SkullCap – An IoT based Smart Helmet for Accident Detection - Report

MAR BASELIOS INSTITUTE OF TECHNOLOGY AND SCIENCE
NELLIMATTOM P.O, KOTHAMANGALAM, KERALA
APJ ABDUL KALAM TECHNOLOGICAL UNIVERSITY
THIRUVANANTHAPURAM
PROJECT REPORT PHASE II – JUNE 2023
SKULLCAP – AN IOT BASED SMART HELMET FOR ACCIDENT
DETECTION
Submitted by
ARJUN SAJI (MBI19CS015)
BASIL PAPPY ROY (MBI19CS018)
BASIL VARGHESE (MBI19CS019)
GEEVARGHESE S ISAAC (MBI19CS030)
In partial fulfilment for the award of the degree
Of
BACHELOR OF TECHNOLOGY
In
COMPUTER SCIENCE AND ENGINEERING

i
MAR BASELIOS INSTITUTE OF TECHNOLOGY AND SCIENCE
NELLIMATTOM P.O, KOTHAMANGALAM, KERALA
DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
BONAFIDE CERTIFICATE
This is to certify that the project report on “SKULLCAP – AN IOT BASED SMART
HELMET FOR ACCIDENT DETECTION” is a bonafide record of the project
presented by ARJUN SAJI (MBI19CS015), BASIL PAPPY ROY (MBI19CS018), BASIL
VARGHESE (MBI19CS019), GEEVARGHESE S ISAAC (MBI19CS030) in partial
fulfilment for the award of the degree of Bachelor of Technology in Computer Science and
Engineering of APJ Abdul Kalam Technological University, Thiruvananthapuram during
the academic year 2022-2023
Project Guide Project Cordinator Head of the Department
Dr. Suji Pramila
Assosiate Professor
Department of CSE
Ms. Teena Skaria
Assistant Professor
Department of CSE
Mr. Midhun Mathew
Assistant Professor
Department of CSE

ii
ACKNOWLEDGEMENT
First and the foremost, we thank ALMIGHTY GOD who gave us the inner
strength, resource and ability to complete our work successfully, without which all our
efforts would have been in vain.
We express our sincere gratitude to Dr. P Sojan Lal, Principal for providing us
the research facilities and the circumstances to complete the project in a grant successful
manner.
We express our deep sense of gratitude to Mr. Midhun Mathew, Assistant
Professor & HOD, Department of CSE for his insightful comments, encouragements, and
critical inquiries, which inspired us to widen our project from various perspectives. We
would like to extend our gratitude to Dr.Suji Pramila, Associate Professor CSE
Department for the creative suggestions, valuable advice, mental support and guidance as
Project Coordinator.
Words are insufficient to express our heartfelt and sincere gratitude to our Project
Guide who have suggested a relevant and interesting project topic and paved the way to the
successful completion of this project. With great respect and gratitude, we acknowledge
our guide Ms. Teena Skaria, Assistant Professor, CSE Department for her invaluable
guidance, great inspiration, and strong moral support throughout our research work. We are
indebted to her for the valuable time she has spared for us during this work.
We use this opportunity to thank all faculties in the department and the friends who
have helped us to complete the project, with their inspiration and co-operation.
Arjun Saji
Basil Pappy Roy
Basil Varghese
Geevarghese S Isaac

iii
ABSTRACT
In order to make driving safer and prevent injuries, riders wear helmets as protective
headgear. In this study, a smart helmet is created that can detect a rider's fall and send an
alert via a GSM module, coupled with the fall's location determined by a GPS device, to
prevent delays in rescue efforts. The helmet is equipped with inertial sensors, such as
gyroscope and accelerometer sensors with a three-axis sensing unit, which are controlled
by an ARDUINO-based microprocessor to detect falls. In 28 experiments, the developed
helmet is employed for data collection, and 22 statistical aspects are gleaned from the
gathered information. Using the wrapper technique, a subset of distinguishing
characteristics is chosen, and four different classifiers—Naive Bayes, k-nearest neighbor,
random forest, and support vector machine—are then applied to the chosen features to
categories the two states, fall and non-fall. The Nave Bayes classifies rider falls with the
maximum accuracy, 98.21%, as shown by the data. The suggested method beats current
cutting-edge fall detection methods in terms of classification precision.

iv
TABLE OF CONTENTS
Pages
CERTIFICATE i
ACKNOWLEDGEMENT
ABSTRACT
TABLE OF CONTENTS
LIST OF FIGURES
LIST OF TABLES
LIST OF ABBREVIATIONS
1. INTRODUCTION
1.1 Background and Motivation
1.2 Objectives
2. LITERATURE REVIEW
2.1 Introduction
2.2 Existing Topologies
2.2.1 Automatic Fall Detection Using Smartphone Accelerometer
2.2.2 Combined Smart Watch and Smart Phone Fall Detection System
2.2.3 Smart Helmet an Intelligent Bike System
2.2.4 Fall Detection for Elderly Persons Using Android-Based Platform
2.2.5 SPEEDY: A Fall Detector in a Wrist Watch
ii
iii
iv
vi
vii
viii
1
2
3
4
4
4
5
7
7

v
2.2.6 Android Based Fall Detection Alert System using multi-Sensor
2.2.7 Falls, Injuries Due to Falls, and the Risk of Admission to a Nursing
Home
2.2.8 A Dynamic Motion Pattern Analysis Approach To F&l Detection
2.2.9 A Smartphone-based Fall Detection System
2.2.10 Recognition of False Alarms in Fall Detection Systems
3. DRAWBACKS OF EXISTING SYSTEM
3.1False alaram
3.2 Cost
3.3 User compliance
3.4 Limited accuracy
3.5 Privacy concerns
3.6 Dependence on technology
4. PROPOSED SYSTEM
4.1 Helmet Design
4.2 Data Acquisition
4.3 Feature Extraction and Selection
4.4 Classification
5. SYSTEM MODULE
5.1 GSM Module
5.2 GPS Module
5.3 Arduino Microcontroller
8
9
9
10
10
13
11
11
11
11
11
11
12
12
13
13
15
14
14
14
14

vi
5.4 Battery 15
5.5 Accelerometer 15
6. ARCHITECTURE
6.1 Naive Bayes (NB) 16
6.2 K-Nearest Neighbor (KNN) 16
6.2 Random Forest (RF) 16
7. SYSTEM IMPLEMENTATION
8. SYSTEM REQUIREMENTS
8.1 Hardware Requirements 19
8.2 Software Requirements 19
9. DATA FLOW DIAGRAM 20
10 WORK PLAN AND TASK ALLOCATION 25
10.1 Phase 1: Planning and research 22
10.2 Phase 2: Hardware assembly and testing 22
10.3 Phase 3: Software development and testing 22
10.4 Phase 4: Integration and testing 22
11. CONCLUSION AND FUTURE WORK 23
11.1 Conclusion
11.2 Scope of Future Researches
REFERENCES 26
APPENDIX A – SAMPLE CODING 28
APPENDIX B-SCREENSHOTS 31

vii
LIST OF FIGURES
Fig 4.1 Proposed methodology for riders fall detection using inertial
sensors embedded in the helmet
14
Fig 9.1 Level-0 Data Flow Diagram of Proposed System 23
Fig 9.2 Level-1 Data Flow Diagram of Proposed System 24

viii
LIST OF TABLES
Table 10.4 Workplan and Task Allocation 26

ix
LIST OF ABBREVIATIONS
GPS Global Positioning System
JPEG Joint Photographic Experts Group
ML Machine Learning
AI Artificial Intelligence
OS Operating System
DFD Data Flow Diagram
IP Image Processing
GSM Global System for Mobile Communication
Wi-Fi Wireless Fidelity
IDE Integrated Development Environment

SkullCap – An IoT based Smart Helmet for Accident Detection
DEPARTMENT OF CSE, MBITS 1
CHAPTER 1
INTRODUCTION
1.1 BACKGROUND AND MOTIVATION
The most important topic in every element of our lives is safety and security. A traffic
accident claims the lives of 1.35 million people every year, according to a World Health
Organization (WHO) data from 2018. Today, there are many incidents of people losing
their lives in two-wheeler traffic accidents. The data indicates that falls are the second most
common cause of fatal accidents and injuries. Injuries resulting from accidental falls
account for more than 2.8 million visits to the emergency room each year, more than 8
million hospital admissions, and 27,000 fatalities. The fall poses a serious risk to
motorcycle and bicycle riders. Millions of individuals have perished worldwide as a result
of riding-related falls. Currently, if a person has an accident, we may not be able to provide
immediate first aid treatment since there is no fall detection device and we are not aware
of the accident's location. As a result, the person may pass away as a result of this delay.
Therefore, the fall detection technique is crucial for saving lives. Thousands of lives could
be saved if the fall injuries could be treated quickly. The mechanism for creating a reliable
fall detection system to aid bikers was proposed in this research. The health care of children
and the elderly, as well as the need to provide them with prompt aid when they encounter
a fall, are the key factors attracting researchers' attention to fall detection.
They need to be rescued in time. There are many mechanisms already introduced for the
fall detection of pedestrians and bicyclists. They include wearable devices, smartphone-
based, wireless-based, ambient sensor-based, vision-based, and floor sensor/electric field
sensor-based fall detection systems. In a two units wireless-based system is developed. One
is a helmet unit, and the other is a bike unit. Both the units communicate with the help of
RF transmitter fitted in the helmet. Bike unit has an onboard accelerometer which senses
the free fall of the bike. In a vision-based fall detection system is proposed. A camera
records the observer’s activity and by the use of image processing techniques, a fall is
detected. The proposed method is a computationally intensive method. Some researchers
have used smartphone built-in accelerometer and gyroscope with other sensors to sense the

fall and sending alert messages for emergency response in case of fall. The position of the
smartphone or wearable sensors in these studies vary.
1.2 OBJECTIVES
This paper presents a fall detection framework using a helmet that has a wearable sensor
MPU 6050 with on chip accelerometer and gyroscope. These sensors have the capability
to track 3-D human motion. Accelerometer is used to track the linear acceleration, whereas,
gyroscope sensor is utilized to measure orientation and angular velocity. These smart
inertial sensors are commonly used for fall detection, activity recognition, and posture
identification. These sensors are fitted inside the helmet of the bikers along with other
components, including the SD-card module to record the data from the sensor. Introducing
some essential features in the helmet made it more smart and helpful for the riders. After
the design of helmet, data acquisition is performed to analyse the performance of the
system. A human dummy is built, and the helmet is fitted on the head.

