengers.
.
Fortnight Duration Work Done Remarks
12-12-2018 To 13-04-2019 ISIE INDIA 85% Work Completed
What is navigation and why we need to navigate?

What if we automate this system? Is it ever

happened before? If so, what advantages it brought
to the society?
What is Data acquisition and we need it?
• Data is information
• Here we mainly deals with sensors data and store them to apply further processing.
• The data may be analog or digital.
• Some of the example of data are
• Temperature of the room
• GPS points
• Nitrogen content of the soil
• Cell count of the blood
• LIDAR data.
Why we developed this autonomous rover?
• To automate the process of soil sampling and to improve the health of the agricultural
field by acquiring the basic parameter like
• Moisture
• Temperature
• Air quality
• Humidity
• Soil nutrient contents like nitrogen, potassium.

• We also compare the results with our database and judge whether the soil is in good
condition to a specific crop. If not, we suggest the farmer to improve the health of the
field.
Technologies we are using to achieve our goal
• Machine learning
• Artificial neural networks
• LIDAR
• Global positioning system
• Google maps
How we navigate autonomously?
• We use machine learning algorithm SVM where train the rover to navigate
autonomously.
• For example we take distances at every 30 degrees angle of the ULTRA SONAR. And
class denoting the state of the rover.
• We consider four states to the rover
• State 1: forward
• State 2: backward
• State 3: right
• State 4:left
Sample distance vectors and corresponding classes
Distance Vectors from ultrasonar and corresponding state of the rover
Clock wise sweep of servo Anti Clock wise Sweep of servo
class 0° 30° 60° 90° 120° 150° 180° 180° 150° 120° 90° 60° 30° 0°
1 51 51 51 51 51 51 51 51 51 51 51 51 51 51
3 51 51 51 51 22 19 51 51 51 51 18 19 51 51
1 51 51 51 51 51 51 31 25 25 26 51 51 51 51
1 51 51 51 51 51 51 51 51 51 51 51 51 51 51
3 51 51 51 51 21 20 51 51 51 51 51 25 51 51
1 51 51 51 51 51 51 26 27 27 27 51 51 51 51
1 51 51 51 51 51 51 51 51 51 51 51 51 51 51
3 51 51 51 51 24 22 51 51 51 51 21 22 51 51
1 51 51 51 51 51 26 25 25 26 26 30 51 51 51
3 51 51 51 51 51 51 26 24 24 24 51 51 51 51
1 51 51 51 51 51 51 51 25 25 26 51 51 51 51
1 51 51 51 51 51 51 51 51 51 51 51 51 51 51
3 51 51 51 51 20 20 21 51 51 24 21 51 51 51
1 51 51 51 51 51 51 27 27 27 27 51 51 51 51
3 51 51 51 51 51 51 17 16 16 17 51 51 51 51
1 51 51 51 51 51 51 51 22 21 51 51 51 51 51
1 51 51 51 51 51 51 51 51 51 51 51 51 51 51
3 51 51 51 51 19 18 20 51 51 51 20 51 51 51
1 51 51 51 51 51 51 26 25 25 25 51 51 51 51
1 51 51 51 51 51 51 51 51 51 51 51 51 51 51
3 51 51 51 51 19 17 18 51 51 21 18 51 51 51
1 51 51 51 51 51 51 25 22 21 22 51 51 51 51
1 51 51 51 51 51 51 51 51 51 51 51 51 51 51
3 51 51 51 51 24 21 51 51 51 51 20 23 51 51
1 51 51 51 51 51 51 28 27 28 29 51 51 51 51
Other alternate algorithms we can use:
• Naïve Bayes
• Random forest
• KNN
• Neural networks
• PCA for dimensionality reduction
Alternate real time applications
Use Case Description

