2nd IEEE International Conference on Computational Systems and Information Technology for Sustainable Solutions 2017
Smart Helmet for Accident Detection and
Notification
Shoeb Ahmed Shabbeer
sashoeb94@gmail.com
Dept. of ISE, RV College of Engineering
Bangalore, India
Abstract — Aim to create a system utilizing the
Internet of Things (IoT) concept to detect accidents in two
wheelers using a microcontroller and accelerometer and
notify the authorities as well as the emergency contact of
the rider. Also, plot the accidents occurring in the city for
further analysis.
Keywords—internet of things; accident detection;
reporting; embedded systems;
I. INTRODUCTION
The Transport research wing of the Ministry of road transport and
highways recently published its report in 2015 on the road accidents
that occur in the country. Various facets of the accidents like trends,
severity, injuries and fatalities among others was described in detail.
During 2015, close to 5 lakh road accidents took place in which 1.46
lakh people were killed and approximately 5 lakh people were
injured. Among these statistics 1.44 lakh accidents belonged to the
two wheeler category in which thirty six thousand people were killed
in thirty four thousand number of accidents.
The public apathy is high in regard to the accidents with most people
choosing to ignore the accident that has happened which leads to
delay in the reporting of the accident to the authorities and
subsequent delay in emergency response leading to loss of life or
impairment.
IoT comprises of smart machines and sensors interacting with other
machines using an underlying network infrastructure. The IoT
devices sense the data from the environment and send the data to
each other or a central server. This leads to huge amount of data
generated which can be further processed and analysed to generate
useful conclusions. This gives the advantage of remote monitoring
the environment and insights into real time data reported from the
environment.
The main aim of the system is to accurately and reliably detect
accidents that occur in two wheelers and also report the accident to
the authorities and the emergency contact of the accident victim. The
accident detection is facilitated by the use of a microcontroller which
gets the data from the accelerometer and sends a notification via a
GSM device if an accident has occurred.
31 978-1-5386-2044-1/17/$31.00 ©2017 IEEE
Merin Meleet
merin.meleet@gmail.com
Dept. of ISE, RV College of Engineering
Bangalore, India
The various devices and services used in this system for the accident
detection and notification purpose are –
1. Arduino Microcontroller
2. MPU6050 Accelerometer
3. Sim900 GSM Module
4. NEO6M V2 GPS Module
5. Web Server
II. RELATED WORK
IoT is a relatively new field which is seeing rapid advancements in
technology due to the immense benefits it offers. S. Chandran et al.
developed a similar system to detect and report accidents [1]. The
system under development holds much more features and
improvements. M. Fogue et al. highlighted the use of V2V for
vehicle assistance [3]. A lot of systems have already been developed
for four wheelers to detect accidents but the field of two wheeler
accident detection is still in its infancy. J. White et al. have
developed a system to detect accidents with the help of smartphones
[4]. This system uses the inbuilt accelerometers in smart phones to
detect accidents and the inbuilt connectivity features of smartphones
for accident notification. This system is not accurate and reliable as
smartphones are susceptible and experience a lot of sudden
acceleration changes which can be triggered as false accidents. S.
Hari Sankar et. al. identify the main cause of fatalities due to
accidents [5]. They have identified the critical golden hour and in
this period dispatch ambulances to the accident spot. E. Nasr et. al.
also have come up with a system which detects accidents and relays
the accident location to the Public Safety Organization [6]. Amit
Meena et. al. have also come up with a system to combat the
increasing mortality rate of two-wheeler accidents in India [7]. The
system however cannot plot accidents and thus identifying trends is
not possible. Bruno Fernandes et. al. aim to detect accidents in four-
wheelers using a smartphone app [8]. The app interfaces the inbuilt
sensors of the smartphone to detect any anomalies which may be an
accident. Using a smartphone to detect accidents lowers the
reliability as false positives might be triggered.
The 802.11p working group had recently passed the IEEE 802.11p
standard for V2V security applications. A lot of inroads have been
made for accident detection in four wheelers but a lot of research is
yet to be made in the two wheeler department. Thus the need of the
 2nd IEEE International Conference on Computational Systems and Information Technology for Sustainable Solutions 2017
hour is to develop a low-cost, affordable and intelligent system
which can be utilized efficiently by everyone.
III. PROPOSED ARCHITECTURE
The paper describes the prototype of a smart helmet which aims to
detect and report accidents occurring in two wheelers.
Microcontroller interfaced with accelerometer and GSM module
and cloud service infrastructures are utilized to achieve the final
objectives of notification and reporting. The helmet is developed in
a way to detect an accident in two wheelers and send the
geographical co-ordinates of the accident to the emergency
authorities and the emergency contacts of the victim. A 6-axis
accelerometer is mounted on the helmet which continuously
monitors the acceleration levels of the helmet. When an accident is
detected due to inconsistent acceleration levels and exceeds the
threshold, it gathers the GPS co-ordinates from the GPS module and
sends a message to the emergency authorities’ web server which
then sends an emergency message to the assigned emergency
contact of the victim.