CHAPTER 2
LITERATURE REVIEW
2.1 INTRODUCTION
There has been a significant amount of research on the use of inertial sensors for fall
detection of riders. Inertial sensors, such as accelerometers and gyroscopes, can provide
information on the motion and orientation of a rider's head, which can be used to detect
falls. One study used an accelerometer and a gyroscope to detect falls in motorcycle riders.
The sensors were placed in a smart helmet, and machine learning algorithms were used to
classify falls based on the data from the sensors. The results of the study showed that the
proposed fall detection system had a high accuracy rate and could effectively detect falls
in real-time. Another study used a similar approach, but with the addition of a
magnetometer to provide information on the orientation of the helmet. The results of this
study showed that the inclusion of the magnetometer improved the accuracy of the fall
detection system. Other studies have explored the use of inertial sensors for fall detection
in other contexts, such as falls in elderly individuals. These studies have also shown
promising results in terms of the accuracy of the fall detection systems. Overall, the use of
inertial sensors for fall detection of riders appears to be a promising approach, with the
potential to improve the safety of riders. It is important to continue researching and refining
these systems to improve their accuracy and reliability.

2.2 EXISTING TOPOLOGIES
2.2.1 Automatic Fall Detection Using Smartphone Accelerometer
There has been a significant amount of research on the use of smartphone
accelerometers for automatic fall detection. Smartphone accelerometers are small devices
that can measure the acceleration and orientation of a smartphone, and this information can
be used to detect falls. One study used a smartphone accelerometer to detect falls in elderly
individuals. The researchers collected data from falls and non-fall activities, and used
machine learning algorithms to classify the data. The results of the study showed that the
proposed fall detection system had a high accuracy rate and could effectively detect falls
in real-time. Another study used a similar approach, but with the addition of a gyroscope
to provide additional information on the motion of the smartphone. The results of this study
showed that the inclusion of the gyroscope improved the accuracy of the fall detection
system. Other studies have explored the use of smartphone accelerometers for fall detection
in other contexts, such as falls in athletes. These studies have also shown promising results
in terms of the accuracy of the fall detection systems. Overall, the use of smartphone
accelerometers for automatic fall detection appears to be a promising approach, with the
potential to improve the safety of individuals. It is important to continue researching and
refining these systems to improve their accuracy and reliability.
2.2.2 Combined Smart Watch and Smart Phone Fall Detection System
There has been research on the use of combined smart watch and smartphone fall detection
systems. These systems use the accelerometers and other sensors in both devices to detect
falls. One study used a smart watch and a smartphone to detect falls in elderly individuals.
The accelerometers in both devices were used to collect data on the motion and orientation
of the devices during falls and non-fall activities. Machine learning algorithms were then
used to classify the data and detect falls. The results of the study showed that the proposed
fall detection system had a high accuracy rate and could effectively detect falls in real-time.
Another study used a similar approach, but with the addition of a gyroscope and a
magnetometer to provide additional information on the motion and orientation of the
devices. The results of this study showed that the inclusion of these additional sensors
improved the accuracy of the fall detection system. Other studies have explored the use of
combined smart watch and smartphone fall detection systems in other contexts, such as

falls in athletes. These studies have also shown promising results in terms of the accuracy
of the fall detection systems. Overall, the use of combined smart watch and smartphone fall
detection systems appears to be a promising approach, with the potential to improve the
safety of individuals. It is important to continue researching and refining these systems to
improve their accuracy and reliability.
2.2.3 Smart Helmet an Intelligent Bike System
A smart helmet is a type of bicycle helmet that is equipped with sensors and other
technology to provide additional functionality beyond traditional helmets protection. One
possible application of a smart helmet is as an intelligent bike system, which can provide
riders with information and assistance to enhance their cycling experience. Some possible
features of a smart helmet as an intelligent bike system include:
• Fall detection: A smart helmet can use inertial sensors, such as accelerometers and
gyroscopes, to detect falls and alert emergency services or a designated contact.
• Navigation: A smart helmet can include a GPS device and display to provide turn-
by-turn directions to help riders navigate to their destination.
• Communication: A smart helmet can include a microphone and speaker to allow
riders to make phone calls or listen to music while cycling.
• Performance tracking: A smart helmet can track a rider's performance, such as
distance travelled, speed, and calories burned, and provide feedback to help riders
improve their cycling.
• Safety features: A smart helmet can include additional safety features, such as lights
or reflectors, to improve visibility and reduce the risk of accidents.
Overall, a smart helmet as an intelligent bike system has the potential to enhance the cycling
experience and improve the safety of riders.
2.2.4 Fall Detection for Elderly Persons Using Android-Based Platform
There has been a significant amount of research on the use of Android-based
platforms for fall detection in elderly persons. Android-based platforms, such as

smartphones and tablets, can be equipped with sensors and other technology to detect falls
and alert caregivers or emergency services. One study used an Android-based smartphone
to detect falls in elderly persons. The smartphone was equipped with an accelerometer and
a gyroscope, and machine learning algorithms were used to classify the data from the
sensors to detect falls. The results of the study showed that the proposed fall detection
system had a high accuracy rate and could effectively detect falls in real-time. Another
study used a similar approach, but with the addition of a magnetometer to provide
information on the orientation of the smartphone. The results of this study showed that the
inclusion of the magnetometer improved the accuracy of the fall detection system. Other
studies have explored the use of Android-based platforms for fall detection in other
contexts, such as falls in athletes. These studies have also shown promising results in terms
of the accuracy of the fall detection systems. Overall, the use of Android-based platforms
for fall detection in elderly persons appears to be a promising approach, with the potential
to improve the safety and well-being of elderly individuals. It is important to continue
researching and refining these systems to improve their accuracy and reliability.
2.2.5 SPEEDY: A Fall Detector in a Wrist Watch
There has been research on the use of wrist watches for fall detection. Wrist watches
can be equipped with sensors, such as accelerometers and gyroscopes, to detect falls and
alert caregivers or emergency services. One study used a wrist watch to detect falls in
elderly individuals. The watch was equipped with an accelerometer and a gyroscope, and
machine learning algorithms were used to classify the data from the sensors to detect falls.
The results of the study showed that the proposed fall detection system had a high accuracy
rate and could effectively detect falls in real-time. Another study used a similar approach,
but with the addition of a magnetometer to provide information on the orientation of the
watch. The results of this study showed that the inclusion of the magnetometer improved
the accuracy of the fall detection system. Other studies have explored the use of wrist
watches for fall detection in other contexts, such as falls in athletes. These studies have also
shown promising results in terms of the accuracy of the fall detection systems. Overall, the
use of wrist watches for fall detection appears to be a promising approach, with the potential
to improve the safety and well-being of individuals. It is important to continue researching
and refining these systems to improve their accuracy and reliability.

2.2.6 Android Based Fall Detection Alert System using multi-Sensor
There has been research on the use of Android-based platforms equipped with
multiple sensors for fall detection and alert systems. These systems can use the data from
the sensors to detect falls and alert caregivers or emergency services. One study used an
Android-based smartphone equipped with an accelerometer, a gyroscope, and a
magnetometer to detect falls in elderly individuals. The data from the sensors was collected
and processed using machine learning algorithms to classify falls and non-fall activities.
The results of the study showed that the proposed fall detection system had a high accuracy
rate and could effectively detect falls in real-time.
Another study used a similar approach, but with the addition of a pressure sensor to provide
additional information on the motion of the smartphone. The results of this study showed
that the inclusion of the pressure sensor improved the accuracy of the fall detection system.
Other studies have explored the use of Android-based platforms with multiple sensors for
fall detection in other contexts, such as falls in athletes. These studies have also shown
promising results in terms of the accuracy of the fall detection systems. Overall, the use of
Android-based platforms with multiple sensors for fall detection appears to be a promising
approach, with the potential to improve the safety and well-being of individuals. It is
important to continue researching and refining these systems to improve their accuracy and
reliability.
2.2.7 Falls, Injuries Due to Falls, and the Risk of Admission to a Nursing
Home
Falls are a leading cause of injuries in older adults, and they can have serious
consequences, including hospitalization and admission to a nursing home. Falls can result
in a variety of injuries, including fractures, head injuries, and soft tissue injuries. The risk
of admission to a nursing home after a fall is significant, as falls can lead to decreased
mobility, decreased independence, and a decline in overall health. In some cases, a fall may
result in a permanent loss of function, which can make it difficult or impossible for an
individual to continue living at home. There are several factors that can increase the risk of
falls and injuries due to falls in older adults, including:

