DESIGN AND DEVELOPMENT

OF LEVEL-3 AUTONOMOUS
CAR

Mo, Tiande, Kin-tak Lau, Yu Li, Chi-kin Poon, Yinghong Wu, Paul K. Chu, and Yang Luo. 2022.
"Commercialization of Electric Vehicles in Hong Kong" Energies 15, no. 3: 942.
https://doi.org/10.3390/en15030942

MEHRAN UNIVERSITY
OF ENGINEERING AND
TECHNOLOGY
DEPARTMENT OF MECHATRONIC ENGINEERING JAMSHORO PAKISTAN
 TABLE OF CONTENTS
Introduction Aims and Objectives

Literature Review Methodology

Problem Statement Sustainable Development Goals

Motivation Workplan

Scope Of Project References and Conclusion

MEET OUR TEAM
GROUP LEADER
SHEERAZ 20MTE011

GROUP MEMBERS
MUFEED 20MTE029
MUHAMMAD AZAM 20MTE046

SUPERVISOR
MADAM MEMONA MEMON
 INTRODUCTION
• Level-3 autonomous cars are also known as conditionally automated vehicles
equipped with sensors that allows them to handle all driving tasks under certain
conditions. Under favorable conditions(good weather, clear lane marking, limited
traffic/organized traffic) the car can take complete control of driving, braking,
maintaining speed and following lane marking.

• Driver takeover request : if the system fails in getting the continued favorable
conditions or encounters a condition it can’t handle then, it will immediately alert
the driver to take the control of the vehicle.
• Level-3 autonomous vehicles are mainly used on highways, motorways, express
ways where the favorable conditions are usually present.
 INTRODUCTION (CONT.)

Levels of
Automation

Ref:
 LITERATURE REVIEW
Title Author Description
Date
Design and Development of 8th Manjunath, R, Kounte., C., In this paper, the author had presented a system consisting
Autonomous Driving car October V., Aswini, Shri., Vs, of a self driving car model which employs Jetracer, an AI
using Jetson Nano 2022 Harshvardhan., Ayushi, framework for autonomous cars, and apply image
Developer Kit Kumari., Shrimali, Dhruv recognition to capture traffic signs and classify them using
AI and ML frameworks and respond to them in real time
through jetson nano interface.

ATAC-Based Car-Following 1st August Tang, Tie-Qiao, Gui, Yong, The Authors developed a car following(CF) model as a
Model for Level 3 2022 Zhang, jian longitudinal control strategy as level 3 based on ATAC
autonomous Driving algorithm, in which 1641 pairs of car following were
Considering Driver’s extracted from the next generation simulation(NGSIM) data
Acceptance were applied to train the reinforcement learning model.
Simulation results showed that the proposed model can
enable vehicles to drive safely, efficiently and comfortably.
A Three-Level Game- 1st Liu, Mushuang, Wan, Yan, A three-level decision making framework is developed to
Theoretic Decision-Making November Lewis, Frank L, Nageshrao, generate safe and effective decisions for autonomous
Framework for Autonomous 2022 Subramanya, Filev, Dimitar vehicles. A neural network is developed to learn the tuning
Vehicles parameters via supervised learning. Simulation studies are
conducted to demonstrate the performance of the
developed algorithms in two-vehicle interchange-crossing
and lane-changing scenarios.
 LITERATURE REVIEW (CONT.)