• In this project we want to develop a autonomous vehicle which can also be

• controlled using internet. The main feature of the vehicle is
1) It can drive itself without assistance from driver
2) It can be controlled with an application in both pc and mobile phone using internet.
3) It has a smart accident avoiding system where it applies brakes itself when it predicts an accident
4) The whole system can be diagnosed visually using the remote application.
5) It can predict the damaged systems itself and gives driver alert about the damage.
6) It can interact with its passengers with smart artificial intelligence technology where it can play songs and
talk to the humans.
7) A driver can set the system in train mode which automatically helps the system to learn new driving skills.
Architecture
.
• A micro controller is connected to processor through serial communication and all the
• sensor data is collected by the controller and sent to the processors to reduce the load on
• processor. Processor process the sensors data and accordingly control the physical system through
• relays.
• This system design is based on closed loop control system. The two major feed back
• systems are speed, temperature of motor and vehicle dynamics like tilting, orientation and
• direction and nearest object distance. By using these feed back parameter the processor decides to
• increase or decrease the vehicle speed, steering or apply brakes accordingly
• Feedback paths:
• Speed , Temperature, Tilting , Orientation and Direction are major feedback systems to the vehicle
• by using these parameter the processor controls the motor drive.
Software Architecture
Physical Prototype Front view
Applications:
1. Autonomous vehicles take quick decision on roads in the situations where response time is very
less. This avoids accidents on road and save thousands of people.
2. The travel time or shipment time is much predictable than conventional driving system. The
navigation from start to destination points become clearer if we connect whole system into
centralized or connected computation.
3. By autonomous vehicles we can navigate autonomously anywhere and get data like temperature,
humidity and climate conditions etc., One of the best example for autonomous navigation are
agriculture rovers. These rovers navigate through out the field and get crucial data elements like
soil fertility, temperature, moisture which helps us understand overall crop health and also
individual areas of fields.
4. We can use autonomous navigation and data acquisition techniques in the field of military and
army applications to get tracks of enemies and their movements.
Conclusion
• By this project we demonstrated the different approach to develop autonomous vehicle using
machine learning algorithms. We also compared two major machine learning algorithms SVC and
Naive bayes. In our observation Naive bayes has better accuracy, prediction time and training time.
We have designed a physical prototype and also acquired precious data. By using remote server we
reduced the processing power required for micro processor or on board vehicle which saves money
for the automobile companies.
REFERENCES

• https://www.raspberrypi.org/

• https://www.arduino.cc/

• M. U. Saleh et al., “Design and Implementation of a Simple, Low-Cost Robotic Arm,”

International Journal of Advanced Research in Electrical, Electronics and Instrumentation

Engineering, vol. 2 Issue 10, p. 46394645, 2013.

[2] C. Hernández, R. Poot, L. Narváez, E. Lianes and V. Chi, “Design and Implementation of a

System for Wireless Control of a Robot,” International Journal of Computer Science Issues, Vol.

7, Issue 5, pg 191197, September 2010.

[3] J.-D. Warren, J. Adams, and H. Molle, Arduino Robotics, New York, NY: Apress, 2011.

[4] W. Wei, Y. Pan, and K. Furuta, “Internet-based telecontrol system for wheeled mobile robot,”

IEEE, doi:10.1109/ICMA.2005.1626715, 2005.

Learning Outcome
• This project report pertains to six months industrial training that I had underwent at ISIE-Imperial
Society of Innovative Engineers, Noida as part of curriculum of degree in Bachelor of technology
in Electronics and Communication Engineering as required by K L University.
• I learnt a lot from professional managers and skilled engineers. I had a great learning experience
as trainee in this firm. I learnt a lot about how different Eco-friendly Technology, Solar and Hybrid
Technology, Robotics in the industry.
• I have learnt about different type of servers like Electric Motor, Controller, Hybrid and Solar
Powered technology. Also I have learnt about the electronics control unit in vehicles. And I was
part of Technical Committee of Eco-friendly Motorsports ISIE.
THANK YOU