• Musculoskeletal disorders: Conditions such as osteoporosis and arthritis can
decrease muscle strength and increase the risk of falls.
• Neurocognitive disorders: Dementia and other conditions that affect cognition can
increase the risk of falls by impairing judgment and balance.
• Medications: Some medications, particularly sedatives and antidepressants, can
increase the risk of falls by causing dizziness or impairing balance.
• Environmental factors: Hazards in the home, such as cluttered floors or inadequate
lighting, can increase the risk of falls.
To reduce the risk of falls and injuries due to falls, it is important for older adults to take
steps to improve their balance and strength, and to identify and address any potential fall
hazards in their environment. Fall prevention interventions, such as exercise programs and
home modifications, can also be effective in reducing the risk of falls.
2.2.8 A Dynamic Motion Pattern Analysis Approach To F&l Detection
Dynamic motion pattern analysis is a method that involves analysing the dynamic patterns
of motion, such as acceleration and orientation, to detect falls. This approach can be used
to develop fall detection systems that use sensors, such as accelerometers and gyroscopes,
to detect falls in real-time. One study used a dynamic motion pattern analysis approach to
develop a fall detection system for elderly individuals. The system used an accelerometer
and a gyroscope to collect data on the motion and orientation of the person, and machine
learning algorithms were used to classify the data as falls or non-fall activities. The results
of the study showed that the proposed fall detection system had a high accuracy rate and
could effectively detect falls in real-time.
Another study used a similar approach, but with the addition of a magnetometer to provide
additional information on the orientation of the person. The results of this study showed
that the inclusion of the magnetometer improved the accuracy of the fall detection system.
Other studies have explored the use of dynamic motion pattern analysis for fall detection
in other contexts, such as falls in athletes. These studies have also shown promising results
in terms of the accuracy of the fall detection systems.
Overall, the use of dynamic motion pattern analysis for fall detection appears to be a
promising approach, with the potential to improve the safety and well-being of individuals.

It is important to continue researching and refining these systems to improve their accuracy
and reliability.
2.2.9 A Smartphone-based Fall Detection System
A smartphone-based fall detection system is a system that uses the sensors and other
technology in a smartphone to detect falls and alert caregivers or emergency services.
Smartphones are equipped with a variety of sensors, such as accelerometers, gyroscopes,
and magnetometers, which can be used to detect falls. One study used a smartphone-based
fall detection system to detect falls in elderly individuals. The system used an accelerometer
and a gyroscope to collect data on the motion and orientation of the smartphone during falls
and non-fall activities. Machine learning algorithms were then used to classify the data and
detect falls. The results of the study showed that the proposed fall detection system had a
high accuracy rate and could effectively detect falls in real-time. Another study used a
similar approach, but with the addition of a magnetometer to provide additional information
on the orientation of the smartphone. The results of this study showed that the inclusion of
the magnetometer improved the accuracy of the fall detection system. Other studies have
explored the use of smartphone-based fall detection systems in other contexts, such as falls
in athletes. These studies have also shown promising results in terms of the accuracy of the
fall detection systems. Overall, the use of smartphone-based fall detection systems appears
to be a promising approach, with the potential to improve the safety and well-being of
individuals. It is important to continue researching and refining these systems to improve
their accuracy and reliability.
2.2.10 Recognition of False Alarms in Fall Detection Systems
False alarms can be a problem in fall detection systems, as they can lead to unnecessary
calls for emergency assistance and can decrease the overall reliability of the system. False

alarms can be caused by a variety of factors, including sensor noise, system malfunctions,
and activities that are mistaken for falls. To address the problem of false alarms in fall
detection systems, researchers have developed a number of approaches for recognizing
false alarms and reducing their occurrence. These approaches can be divided into two
categories: pre-processing techniques and classification techniques. Pre-processing
techniques involve filtering or smoothing the data from the sensors to reduce noise and
improve the accuracy of the fall detection system. These techniques can include techniques
such as median filtering, Kalman filtering, and wavelet filtering. Classification techniques
involve using machine learning algorithms to classify falls and non-fall activities based on
the data from the sensors. These techniques can include techniques such as decision tree
algorithms, support vector machines, and neural networks. Overall, recognition of false
alarms in fall detection systems is an important area of research, as it can improve the
reliability and effectiveness of these systems. It is important to continue researching and
refining these approaches to reduce the occurrence of false alarms and improve the
accuracy of fall detection systems.

CHAPTER 3
DRAWBACKS OF EXISTING SYSTEM
3.1 False alarms
Fall detection systems can sometimes generate false alarms, which can lead to unnecessary
calls for emergency assistance and decrease the overall reliability of the system.
3.2 Cost
Fall detection systems can be expensive, particularly if they require specialized equipment
or monitoring services.
3.3 User compliance
Fall detection systems require the user to wear or carry the device at all times, and some
users may be resistant to using the device or may forget to wear it.
3.4 Limited accuracy
Fall detection systems may not always accurately detect falls, particularly if the fall is
unusual or the sensors are not functioning properly.
3.5 Privacy concerns
Some users may be concerned about the privacy implications of using a fall detection
system, particularly if the system requires continuous monitoring or the transmission of
personal data.
3.6 Dependence on technology
Fall detection systems rely on technology, and if the device or system fails, it may not be
able to detect falls.
Overall, while fall detection systems can be a useful tool for improving the safety and well-
being of individuals, it is important to consider the potential drawbacks and limitations of
these systems when deciding whether to use one.

CHAPTER 4
PROPOSED SYSTEM
The proposed methodology for riders fall detection is divided into four stages,
i.e., helmet design, data acquisition, feature extraction and selection, and finally,
classification as shown in Fig. 1. The detail of each stage is given as follows.
4.1 Helmet Design
A smart helmet is created in the first stage utilizing the MPU6050, which contains on-chip
3-axis accelerometer and gyroscope sensors. The MPU-6050 is controlled by an
ARDUINO-based microcontroller, which collects sensor data and Bluetooth-transmits it to
the smartphone application. As seen in Fig. 2, this circuitry is installed inside the helmet
unit and positioned such that it is parallel to the helmet's head surface. To prevent excessive
movement of the components, the NodeMCU12 microcontroller, microSD card, battery,
and cables are all tightly packed inside the helmet unit.
Fig 4.1 - Proposed methodology for riders fall detection using inertial sensors embedded in the
helmet

4.2 Data Acquisition
The designed helmet is fastened to a dummy that is secured to the seat of a bicycle. The
cycle is propelled forward along a straight path while the helmet is powered on. The cycle
is allowed to move freely for about 5 seconds before falling, and in that time, accelerometer
and gyroscope data are recorded at a sampling rate of 10 Hz. When fall occurs, additional
checkpoints are introduced. There are 30 trials in total, with 15 left and 15 right side falls.
Due to the damaged data file, two trials were excluded. The obtained data is then split into
two distinct pieces, referred to as fall and non-fall, based on checkpoints. 56 incidents made
up the data under investigation.
4.3 Feature Extraction and Selection
The obtained accelerometer and gyroscope data are transformed into 22 different statistical
features, as given in Table I. The majority of these characteristics have been applied to fall
detection in previous research. These features are taken from the obtained signal, which
can either be an accelerometer or a gyroscope, and are extracted along each axis, or x, y,
and z. Against each case, the total feature vector length is 22 3 2, or 132. The wrapper
approach is used to choose the best features after feature extraction. Based on the
effectiveness of the classifiers, the wrapper method evaluates the value of the features. It is
employed in machine learning techniques and assesses features using a learning scheme.
4.4 Classification
The last stage is to identify a fall using a set of features. This study evaluates the following
classifiers for smart helmet-based rider fall detection.

CHAPTER 5
SYSTEM MODULES
A smart helmet is a type of protective headgear equipped with various features
and technologies, such as sensors, communication modules, and computing capabilities.
These features allow the helmet to perform various functions, such as detecting impacts,
tracking the user's location, and communicating with other devices. In this article, we will
discuss the different system modules that can be used in a smart helmet to detect impact,
including a GSM module, GPS module, Arduino microcontroller, battery, and
accelerometer.
5.1 GSM Module
A GSM (Global System for Mobile Communications) module is a device that allows the
helmet to connect to a mobile network and communicate with other devices through SMS
or data transmission. The GSM module can be used to send an emergency message or alert
to a predetermined contact or emergency services in the event of an impact or other
emergency situation.
5.2 GPS Module
A GPS (Global Positioning System) module is a device that uses satellite signals to
determine the helmet's location and track its movement. The GPS module can be used to
track the user's location and provide real-time tracking information to a predetermined
contact or emergency services in the event of an impact or other emergency situation.
5.3 Arduino Microcontroller
An Arduino is a type of microcontroller that can be used to control and monitor various
sensors and devices within the helmet. The Arduino can be programmed to perform various
tasks, such as collecting data from the accelerometer and sending an emergency message
or alert in the event of an impact.

5.4 Battery
The battery is a critical component of the smart helmet, as it provides power to all of the
other components. The battery should have a long lifespan and be able to hold a charge for
a sufficient amount of time to ensure that the helmet can perform its functions even in the
event of a power outage or other emergency.
5.5 Accelerometer
An accelerometer is a device that measures acceleration, or the rate of change in velocity.
In a smart helmet, the accelerometer can be used to detect impacts and determine the
severity of the impact. The data collected by the accelerometer can be used to trigger an
emergency message or alert in the event of a severe impact.
In summary, a smart helmet can be equipped with various system modules, including a
GSM module, GPS module, Arduino microcontroller, battery, and accelerometer, to detect
impacts and provide real-time tracking and communication capabilities. These modules
work together to provide a comprehensive solution for protecting the user and ensuring
their safety in the event of an impact or other emergency situation.