Title Date Author Description

A Novel Design of 1st August B.Padmaja, P V Narasimha, Myneni The proposed system utilizes mathematical models
Autonomous Cars using 2018 Madhu Bala, E Krishna Rao Patro like Neural networks and image processing
IOT and Visual features techniques. This is implemented as three major
components: Curved road detection, road sign and
signal detection and obstacle detection(collision
avoidance).The system can attain high accuracy with
real-time constraints.
Development of an Ultra 25 January S.Mohd Redzuan, M.H.Nuraido, In this paper the authors had developed the
Sonic sensor driven 2019 O.Norzalina, A.B.Noorhelinahanii prevention system during vehicle reverse mode using
Automatic Vehicle Arduino controller, Linear actuator and ultra sonic
reverse safety system sensors which can avoid casualties and it also gives
confidence to the driver as reverse is typically worse
in larger vehicles like trucks and SUVs.
Investigation on 1st January Muhammd Aizzat Iqbal Bin Abd The author integrated the processed data from
Integration of sensors 2022 Rashid, Ahmad Shahrizan Abdul ultrasonic sensor, together with the collected data
and Vision-based Vehicle Ghani from pixy sensor to validate the accuracy of the
Detection System information received by the controller to navigate the
vehicle. The comparison of data received from pixy
camera and ultrasonic sensor is done to determine
safety status.
 LITERATURE REVIEW (CONT.)
Title Date Author Description
Artificial Neural Network 26 August Eka Nuryanto Budisusila The paper focuses on the application of artificial neural
Algorithm for Autonomous 2020 , Bustanul Arifin, network algorithm that get input in the form of signal
Vehicle Ultrasonic Multi-Sensor Sri Arttini Dwi Prasetyowat that refer to the output signal of the ultrasonic sensors,
system i which has been already assembled into a multi-sensor,
, Bhakti Yudho Suprapto, which is a eight ultrasonic sensors positioned around
Zainuddin Nawawi the vehicle supporting the autonomous vehicle system.

Application of Ultrasonic 18 January Maoyan Fan, Lifang Zhang The author had discussed about the application of
Sensors in Vehicle Collision 2019 ultrasonic sensors in a vehicle collision avoidance and
Avoidance and Early Warning early warning system. The system is based on Arduino
System. Mega microcontroller, sensors that detect obstacles,
optimize ranging algorithm and control emergency
brakes which can help the driver to acquire good
knowledge of the surrounding conditions of the vehicle
and thus being of great significance and ensuring driver
safety.
Design of Multifunctional 1st Ayesha Iqbal, The model of autonomous car presented in this paper
Autonomous car using November Syed Shaheryar Ahmed, aims to solve issues by taking humans off the wheels,
Ultrasonic and Infrared Sensors 2017 Muhammad Danish Tauqee so that they do not have to drive anymore, risk of
r accident, traffic congestion are reduced to minimal
, Ali Sultan, quantity. Ultrasonic and infrared sensors are used for
Syed Yasir Abbas distance measurement and obstacle detection. Arduino
and stepper motors are used for decision making and
sharp turn respectively.
 LITERATURE REVIEW (CONT.)

Title Date Description

Author
Design and 22 Apeksha Kamble, In this paper a 360-surrounding view of the
Implementation of September Akanksha Singh, vehicle, to obtain information about possible
Surround View 2022 Damini Thusoo, obstacles present in the blind spot area of the
Monitoring System in Amol Madane. vehicle and to assist the driver while parking, for
View Autonomous this purpose 4 fish-eye cameras were used to
system. cover the whole area. Distortion modification and
stitching method were used to generate high
quality results.
Driver Monitoring 18 February Prof. M. D. Ingle, The author has used a convolution neural
System using Machine 2023 Shubham Khedekar, network(CNN) and image processing techniques
Learning. Shubham Salunkhe, to classify the state of eyes and mouth and alert
Shailesh Bhosale the driver if fatigue or drowsiness is detected by
facial and body gesture.

Convolution Neural 11 April 2023 M. Chidanand, The author has proposed a model that considers
Network(CNN) and Facial S. L. Jany Shabu, various Facial Characteristics like PERCLOS, Eyes
Features based Smart L. K. Joshila Grace, Aspect Ratio(EAR), blink values and yawning to
Driver Alertness D. Poornima monitor the alertness of driver.
 SUMMARY
From the above research we can conclude that level 3 autonomous vehicles were developed by various
algorithms like Neural Networks, CNN and, also by different AI and ML frameworks. Ultrasonic sensors
combined with IR were also used for obstacle detection(collision avoidance) and emergency brakes. Some
models also consisted of Lidar combined with a camera for navigation of autonomous vehicle, where as
GPS and radar sensors were also used for proposed models of autonomous cars. All the data obtained
from sensors will be interfaced by microcontrollers or microprocessors like Arduino, raspberry pi, Jetson
nano.

Beyond Traditional
boundaries
Our proposed model will be different in a way that we have selected low-cost sensors and
microcontrollers for the autonomous operation to make the project low-cost, another difference is
that we will also develop a machine learning model for fog detection. If detected the proposed model
will suddenly generate a Driver Takeover Request to alert the driver for taking the control of car.
Finally, we will also develop an arena for real time operation of the level 3 car.
 MOTIVATION

• Observing enormous number of people dying everyday globally in road

accidents due to human errors like inattentiveness due to constant driving on
same patched road motivated us to develop a car that can reduce the
fatalities caused by human error.
• Considering the fuel prices and fuel consumption of normal cars we came
across an idea to develop a car that reduces the fuel consumption by
considerable amount the first think that came to our mind was of level-3
autonomous car as it consist of features like cruise control that reduces the
fuel consumption by huge amount.
• Another motivation is that level 3 cars can assist elderly or physically
challenged individuals with driving, granting them with greater independence
and mobility.
 PROBLEM
STATEMENT
“Development of an Level-3 Autonomous car is need of the time considering
the fuel prices and fatalities caused by human error while driving normal(non-
autonomous) cars.Level-1 and Level-2 cars are only good till reducing fuel bills
but when it comes to reduce the death toll of accidents caused by human
error, level 1 and 2 completely fails, Whereas level-4 and level-5 cars are quite
expensive and are not practical till now, so we are left with the option of
developing a level-3 autonomous car, most importantly it will be affordable.”
 AIM

• To develop a cost effective, safe, reliable

and functional level-3 based autonomous
car.
 OBJECTIVES

Objective 1 To Develop strategic solutions for

autonomous operations.

Objective 2 To Assemble a physical car model.

To Develop machine learning

Objective 3 models for Fog Detection and
Driver Monitoring System.

To integrate the developed strategic

Objective 4 solutions and models for real-world
environment testing.
SCOPE OF PROJECT
The project will focus on achieving the following functions:
Driver monitoring system alerting the driver if drowsiness or
fatigue is detected
Maintaining a constant speed according to the distance of the car
infront (Adaptive Cruise Control)
Fog detection
Emergency stops(incase if sudden obstacle is detected)
Blind spot monitoring
Parking assistance
 COMPONENTS LIST
Name Quantity Cost
IR sensors 4 440
Ultrasonic Sensors 4 800
Ultrasonic sensors mounting brackets 4 120
Arduino Mega 1 4500
5 channel IR reflectance sensor 1 600
6v DC geared motors 4 600
L298 Motor driver 2 1000
ESP-32 CAM module 2 2500

Car body (upper body) 1 3000

Arduino nano 1 1200
Jumper wires 100 1000
4WD car chassis with tires 1 3000
Blind Spot Monitoring Bind Spot Monitoring

PROGRESS

Bind Spot Monitoring Bind Spot Monitoring

Emergency Braking

Parking sensors
PROGRES
S
Adaptive Cruise Adaptive Cruise
Control Control

Adaptive Cruise Control

PROGRESS
PROGRES
S
•PROGRESS
•PROGRESS
PROGRESS
CODE FOR DRIVER MONITORING SYSTEM

import cv2
import dlib
import numpy as np
from imutils import face_utils
from scipy.spatial import distance as dist

# Load the pre-trained models

face_detector = dlib.get_frontal_face_detector()
landmark_predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")