CHAPTER 6
ARCHITECTURE
The smart helmet system consists of four main algorithms: Naive Bayes (NB), K-
Nearest Neighbor (KNN) and Random Forest (RF)
6.1 Naive Bayes (NB)
One of the most straightforward yet efficient classification algorithms based on the
probability model is the Bayesian classifier. It creates the efficient classification using the
maximization posteriori rule. This algorithm forecasts the categorization based on
probability (P), where A stands for the hypothesis and C for the supporting
data.
6.2 K-Nearest Neighbor (KNN)
One of the straightforward classification algorithms, the KNN algorithm is frequently used
in learning algorithms. This algorithm divides data points into various classes to forecast
the classification of a future sample point. There are no presumptions about the data in this
statistical approximation approach. It obtains the cases and groups them based on similarity
using a distance function. Minkowski, Manhattan, and Euclidean distance functions are
available. A popular distance estimate is the euclidean distance, which is calculated as
follows: .
6.2 Random Forest (RF)
The supervised learning algorithm RF is a type of search tree. By creating numerous
decision trees and integrating them, this technique predicts reliable and stable
classification. It employs averaging to increase the resultant data's precision for more
precise classification.

CHAPTER 7
SYSTEM IMPLEMENTATION
The proposed smart helmet that can detect impacts works by using an accelerometer
to measure the acceleration experienced by the helmet during an impact, and a GSM
module to send a message with the location and acceleration data to a remote location (such
as a server or a phone). The helmet is equipped with an Arduino microcontroller, which
controls the different hardware components and performs the desired functionality.
• The helmet is worn by the rider and the hardware components (Arduino,
GSM module, GPS module, accelerometer, and battery) are powered on.
• The accelerometer continuously measures the acceleration experienced by
the helmet along one or more axes.
• The Arduino reads the acceleration data from the accelerometer and uses it
to detect impacts. When an impact is detected, the Arduino triggers the GSM
module to send a message with the location (obtained from the GPS module)
and acceleration data.
• The GSM module sends the message via the cellular network to the desired
recipient (such as a server or a phone).
• The recipient receives the message and can use the location and acceleration
data to provide assistance or take any other desired action.
A smart helmet that can detect impacts is a protective device that uses an
accelerometer to measure the acceleration experienced by the helmet during an impact, and
a GSM module to send a message with the location and acceleration data to a remote
location (such as a server or a phone). The helmet is equipped with an Arduino
microcontroller, which controls the different hardware components and performs the
desired functionality.
The accelerometer is a sensor that measures acceleration along one or more axes. It
is commonly used in applications where the acceleration of an object needs to be measured,
such as in smartphones to detect the orientation of the device or in automobiles to deploy
airbags in the event of a collision. In a smart helmet, the accelerometer is used to detect

impacts. When an impact occurs, the accelerometer will detect the sudden change in
acceleration and send a signal to the Arduino.
The Arduino is a microcontroller that can be programmed to control hardware
components and perform a variety of tasks. It is a popular choice for DIY electronics
projects due to its low cost and flexibility. In a smart helmet that can detect impacts, the
Arduino is responsible for reading the acceleration data from the accelerometer and using
it to detect impacts. When an impact is detected, the Arduino can trigger the GSM module
to send a message with the location and acceleration data.
The GSM (Global System for Mobile Communications) module is a device that
allows the helmet to communicate via a cellular network. It can be used to send text
messages or make phone calls, allowing the rider to call for help in the case of an emergency
or report the impact to a remote location. The GSM module is typically connected to the
Arduino via a serial interface, allowing the Arduino to send commands to the module and
receive data from it.
The GPS (Global Positioning System) module is a device that determines the
helmet's location using satellite signals. It can be used to track the helmet's location in the
event it is lost or stolen, or to provide turn-by-turn navigation for the rider. The GPS module
is also typically connected to the Arduino via a serial interface, allowing the Arduino to
read the location data from the module.
The battery is a power source for the helmet, providing electricity to the Arduino,
GSM module, GPS module, and accelerometer. Depending on the specific requirements of
the application, the battery could be a rechargeable lithium-ion battery or a disposable
battery such as AA or AAA cells.
The code for the Arduino includes functions for reading the acceleration data from
the accelerometer and using it to detect impacts. One way to do this is to set a threshold for
the acceleration that indicates an impact has occurred. For example, if the acceleration
exceeds a certain level (e.g. 10 g's), the Arduino can consider it an impact. The code can
then trigger the GSM module to send a message with the location and acceleration data.

CHAPTER 8
SYSTEM REQUIREMENTS
In our proposed system, we are using a range of hardware and software components
to create the smart helmet. The system uses machine learning and IoT technologies to
provide real-time insights into the performance of the helmet.
8.1 Hardware Requirements
The Arduino Uno is a microcontroller board that can be used to control various devices and
sensors. A GSM (Global System for Mobile Communications) module allows the Arduino
to send and receive text messages and make phone calls using a SIM card. A GPS (Global
Positioning System) module can be used to determine the location of the Arduino. An
accelerometer is a sensor that can measure acceleration and tilt, and can be used to detect
movement or orientation. A buzzer is a device that can produce a sound, and can be used
to create alarms or alerts. A 9V battery can be used to power the Arduino and other
components. A push button is a type of switch that can be used to control and turn off false
alarms.
8.2 Software Requirements
The software requirements for our system include the Arduino IDE, which is a software
development environment for the Arduino that is used to program the microcontroller
board. The Windows OS is the operating system that is used to develop the admin
application for monitoring and managing the data collected by the system.

CHAPTER 9
DATA FLOW DIAGRAM
The data flow diagram that illustrates the flow of data in a smart helmet that can
detect impacts using a GSM module, GPS module, Arduino, battery, and accelerometer:
The helmet is worn by the rider and the hardware components (Arduino, GSM module,
GPS module, accelerometer, and battery) are powered on. The accelerometer continuously
measures the acceleration experienced by the helmet along one or more axes. The
acceleration data is sent to the Arduino. The Arduino reads the acceleration data and uses
it to detect impacts. If an impact is detected, the Arduino triggers the GPS module to obtain
the helmet's location. The location data is sent to the Arduino. The Arduino combines the
location and acceleration data and sends it to the GSM module. The GSM module sends
the data via the cellular network to the desired recipient (such as a server or a phone).
Figure 9.1 - Level-0 Data Flow Diagram of Proposed System.

Figure 9.2 - .Level-1 Data Flow Diagram of Proposed System.

CHAPTER 10
WORK PLAN AND TASK ALLOCATION
The purpose of this workplan is to outline the steps and tasks necessary to complete.
The goal of the project is to develop and implement a system that utilizes smart helmet to
more effectively collect and classify resulting in a safe ride for the user
10.1 Phase 1: Planning and research
Research the different hardware components and their capabilities and limitations.
Determine the specific requirements and goals for the helmet (e.g. type of impacts to detect,
communication methods, location accuracy, battery life)Develop a high-level design for
the helmet, including the hardware and software components and their interactions
10.2 Phase 2: Hardware assembly and testing
Assemble the hardware components according to the design. Test the hardware components
individually to ensure they are functioning correctly. Connect the hardware components
and test them together to ensure they are communicating and working as intended
.
10.3 Phase 3: Software development and testing
Write the code for the Arduino to control the different hardware components and perform
the desired functionality (e.g. detect impacts, send a message with the location and
acceleration data). Test the code using the Arduino IDE and debug any issues. Integrate the
code with the hardware and test the system as a whole.
10.4 Phase 4: Integration and testing
Integrate the hardware into the helmet by mounting the components inside the helmet and
enclosing them in a protective casing. Test the helmet in various scenarios to ensure it is
functioning correctly and meeting the requirements.

Table 10.4 - Workplan
Phase Done By Expected Completion
Planning and research Basil Varghese December, 2022
Hardware assembly
and testing
Arjun Saji February, 2023
Software development
and testing
Geevarghese S Isaac March, 2023
Integration and testing Basil Pappy Roy March, 2023

CHAPTER 11
CONCLUSION AND FUTURE WORK
11.1 CONCLUSION
When riding, a helmet is worn to prevent serious head injuries. A helmet's fall detection
and automated rescue delay prevention features. Accelerometer and gyroscope sensing
units are integrated into the helmet to detect riders falling and automatically send a message
to a nearby rescue. In this study, four distinct classification algorithms—Naive Bayes, k-
nearest neighbour, random forest, and support vector machine—are used to categorise fall
and non-fall data. With the maximum classification accuracy of 98.21%, Nave Bayes is
discovered to be the best classification algorithm for fall detection. Additionally, data show
that NB performs better than other classifiers and has higher precision, recall, and F-score
as well as lower absolute and relative error rates than KNN, RF, and other classifiers.
A smart helmet that can detect impacts is a protective device that uses an accelerometer to
measure the acceleration experienced by the helmet during an impact, and a GSM module
to send a message with the location and acceleration data to a remote location (such as a
server or a phone). The helmet is equipped with an Arduino microcontroller, which controls
the different hardware components and performs the desired functionality.
The accelerometer is a sensor that measures acceleration along one or more axes, and is
used to detect impacts in the helmet. The Arduino microcontroller reads the acceleration
data from the accelerometer and uses it to detect impacts, and can trigger the GSM module
to send a message with the location (obtained from the GPS module) and acceleration data.
The GSM module allows the helmet to communicate via a cellular network, while the GPS
module determines the helmet's location using satellite signals. The battery is a power
source for the helmet, providing electricity to the Arduino, GSM module, GPS module, and
accelerometer.
A smart helmet that can detect impacts is a useful tool for protecting riders and providing
them with information and assistance in the case of an accident. It can alert emergency
services or a designated contact of the impact and location, allowing for timely assistance.

The helmet can also provide valuable data on the acceleration experienced during the
impact, which can be used for research or to improve the design of protective devices.
11.2 SCOPE OF FUTURE RESEARCHES
There are several areas where future research could be conducted on a smart helmet that
can detect impacts using a GSM module, GPS module, Arduino, battery, and
accelerometer. Some possible directions for future research include:
Improving the accuracy and reliability of the impact detection: Research could be
conducted to improve the algorithms used to detect impacts and reduce the number of false
positives or false negatives. This could involve using machine learning techniques to
analyse large datasets of acceleration data from impacts and non-impacts, and developing
models that can accurately classify them.
Enhancing the communication capabilities: Research could be conducted to improve the
communication range and reliability of the GSM module, or to explore the use of alternative
communication technologies such as low-power wide-area networks (LPWANs) or
satellite communication.
Optimizing the power consumption: Research could be conducted to reduce the power
consumption of the helmet and extend the battery life. This could involve optimizing the
hardware components and the code, and exploring the use of energy-efficient batteries or
alternative power sources such as solar panels.
Improving the user interface: Research could be conducted to enhance the user interface
(UI) of the helmet and make it more user-friendly. This could involve designing an intuitive
UI with clear and concise information, and developing features such as voice control or
gesture recognition.
Expanding the capabilities of the helmet: Research could be conducted to add additional
features to the helmet, such as a rear-view camera and a heads-up display. These features
could improve the safety and convenience of the helmet for the rider.

REFERENCES
[1] Z. T. Al-Dahan, N. K. Bachache, and L. N. Bachache, “Design and implementation of
fall detection system using MPU6050 Arduino,” in Lecture Notes in Computer Science
(including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in
Bioinformatics), 2016.
[2] K. Chaccour, R. Darazi, A. H. El Hassani, and E. Andrès, “From fall detection to fall
prevention: A generic classification of fallrelated systems,” IEEE Sens. J., vol. 17, no. 3,
pp. 812–822, 2017.
[3] M. Ehatisham-ul-Haq and M. A. Azam, “Opportunistic sensing for inferring in-the-wild
human contexts based on activity pattern recognition using smart computing,” Futur.
Gener. Comput. Syst.,vol. 106, pp. 374–392, 2020.
[4] M. Ehatisham-Ul-Haq, M. A. Azam, Y. Amin, and U. Naeem, “C2FHAR: Coarse-to-
Fine Human Activity Recognition with Behavioral Context Modeling Using Smart Inertial
Sensors,” IEEE Access, vol. 8, pp. 7731–7747, 2020.
[5] A. Z. Rakhman, L. E. Nugroho, Widyawan, and Kurnianingsih, “Fall detection system
using accelerometer and gyroscope based on smartphone,” in 2014 The 1st International
Conference on Information Technology, Computer, and Electrical Engineering, 2014, pp.
99–104.
[6] F. Hussain, F. Hussain, M. Ehatisham-ul-Haq, and M. A. Azam, “Activity-Aware Fall
Detection and Recognition based on Wearable Sensors,” IEEE Sens. J., 2019.
[7] F. Hussain, M. Ehatisham-ul-Haq, M. A. Azam, and A. Khalid, “Elderly Assistance
Using Wearable Sensors by Detecting Fall and Recognizing Fall Patterns,” 2018, pp. 770–
777.
[8] N. Friedman, D. Geiger, and M. Goldszmidt, “Bayesian Network Classifiers,” Mach.
Learn., vol. 29, no. 2, pp. 131–163, 1997.

[9] G. Guo, H. Wang, D. Bell, Y. Bi, and K. Greer, “kNN Model Based Approach in
Classification,” Move to Meaningful Internet Syst. 2003 CoopIS, DOA, ODBASE, vol.
2888, pp. 986–996, 2003.
[10] L. Breiman, “Random Forrest,” Mach. Learn., pp. 5–32, 2001.
[11] C. Cortes and V. Vapnik, “Support-Vector Networks,” Mach. Learn., vol. 20, no. 3,
pp. 273–297, 1995.
[12] D. M. W. Powers, “Evaluation: from precision, recall and F measure to ROC,
informedness, markedness and correlation,” J.Mach. Learn. Technol., vol. 2, no. 1, pp. 37–
63, 2011.

APPENDIX A - SAMPLE CODING
// Include Libraries
#include <Arduino.h>
#include <SoftwareSerial.h>
#include <TinyGPS.h>
// Pin Definitions
#define xPin A3
#define yPin A4
#define zPin A1
#define buzzer4
#define gpsTX11
#define gpsRX 10
#define SIM800L_PIN_TX 2
#define SIM800L_PIN_RX 3
#define button 13
int xVal, yVal, zVal, total_acc1, total_acc2, total_acc3;
int tolerance = 0.5;
String msg, location;
float latitude, longtitude;
SoftwareSerial SIM800L(SIM800L_PIN_TX, SIM800L_PIN_RX); //SoftwareSerial for
GSM Module
SoftwareSerial gpsSerial(gpsTX, gpsRX); //SoftwareSerial for GPS
TinyGPS gps;
void setup() {
SIM800L.begin(9600);
gpsSerial.begin(9600);
Serial.begin(9600);
pinMode(xPin, OUTPUT);
pinMode(yPin, OUTPUT);

pinMode(zPin, OUTPUT);
pinMode(buzzer, OUTPUT);
pinMode(button, INPUT);
}
void loop() {
if (digitalRead(button) == LOW){
digitalWrite(buzzer, LOW);
} else if (digitalRead(button) == HIGH){
digitalWrite(buzzer, HIGH);
}
xVal = analogRead(xPin);
yVal = analogRead(yPin);
zVal = analogRead(zPin);
total_acc1 = sqrt((xVal * xVal) + (yVal * yVal) + (zVal * zVal)); //calculating net
"normal" acceleration
delay(2000);
acceleration for the 1st check
if ((abs(total_acc2) >= abs(13/10 * total_acc1)) || (abs(total_acc2) <= abs(7/10 *
total_acc1))){
delay(10000);
acceleration for the 2nd check

if(((abs(total_acc3) >= abs(11.5/10 * total_acc2)) || (abs(total_acc3) <= abs(0.5/10 *
total_acc2)))){
digitalWrite(button, HIGH);
SendMessage();
}
}
}
void SendMessage(){
GetLocation();
SIM800L.println("AT+CMGF=1"); //setting SMS mode to TEXT
delay(500);
SIM800L.println("AT+CMGS="+919526680261"r"); //+91 9526680261 (friend A)
is the number where we'll message
delay(500);
msg = "Hello, I'm injured in a traffic accident in the coordinates of " + location + "
please save me!!!";
SIM800L.println(msg); //finalizing the message
delay(500);
SIM800L.println((char)24); //sending the message
}
void GetLocation(){
if(gpsSerial.available()){ // check for gps data
if(gps.encode(gpsSerial.read())){ //encoding GPS data
gps.f_get_position(&latitude, &longtitude); //getting coordinates
}
location = "Latitude: " + String(latitude) + " Longtitude: " + String(longtitude);
}
}

APPENDIX B – SCREENSHOTS
Fig 1: Alert message

Fig 2. Device Prototype

MAR BASELIOSINSTITUTE OF
TECHNOLOGY AND SCIENCE [MBITS]
NELLIMATTOM P.O., KOTHAMANGALAM, ERNAKULAM, KERALA -686 693, INDIA
(NAAC Accredited)
NATIONAL CONFERENCE ON INNOVATIONS
IN COMPUTER SCIENCE &TECHNOLOGY (NCICST'23)
CERTIFICATE OF PARTICIPATION
Thisis tocertify that Mr/Ms/Dr..ARJ.UN.6AJ.
of.Maa.Baseltes.anekhsk.g. Aeahnaleg4.amd.Sekens.(Gnets). Neltiraatfrns.:
hasparticipated and presentedthe papertitled.SKyALGA.-.A..9T.Bassd..Smast..tdmet..ko...
.Acctdent..Delecion:<papea...NGISSI.23.026lat the National Conference on Innovations in
Computer Science&Technology (NCICST'23) organized by Department of ComputerScience and Engineering in
associationwithCSICochin Chapteron 25*and26" May2023.
Mr. Midhun Mathew
HOD (CSE),
Organizing Chair
Sojan Lat
Principal
Sr/ C. A. Kunjachan
Secretary
........