# Function to calculate the Eye Aspect Ratio (EAR)

def calculate_ear(eye):
A = dist.euclidean(eye[1], eye[5])
B = dist.euclidean(eye[2], eye[4])
C = dist.euclidean(eye[0], eye[3])
ear = (A + B) / (2.0 * C)
return ear
# Function to calculate head pose
def get_head_pose(shape):
image_pts = np.float32([shape[33], shape[8], shape[36], shape[45], shape[48], shape[54]])
model_pts = np.float32([
(0.0, 0.0, 0.0), # Nose tip
(0.0, -330.0, -65.0), # Chin
(-225.0, 170.0, -135.0), # Left eye left corner
(225.0, 170.0, -135.0), # Right eye right corner
(-150.0, -150.0, -125.0), # Left Mouth corner
(150.0, -150.0, -125.0) # Right mouth corner
])

focal_length = 1
camera_matrix = np.array(
[[focal_length, 0, 0.5],
[0, focal_length, 0.5],
[0, 0, 1]], dtype = "double"
)

dist_coeffs = np.zeros((4,1))
success, rotation_vector, translation_vector = cv2.solvePnP(model_pts, image_pts, camera_matrix, dist_coeffs)

rmat, _ = cv2.Rodrigues(rotation_vector)
angles, _ = cv2.decomposeProjectionMatrix(rmat)
pitch, yaw, roll = [np.asscalar(a) for a in angles]
return pitch, yaw, roll
# Main driver monitoring loop
cap = cv2.VideoCapture(0) # Replace with the appropriate video source

while True:
ret, frame = cap.read()
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
faces = face_detector(gray)

for face in faces:

shape = landmark_predictor(gray, face)
shape = face_utils.shape_to_np(shape)

# Calculate EAR for both eyes

left_eye = shape[face_utils.FACIAL_LANDMARKS_IDXS["left_eye"]]
right_eye = shape[face_utils.FACIAL_LANDMARKS_IDXS["right_eye"]]
left_ear = calculate_ear(left_eye)
right_ear = calculate_ear(right_eye)
ear = (left_ear + right_ear) / 2.0
# Eye closure detection
if ear < EAR_THRESHOLD:
# Implement eye closure duration tracking here
pass

# Head pose estimation

pitch, yaw, roll = get_head_pose(shape)

# Display head pose angles

cv2.putText(frame, f"Pitch: {pitch:.2f}", (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255,
0), 2)
cv2.putText(frame, f"Yaw: {yaw:.2f}", (10, 60), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0),
2)
cv2.putText(frame, f"Roll: {roll:.2f}", (10, 90), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0),
2)

cv2.imshow("Driver Monitoring System", frame)

if cv2.waitKey(1) & 0xFF == ord('q'):

break
 METHODOLOGY
Motor Connections
Problem Statement with controlller

Literature Review Sensor Selection Car Chassis attachment

Model with motors

Software solutions
for autonomous
operations
Integration of
Development of machine
sensors
learning model for fog
detection on
expressways

Development of a vision-
based driving monitoring Development of a
system with the help of real-world type Testing
pre-trained models. arena /environment
From the above-mentioned SDG’s we have selected 5 Sustainable Development Goals
for our project .

Level 3 cars have the potential Improved traffic flow and

to significantly reduce road reduced congestion making
traffic accidents by eliminating transportation more accessible
human error, also the driver can and sustainable within cities.
take the eyes off the
road(driving) thus contributing
to “Good Health and Well-
Being”.
Development and Deployment of By reducing traffic congestion
level 3 cars can contribute to and fuel bills, thus the green
innovations in transportation house gas emissions are
infrastructure and can enhance lowered mitigating climate
overall industry efficiency. change impacts.

Improved Accessibility of
transportation through
Autonomous Vehicles, especially
for the elderly or disabled can
foster equality and social
inclusion.
 WORK PLAN
NOV DEC JAN FEB MARCH APRIL MAY JUNE JULY AUG SE OCT
P
Literature
Review

Development of
software solutions for
autonomous operations

Development of
software solutions for
autonomous operations

Assembly of 4wd car

model

Development of a
machine learning
model for fog detection
on express ways

Development of a
driver monitoring
system

Development of Arena
for real time operation

Testing
 CONCLUSION
This project aims to contribute to the advancement of Level-3 autonomous driving
technology by demonstrating its feasibility using Arduino and sensor technology. By
overcoming the presented challenges, this project has the potential to contribute to:

• Reduced cost of developing autonomous vehicles.