MAR BASELIOS INSTITUTE OF
NELLIMATTOM P.O., KOTHAMANGALAM, ERNAKULAM, KERALA- 686 693, INDIA
(NAAC Accredited)
IN COMPUTER SCIENCE &TECHNOLOGY (NCICST23)
CERTIFICATEOF PARTICIPATION
Thisis tocertifythat Mr/Ms,Pr.BASIL..PAPPY Roy
of.Maa.Baseltes.atabe.a.Aeehnalag4.s1.Skenc.MBIT). Nelimaats...
hasparticipatedand presentedthepaper titld.£ULAGAP..Am.oT. Based.Smatk..Heloet. Be...
.Aectdent.Dcteekon:.CPapa..D.:..NSICSI.23.D24.lat the National Conference on Innovations in
ComputerScience&Technology(NCICST 23) organized by Department of Computer Science and Engineering in
associationwith CSICochin Chapteron25"and26*May2023.
Mr. Midhun Mathew
HOD (CSE),
Organizing Chair
Dr. P. Sojan Lal
Principal
Sri. C. A. Kunjachan
Secretary

MAR BASELIOS INSTITUTE OF
NELLIMATTOM P.O., KOTHAMANGALAM, ERNAKULAM, KERALA -686 693, INDIA
(NAAC Accredited)
IN COMPUTER sCIENCE &TECHNOLOGY (NCICST'23)
This is to certify that Mr/Ms,/Dr....5A34.VARGHESE
of.Mas.Baseltes.tnalitak.,..kshnaleg4.and.Setemss. CMALT).NeltiraaHrens..
hasparticipatedandyresentedthepapertitied SkuLuSAP..An.le...BAsed.Emas.elmek.fe
. .
Acstdem...Dsksekon.(pepm.LA..NcICST.23.2êat the National Conference on Innovations in
Mr. Midhun Mathew
HOD (CSE),
Organizing Chair
Dr. P. Sojan Lal
Principal
Sri. C.A. Kunjachan
Secretary
ComputerScience&Technology(NCICST23) organizedby Department of Computer Scienceand Engineeringin
association withCSICochin Chapteron 25"and26"May2023.

MARBASELIOS INSTITUTE OF
NELLIMATTOM P.O.,KOTHAMANGALAM, ERNAKULAM, KERALA-686 693, INDIA
(NAAC Accredited)
IN COMPUTER SCIENCE &TECHNOLOGY (NCICST23)
Thisis tocertifythat Mr/M»/Dr....CAEENA.RGHESE.S.SSAG.
of. Ma.Ba.seltos.mathak..sshnelag.skand, Seience.CMans).Neltmates.
hasparticipated andpresentedthepaper titled.aKMLASAP...A..eI.Based..Brnest..telnaet.ke...
Aeeldent...Detec.emCPApe.AR...NSISAT.23-029kt the National Conference on Innovations in
ComputerScience&Technology (NCICST23) organizedby Departmentof ComputerScienceandEngineeringin
associationwithCSICochinChapteron25*and26 May2023.
Mr. Midhun Mathew
HOD (CSE),
Organizing Chair
Dr. P. Sojan Lal
Principal
Sri. C, A. Kunjachan
Secretary

SKULLCAP – An IOT Based Smart Helmet for
Accident Detection
Arjun Saji
Computer Science and Engineering Dept.
Mar Baselios Institute of Technology and Science(MBITS)
Kothamangalam, Kerala, India
arjunsaji149@gmail.com
Basil Pappy Roy
basilpappyroy@gmail.com
Basil Varghese
basilvarghese3737@gmail.com
Geevarghese S Issac
geevarghesesisaac@gmail.com
Ms.Teena Skaria
Computer Science and Engineering Dept
teenaskaria@mbits.ac.in
Ms.Nisreen M Ali
Computer Science and Engineering Dept
nisreenm@mbits.ac.in
Abstract—Riders wear helmets as protective headgear to make
driving safer and reduce injuries. In this study, a smart helmet
is created that can detect riders’ falls and then using the global
positioning system (GPS) to determine the location of the fall
and deliver a message through a GSM module to prevent rescue
delays. For fall detection, inertial sensors, such as gyroscope
and accelerometer sensors with a three-axis sensing unit, are
mounted inside the helmet and managed by an ARDUINO-
based microcontroller. In 28 trials, the constructed helmet is
utilised for data collection, and the recorded data is used
to derive 22 statistical features.Using the wrapper technique,
a subset of distinguishing characteristics is chosen, and four
different classifiers—Naive Bayes, k-nearest neighbour, random
forest, and support vector machine—are then used to the chosen
features to categorise the two states—fall and non-fall. The results
clearly show that the Nave Bayes classifies cyclists’ falls with the
maximum accuracy of 98.21
Index Terms—GPS module,GSM module,ARDUINO, prevent
injuries
I. INTRODUCTION
In every aspect of our life, ensuring safety and security
is a vital priority. The growing prevalence of road accidents
around the world is one of the most urgent issues in this
regard. A startling 1.35 million people every year may away in
such catastrophes, according to a World Health Organisation
(WHO) report from 2018. In this setting, two-wheeler traffic
accidents are responsible for a growing number of fatalities.
Unsettlingly, falls are ranked as the second most common
cause of fatal accidents and injuries.
Accidental falls have serious consequences since they cause
numerous injuries and negative results. Every year, approxi-
mately 2.8 million falls-related injuries result in visits to the
emergency room, more than 8 million hospitalisations, and
27,000 fatalities in the United States alone. This highlights
the seriousness of the issue, especially for motorcyclists and
cyclists who have a higher chance of falling. Tragically, when
accidents happen, the absence of a trustworthy fall detection
mechanism and rapid location information frequently results
in delays in giving urgent first aid, which causes avoidable
deaths.
Therefore, it is crucial to create a reliable fall detection
system in order to save lives. Injuries caused by falls can
be treated right away, saving thousands of lives that would
otherwise be lost. Researchers have focused on creating ef-
ficient fall detection techniques, especially for pedestrians
and bikers, as a result of the urgent need for intervention.
Wearable technology, smartphone-based solutions, wireless-
based systems, ambient sensor-based strategies, vision-based
techniques, and floor sensor/electric field sensor-based systems
are some of these mechanisms.
A wireless fall detection system using two units, a bike
unit and a helmet unit, is one example. The bike unit has an
onboard accelerometer to detect free falls, and these devices
communicate via an RF transmitter built into the helmet.
Similar to this, cameras are used by vision-based fall detection
systems to monitor and assess activities while using image
processing to spot falls. However, these techniques could
need a lot of processing. Another method involves using
the smartphone’s built-in accelerometer, gyroscope, and other
sensors to detect falls and deliver emergency alarms. These
Proceedings of National Conference on Innovations in Computer Science & Technology (NCICST '23)
ISBN: 978-93-5906-287-7 131

smartphone or wearable sensor placements can change from
study to study.
This research introduces a unique method for fall detection
and accident avoidance for cyclists that is based on the
SkullCap helmet-based system. To improve safety and offer
quick help in the event of falls, the suggested solution inte-
grates Internet of Things (IoT) technology with smart helmet
design. The MPU 6050 wearable sensor, which has an on-
chip accelerometer and gyroscope, is installed in the helmet.
With the help of these sensors, which record information
on linear acceleration, direction, and angular velocity, human
motion may be tracked in three dimensions. A complete and
sophisticated safety solution for riders, the collected data is
recorded using an SD-card module built inside the helmet.
Using a human dummy to simulate falls while cycling,
thorough data collection is done to assess the system’s efficacy.
The system is then trained using this data to distinguish
between falls and other events with accuracy. The experimental
findings show how the suggested fall detection system works
and how it has the potential to greatly increase bike safety.
In order to reduce the risks connected with falls while
cycling, it is crucial to design an effective fall detection
system for bikers. It becomes feasible to quickly detect falls
and provide timely aid, minimising needless loss of life,
by integrating wearable sensors into a smart helmet. The
remaining sections of this paper will focus on the suggested
approach, the outcomes of the experiment, and some last
thoughts.
II. RELATED WORK
Smartphone accelerometers are tiny gadgets that can gauge a
smartphone’s acceleration and orientation, and this knowledge
can be applied to the detection of falls. One study employed
machine learning algorithms to classify the data from a
smartphone accelerometer to identify falls in elderly people. A
gyroscope was added to a different study’s identical methodol-
ogy to provide more details on the motion of the smartphone.
In various circumstances, such as falls in sports, additional
research have investigated the use of smartphone accelerom-
eters for fall detection. Overall, the potential to increase
individual safety is highlighted by the use of smartphone
accelerometers for automatic fall detection. Continued study
and development of these systems are necessary to enhance
theirTo increase these systems’ accuracy and dependability, it
is crucial to carry out additional study and refinement.[1]Smart
watch and smartphone fall detection systems that employ the
accelerometers and other sensors in both devices to detect falls
have been the subject of research. One study found that the
suggested fall detection system had a high accuracy rate and
could successfully detect falls in real-time in older people
using a smart watch and a smartphone. A gyroscope and
a magnetometer were added to a different study’s identical
methodology in order to provide more details on the motion
and orientation of the devices. The application of coupled
smart watch and smartphone fall detection systems in other
studies has focused onamong various circumstances, such as
falls among sports, additional research have investigated the
use of integrated smart watch and smartphone fall detection
systems. In general, using a combination of smart watch
and smartphone fall detection technologies seems to be a
good strategy, with the potential to increase people’s safety.
To increase these systems’ accuracy and dependability, it is
crucial to carry out additional study and refinement.[2]
A smart helmet is a type of bicycle helmet that offers
additional functionality in addition to the protection that
standard helmets provide through the use of sensors and other
technology. One potential application for a smart helmet is a
smart bike system that might provide bikers with information
and help to enhance their cycling experience.[3]The usage of
Android-based systems for fall detection in older people has
been studied. In one study, machine learning techniques were
employed to categorise the sensor data using an Android-based
smartphone with an accelerometer and gyroscope. A magne-
tometer was added to a different study’s identical methodology
to provide data on the smartphone’s orientation. In other
research, such as those looking at falls in athletes, Android-
based platforms have been investigated for application in fall
detection. Overall, using Android-based platforms to detect
falls in elderly people seems like a viable strategy, with the
potential to increase their safety and wellbeing. To increase
these systems’ accuracy and dependability, it is crucial to
carry out additional study and refinement.[4] The use of
wrist watches for fall detection has been studied. In one
study, a wristwatch was used to identify older people who
had fallen, and machine learning algorithms were used to
categorise the sensor data. A magnetometer was added to
another study’s comparable methodology to provide data on
the orientation of the. In general, using wrist watches to detect
falls seems to be a viable strategy with the potential to increase
people’s safety and wellbeing. To increase these systems’
accuracy and dependability, it is crucial to carry out additional
study and refinement.[5]The usage of Android-based platforms
with several sensors for fall detection and alert systems has
been studied. One study employed a smartphone running the
Android operating system and outfitted with an accelerometer,
gyroscope, and magnetometer to monitor elderly people for
falls. To categorise falls and non-fall activities, the sensor data
was gathered and analysed using machine learning techniques.
A pressure sensor was added to a different study’s method
in order to provide more details about the movement of the
smartphone.[6] Elderly people frequently get injuries from
falls, which can have major repercussions like hospitalisation
and nursing home admission. Falls can cause soft tissue, bone,
and head injuries as well as diminish independence, mobility,
and overall health. They can also result in fractures. Some falls
may cause a lifelong loss of function, making it difficult or
impossible for the person to live at home going forward.[7]
In order to identify falls, a technique known as ”dynamic
motion pattern analysis” analyses dynamic patterns of motion,
such as acceleration and orientation. It can be used to create
fall detection systems that track falls in real time using
sensors like accelerometers and gyroscopes. In one study, a
ISBN: 978-93-5906-287-7 132

fall detection system for older people was created using a
dynamic motion pattern analysis method. This system had
a high accuracy rate and could successfully detect falls in
real-time. A magnetometer was added to a different study’s
identical methodology in order to provide more details about
the subject’s orientation. In various circumstances, such as
falls in sports, additional research have investigated the use
of dynamic motion pattern analysis for fall detection.Overall,
employing dynamic motion pattern analysis to detect falls
seems to be a viable strategy with the potential to increase peo-
ple’s safety and welfare. To increase these systems’ accuracy
and dependability, it is crucial to carry out additional study
and refinement.[8] A system that employs a smartphone’s
sensors and other technology to detect falls and notify carers
or emergency services is known as a smartphone-based fall de-
tection system. An accelerometer and gyroscope were utilised
in one study to capture data on the motion and orientation
of the smartphone during falls and non-fall activities. This
technology was used to identify falls in older people.In various
circumstances, such as falls in sports, additional research have
investigated the application of smartphone-based fall detection
systems. In general, using smartphone-based fall detection
systems seems like a viable strategy, having the potential
to increase people’s safety and wellbeing. To increase these
systems’ accuracy and dependability, it is crucial to carry
out additional study and refinement.[9] False alarms can be
a concern with fall detection systems, resulting in calls for
emergency help and lowering the system’s dependability and
efficiency. In order to combat this, researchers have created
methods to identify false alarms and lower the likelihood
that they will occur, which may be broken down into pre-
processing and classification strategies. While classification
strategies use machine learning algorithms to categorise falls
and non-fall activities based on sensor data, pre-processing
techniques entail filtering or smoothing data from the sensors
to minimise noise and increase accuracy. There is ongoing
research to further improve these methods and the systems’
accuracy.[10]
III. SCOPE AND OBJECTIVE
The purpose of this study is to demonstrate a proof of
concept for the application of wearable sensors to the detection
of falls in the context of motorcycle riders. This fall detection
framework intends to improve the safety and wellbeing of
motorcycle riders by utilising modern sensor technology built
into a rider’s helmet. It does this by enabling real-time fall
detection and quick notification of emergency services.
The development and assessment of the viability and effi-
cacy of the suggested fall detection framework are the main
objectives of this study. The use of wearable sensors—more
particularly, the MPU 6050 sensor, which has an on-chip
accelerometer and gyroscope—is the main focus. Due to the
precise tracking of the rider’s three-dimensional motion made
possible by these sensors, falls-related aberrant motions can be
identified. The main goal is to show how wearable sensors can
effectively detect falls in real-time while simultaneously alert-
ing emergency personnel. The suggested fall detection system
can dramatically shorten response times, increase survival
chances, and lessen the degree of injuries caused by motorbike
accidents by providing prompt aid in urgent situations.
Although this study offers a proof of concept, there are nu-
merous opportunities for additional research and development
that could improve the system’s accuracy, dependability, and
usability. Increasing the precision and dependability of fall
detection systems is an important consideration. It is feasible
to train the system using a bigger dataset of fall occurrences
by fine-tuning the algorithms and utilising machine learning
approaches, boosting performance.Furthermore, effective and
timely warning of emergency services depends greatly on
communication skills. The system may seamlessly transmit fall
detection alerts by integrating with mobile networks or specific
communication protocols, ensuring that emergency services
are informed right away.Another issue that needs consideration
is the optimisation of power consumption. The wearable sensor
system’s battery life can be increased by putting effective
power management strategies into practise, such as optimising
sensor sampling rates and investigating low-power modes.
Investigating energy harvesting possibilities can also improve
the system’s sustainability and power efficiency.In conclusion,
this paper aims to present a proof of concept for the use
of wearable sensors in fall detection for motorcycle riders.
By leveraging advanced sensor technology integrated into a
rider’s helmet, the proposed fall detection framework has the
potential to accurately detect falls in real-time and promptly
alert emergency services. Future research and development
efforts should focus on enhancing accuracy, optimizing power
consumption, improving communication capabilities, refining
the user interface, and expanding the functionality of the
helmet. By addressing these aspects, the proposed fall de-
tection framework can significantly contribute to enhancing
rider safety and reducing the risks associated with motorcycle
accidents.
IV. PROPOSED SYSTEM
In the planned study, a smart helmet system is introduced
that can recognise when a rider has fallen and instantly inform
emergency personnel, notifying them of the exact location
of the incident. Inertial sensors built inside the helmet are
used in the system, and an ARDUINO-based microcontroller
controls them. 28 experiments were conducted to assess the
system’s efficacy, and 22 statistical features of fall detection
were examined.
By detecting falls in real-time and facilitating prompt as-
sistance, the suggested system’s main goal is to increase rider
safety. The system’s goal is to speed up emergency response
times and increase survivability rates in the case of an accident
by detecting falls and alerting authorities as soon as possible.
The severity of injuries can be negatively impacted by slow
response times. A set of 28 experiments were run to gauge
the system’s performance. These tests attempted to imitate
different fall scenarios and gather information about how
well the system detected falls. The system’s performance and
ISBN: 978-93-5906-287-7 133

possible areas for development were revealed via statistical
analysis of 22 different features. The system’s capabilities and
constraints may be fully understood thanks to this evaluation
method.
The proposed system shows tremendous promise, but there
are still a number of areas that might use additional study
and development. The fall detection algorithm’s accuracy and
dependability could be improved, which is one area that
merits additional investigation. The system may distinguish
between actual falls and other motion-related events bet-
ter by modifying the algorithm and implementing machine
learning techniques, lowering false positives and improving
overall.Optimising power usage is also essential to guaran-
teeing the system’s functionality and lifetime. The system can
improve the battery life, enabling continuous operation without
frequent recharging or replacement, by using effective power
management approaches, such as modifying sensor sampling
rates and investigating low-power modes.
Additionally, improving the user interface can help users
embrace the system and make use of it. Riders can be helped
to comprehend and react appropriately to fall detection alarms
by designing a simple and user-friendly interface with clear
visual and audio clues. The efficiency of the safety measures
offered can be increased with a smooth connection between
the system and the rider.
Finally, it can be said that the suggested smart helmet
system has the potential to increase rider safety by seeing falls
and rapidly notifying emergency personnel of their precise
location. The system’s effectiveness was assessed through
meticulous experimentation and statistical analysis, revealing
both its benefits and potential weaknesses. To further improve
the system’s accuracy, dependability, and general usefulness,
future research and development efforts can concentrate on
streamlining the fall detection algorithm, maximising power
consumption, and improving the user interface. In the end,
these developments may aid in reducing response times,
enhancing rescue operations, and maybe saving lives in mo-
torbike accidents.
The proposed methodology for riders fall detection is di-
vided into four stages, i.e., helmet design, data acquisition,
feature extraction and selection, and finally, classification as
shown in Fig. 1.
A. Helmet Design
The MPU6050 sensor, which includes a 3-axis accelerom-
eter and gyroscope, is used in the proposed smart helmet.
These sensors are essential for detecting and monitoring the
motion and orientation of the helmet wearer. The ARDUINO-
based microcontroller serves as the control module and is in
charge of gathering sensor data and Bluetooth transmission to
a smartphone application.
The system’s different parts, such as the NodeMCU12
microcontroller, microSD card, batteries, and connections, are
safely and firmly packed inside the helmet unit to guarantee
stability and dependability. Due to the layout, there is less
Fig. 1. Architecture of proposed system.
chance of excessive component movement leading to damage
or disconnection when in use.
In order to analyse and analyse the sensor data in real-
time, the microcontroller built on the ARDUINO platform is
essential. It gathers data from the MPU6050 sensor, applies
pertinent thresholds and algorithms to recognise fall events,
and starts notifications from being transmitted to the associated
smartphone app.
The smart helmet establishes a wireless link with a smart-
phone application by utilising Bluetooth technology. The
seamless data transfer provided by this connection enables the
transmission of fall detection alarms and the position data of
the helmet to the smartphone. In order to facilitate quicker
response times and potentially life-saving treatments, the
smartphone application can use this data to immediately alert
emergency services or specified contacts about the autumn
occurrence.Overall, the smart helmet system’s integration of
the MPU6050 sensor, ARDUINO-based microcontroller, and
Bluetooth technology makes it possible to detect falls in real-
time and transmit pertinent data to a smartphone application.
The helmet unit’s small configuration of parts maintains sta-
bility and reduces the possibility of component displacement.
The smart helmet provides a viable alternative for boosting
rider safety and facilitating quick help in the event of a fall or
accident by fusing various technologies.
B. Data Acquisition
By fastening the proposed helmet on a dummy, which is
then fastened to a bicycle seat, it is submitted to testing. During
the tests, the helmet is activated and the bicycle is moved
ahead along a straight course. At a sampling rate of 10 Hz,
the helmet’s accelerometer and gyroscope sensors gather data.
There are a total of 30 trials, with 15 left-side falls and 15
right-side falls. Due to corrupted data files, two trials are not
included in the study. The gathered information is then divided
into fall and non-fall categories based on certain checkpoints,
producing a dataset of 56 occurrences for additional analysis.
ISBN: 978-93-5906-287-7 134