• Increased accessibility of autonomous driving technology to a wider range of
researchers and developers.
• Further advancement of algorithms and approaches for Level-3 autonomous driving.
• The project's success will pave the way for further research and development in this
exciting field, potentially leading to safer, more efficient, and more convenient
transportation for all.
REFERENCES
• [1] Manjunath, R, Kounte., C., V., Aswini, Shri., Vs, Harshvardhan., Ayushi, Kumari., Shrimali, Dhruv. (2022).
Design and Development of Autonomous Driving Car Using NvidiaJetson Nano Developer Kit. Available from:
10.1109/ICCCMLA56841.2022.9989127

• [2] (2022). ATAC-Based Car-Following Model for Level 3 Autonomous Driving Considering Driver’s Acceptance.
IEEE Transactions on Intelligent Transportation Systems, Available from: 10.1109/tits.2021.3090974

• [3] Mushuang, Liu., Yan, Wan., Frank, L., Lewis., Subramanya, Nageshrao., Dimitar, Petrov, Filev. (2022). A Three-
Level Game-Theoretic Decision-Making Framework for Autonomous Vehicles. Available from:
10.1109/TITS.2022.3172926

• [4] B., Padmaja., P, V, Narasimha, Rao., Myneni, Madhu, Bala., E, Krishna, Rao, Patro. (2018). A Novel Design of
Autonomous Cars using IoT and Visual Features. Available from: 10.1109/I-SMAC.2018.8653736
 REFERENCE
S
• [5] S.Mohd, Redzuan., M.H.Nuraida., O., Norzalina., A.B., Noorhelinahani. (2019). Development Of Ultra Sonic
Sensor Driven Automatic Vehicle Reverse Safety System. International Journal of Scientific & Technology
Research.

• [6] Muhammad, Aizzat, Iqbal, Bin, Abd, Rashid., Ahmad, Shahrizan, Abdul, Ghani. (2022). Investigation on
Integration of Sensors and Vision-Based Vehicle Detection System for Autonomous Vehicle. Available from:
10.1007/978-981-33-4597-3_71

• [7] Eka, Nuryanto, Budisusila., Bustanul, Arifin., Sri, Arttini, Dwi, Prasetyowati., Bhakti, Yudho, Suprapto.,
Zainuddin, Nawawi. (2020). Artificial Neural Network Algorithm for Autonomous Vehicle Ultrasonic Multi-Sensor
System. Available from: 10.1109/EECCIS49483.2020.9263459

• [8] Maoyan, Fan., Lifang, Zhang. (2019). Application of Ultrasonic Sensors in the Vehicle Collision Avoidance and
Early Warning System. DEStech Transactions on Engineering and Technology Research, Available from:
 REFERENCES

• [9] Ayesha, Iqbal., Syed, Shaheryar, Ahmed., Muhammad, Danish, Tauqeer., Ali, Sultan., Syed, Yasir, Abbas.
(2017). Design of multifunctional autonomous car using ultrasonic and infrared sensors. Available from:
10.1109/ISWSN.2017.8250023

• [10] Kamble, A., Singh, A., Thusoo, D., Madane, A.R. (2023). Design and Implementation of Surround View
Monitoring System in View of Autonomous Vehicle. In: Goswami, S., Barara, I.S., Goje, A., Mohan, C., Bruckstein,
A.M. (eds) Data Management, Analytics and Innovation. ICDMAI 2022. Lecture Notes on Data Engineering and
Communications Technologies, vol 137. Springer, Singapore. https://doi.org/10.1007/978-981-19-2600-6_36

• [11] Prof., M., D., Ingle., Shubham, Khedekar., Shubham, Salunkhe., Shailesh, Bhosale. (2023). Driver Monitoring
System Using Machine Learning. Proceeding international conference on science and engineering, Available
from: 10.52783/cienceng.v11i1.119

• [12] (2023). Convolutional Neural Networks (CNN) and Facial Features based Smart Driver Alertness Detection.
Available from: 10.1109/icoei56765.2023.10125701