C. Feature Extraction and Selection
The process of turning the raw data into useful features
that can be used for fall detection comes after acquiring the
accelerometer and gyroscope data from the trials. 22 statistical
features are taken from the data in this study. These features
record different facets of the experimenter’s helmet’s motion
and orientation.
Each axis of the accelerometer and gyroscope
data—namely, x, y, and z—is subjected to a different
feature extraction procedure. The 22 statistical features
are computed for each axis, yielding a feature vector
length of 22 * 3 * 2, or 132 features, in total. Numerous
statistical properties, including the mean, standard deviation,
variance, maximum, minimum, skewness, kurtosis, and other
descriptive statistics, are included in these aspects.The process
of choosing the most pertinent features for fall detection
follows feature extraction. In this work, the wrapper strategy
is used, which entails utilising a machine learning algorithm
to assess the performance of various subsets of features. With
this method, the predictive potency of both individual features
and feature combinations may be evaluated. In the wrapper
approach, various feature subsets are iteratively chosen before
being used to train a machine learning model. The model’s
performance is then assessed using particular measures, like
accuracy or F1 score. This procedure is repeated for several
feature subsets, enabling comparison and the choice of the
best feature set.The study uses the wrapper approach to
choose the subset of features that offers the most precise and
trustworthy fall detection. The effectiveness of the model is
assessed based on its combined effect, hence this method
considers the interdependencies and interactions among the
features.
Overall, the use of machine learning techniques in conjunc-
tion with the wrapper approach enables the evaluation and
selection of the most useful characteristics for fall detection.
By concentrating on the most pertinent and discriminative
features, this method helps to optimise the performance of the
fall detection system, improving its accuracy and reliability in
detecting falls and distinguishing them from regular motion
patterns.
D. Classification
The last stage is to identify a fall using a set of features.
This study evaluates the following classifiers for smart helmet-
based rider fall detection.
E. Naive Bayes (NB)
In order to estimate the likelihood of each class given a
collection of input features, the Naive Bayes method is used.
Using training data, it determines the conditional probabilities
of the features given each class. The approach creates a
probabilistic model during the training stage by estimating the
class priors and feature likelihoods.
The technique uses the Bayes theorem to determine the
posterior probability of each class given the input features
while making predictions on new data. The projected class
for the input data is then chosen to have the highest posterior
probability.
Naive Bayes’ efficiency in handling high-dimensional data,
even when the number of features is vast, is one of its
key advantages. Additionally, it works effectively when the
independence assumption is true.
F. K-Nearest Neighbor (KNN)
Popular supervised machine learning algorithms for classi-
fication and regression tasks include the k-nearest neighbours
(KNN) technique. Being non-parametric, it makes no assump-
tions on the distribution of the underlying data.
The foundation of KNN is the idea that objects or data
points with similar properties likely to be members of the same
class or have comparable values. To perform predictions, the
algorithm keeps the complete training dataset in memory. In
order to identify the class or value of a new data point, KNN
first locates the k data points in the training set that are nearest
to the new one in terms of a distance metric (such as Euclidean
distance), assigns the majority class to those k neighbours, or
averages their values.
G. Random Forest (RF)
The strong and adaptable ensemble learning method Ran-
dom Forest (RF) is frequently used for both classification
and regression applications. It is a development of decision
tree algorithms that integrates the forecasts of various distinct
decision trees to produce more precise and reliable forecasts.
A group of decision trees are created using the Random For-
est algorithm, and each tree is trained using a random subset
of the input features and training data. This unpredictability
lowers the possibility of overfitting while introducing variation
among the individual trees.The following are the main phases
in creating a Random Forest:
Random sampling with replacement (bootstrapping): Create
several bootstrap samples by randomly choosing portions of
the training data with replacement.
Random feature selection: Choose a subset of features to
take into account when dividing each node of a decision tree
at random. This adds variability and prevents the forecasts
from being dominated by a few features.
Construction of decision trees: Build decision trees utilis-
ing the bootstrap samples and chosen features. Recursively
dividing the data based on feature thresholds to maximise
information gain or minimise impurity (such as Gini index
or entropy) is how each tree is constructed.
After the trees have been trained, predictions are made by
combining their individual forecasts.In conclusion, Random
Forest is a flexible ensemble learning system that mixes the
forecasts from several decision trees. It is renowned for its
dependability, capacity to handle different data circumstances,
and evaluation of feature relevance. Due to its excellent
prediction accuracy and adaptability to a variety of machine
learning problems, Random Forest is widely employed in
practise.
ISBN: 978-93-5906-287-7 135

CONCLUSION
In conclusion, the development of a smart helmet with fall
detection and automated rescue delay prevention features is
a significant advancement in the field of rider safety. The
integration of accelerometer and gyroscope sensors, coupled
with an Arduino-based microprocessor, has allowed for the
detection of falls and the immediate transmission of location
information via a GSM module to nearby rescues.Additionally,
the use of a smart helmet with impact detection capabilities
can provide riders with valuable information and assistance in
the event of an accident, potentially saving lives and reducing
the severity of injuries. The ability to collect data on the
acceleration experienced during impacts can also be used to
improve the design of protective devices and advance research
in the field of rider safety. Overall, the development and im-
plementation of smart helmets with fall detection and impact
detection capabilities can greatly contribute to the prevention
of injuries and the promotion of safer riding practices.
ACKNOWLEDGEMENT
We express our sincere gratitude to Dr. P Sojan Lal,
Principal for providing us the research facilities and the
circumstances to complete the project in a grant successful
manner. We would like to express our sincere gratitude to
the following individuals and organizations for their valuable
contributions to this project. First and foremost, we would
like to thank our guide Ms.Teena Skaria, for her guidance and
support throughout the project.
REFERENCES
[1] https://www.researchgate.net/publication
[2] T. Vilarinho et al., ”A Combined Smartphone and Smartwatch Fall
Detection System,” 2015 IEEE International Conference on Computer
and Information Technology; Ubiquitous Computing and Communi-
cations; Dependable, Autonomic and Secure Computing; Pervasive
Intelligence and Computing, Liverpool, UK, 2015, pp. 1443-1448, doi:
10.1109/CIT/IUCC/DASC/PICOM.2015.216.
[3] N. Nataraja, K. S. Mamatha, Keshavamurthy and Shivashankar,
”SMART HELMET,” 2018 3rd IEEE International Conference on Re-
cent Trends in Electronics, Information Communication Technology
(RTEICT), Bangalore, India, 2018, pp. 2338-2341, doi: 10.1109/RTE-
ICT42901.2018.9012338.
[4] S. -H. Fang, Y. -C. Liang and K. -M. Chiu, ”Developing a mobile phone-
based fall detection system on Android platform,” 2012 Computing,
Communications and Applications Conference, Hong Kong, China,
2012, pp. 143-146, doi: 10.1109/ComComAp.2012.6154019.
[5] S. Li, ”Fall Detection With Wrist-Worn Watch by Observations in
Statistics of Acceleration,” in IEEE Access, vol. 11, pp. 19567-19578,
2023, doi: 10.1109/ACCESS.2023.3249191.
[6] https://www.researchgate.net/publication/317672343
[7] Tinetti ME, Williams CS. Falls, injuries due to falls, and the risk of
admission to a nursing home. N Engl J Med. 1997 Oct 30;337(18):1279-
84. doi: 10.1056/NEJM199710303371806. PMID: 9345078.
[8] Suhuai Luo and Qingmao Hu, ”A dynamic motion pattern analysis
approach to fall detection,” IEEE International Workshop on Biomed-
ical Circuits and Systems, 2004., Singapore, 2004, pp. 1-5, doi:
10.1109/BIOCAS.2004.1454088.
[9] https://doi.org/10.1016/j.pmcj.2012.08.003
[10] S. Abbate, M. Avvenuti, G. Cola, P. Corsini, J. Light and A. Vecchio,
”Recognition of false alarms in fall detection systems,” 2011 IEEE
Consumer Communications and Networking Conference (CCNC), Las
Vegas, NV, USA, 2011, pp. 23-28, doi: 10.1109/CCNC.2011.5766464.
ISBN: 978-93-5906-287-7 136

SkullCap – An IoT based Smart Helmet for Accident Detection - Report

More Related Content

Similar to SkullCap – An IoT based Smart Helmet for Accident Detection - Report (20)

Recently uploaded (20)

SkullCap – An IoT based Smart Helmet for Accident Detection - Report