Figure.1 Architecture of the system
A. Challenges faced for the development of the system
1. Cost of the system
Affordability of the system is one of the major objectives. If the end
cost of the system is very high then adoption rate will be very low.
The end cost of the system would not be more than a couple
thousand Indian rupees.
2. Accuracy in detection of accidents
Change in acceleration is the key indicator for detection of accident.
If the user accidently drops his helmet or gives sudden acceleration
or braking then there are chances that it might be treated as an
accident.
3. Battery considerations
The system should have a decent battery life and be able to work all
day long. For this purpose various hardware devices and algorithm
changes have been made to maximise the battery life.
B. Major Objectives
The major objectives of the system can be summarized into –
1. Force the riders to wear their helmets. As per the law and
to maximise the survivability of the rider in case of an
978-1-5386-2044-1/17/$31.00 ©2017 IEEE
accident, wearing helmet is must. If the rider is not
wearing his helmet, he should not be able to start his two-
wheeler.
2. Detect accidents reliably by removing false positives. The
threshold for acceleration change to be classified as an
accident should be suitable. Too high a threshold would
discard the accidents and too low a threshold would lead
to false positives.
3. Report accidents to authorities and to the assigned
emergency contact. The notification should also have the
location of the accident so that help can be sent
immediately.
4. Consolidate and plot the accidents on a map to prevent
future accidents by identifying accident hot-spots. This is
done with the help of Google maps.
C. Working Principle
The smart helmet provides an affordable and effective accident
detection and notification system to address the problems faced by
the two wheeler riders. The helmet contains a microcontroller which
gets the acceleration data from the 6 axis accelerometer which also
contains a gyroscope. A GSM module is also present in the system
which is responsible for communication with the web server. The
smart helmet communicates with the web server using a REST
HTML architecture. The architecture of the system is described in
Figure 1.
1. Accident Detection
Accident detection is facilitated with the help of MPU6050 which is
a 6-axis accelerometer and gyroscope which detects the acceleration
values along the positive and negative x, y and z axis and the angle
respectively.
Consider a bike which is moving along the road. The road ahead of
the bike is the positive x-axis while the road behind is the negative
x-axis, the area to the right and the left of the rider is the positive
and negative y-axis respectively. The road immediately below the
two-wheeler represents the y-axis.
When the bike is stationary, there are no forces on the bike apart
from the acceleration due to earth’s gravity (~1g) which acts along
the negative y-axis.
Figure 2 Axis representation of two-wheeler
Consider the case when the two-wheeler is travelling on a road and
a collision happens when a vehicle impacts on to the back of the two
wheeler. Figure 3 represents the forces involved in the collision.
32
 2nd IEEE International Conference on Computational Systems and Information Technology for Sustainable Solutions 2017
Time period 0-200ms shows the constant acceleration experienced
by the two wheeler when it is moving at a constant speed. Time
period 200-250ms shows the added acceleration experienced by the
rear end collision to the two wheeler. This period shows a great
change in acceleration from 2.4m/s at 200ms to 6.5m/s at 250ms. An
acceleration change of 4.1m/s is experienced by the helmet of the
two wheeler rider in a time frame of 50ms which is more than
enough force to cause an accident. Time period 250-300ms is the
negative acceleration experienced due to the whiplash effect on the
neck and helmet.
Figure 3 Forces involved in a rear end collision
A similar change in acceleration also happens when the two-wheeler
meets a frontal collision or a side collision and the respective axes
show the impact.
This large change in acceleration in a short span of time is what the
microcontroller aims to detect which would signify an accident. The
threshold has to be set appropriately so as to not cause any false
alarms.
The lean angle also plays an important part as the two-wheeler can
also tip over sliding along the road without any significant change
in acceleration.
Figure 4 Motorcycle Lean Angle
If the change in angle is also high in a short period of time or the
angle of the two-wheeler goes above a certain angle (most road
going two wheelers have a max lean angle of around 45 degrees) it
33 978-1-5386-2044-1/17/$31.00 ©2017 IEEE
signifies that an accident has occurred due to the vehicle tipping over
as depicted in Figure 4.
The algorithm for accident detection can be summarised as, (the
initial values of acceleration and angles are 0)
1. Get acceleration and gyroscope values from the
accelerometer.
2. Save the acceleration and the gyroscope values as new
values.
3. Check if the change in acceleration exceeds the threshold
or the lean angle exceeds the threshold by the help of the
current and new acceleration and gyroscope values.
4. Store the new acceleration and gyroscope values as the old
values.
5. Repeat from step 1 every 50ms.
2. Sending notification
The exact location of accident is provided by a NEO6M-V2 GPS
module. It consumes a voltage of 5V DC which is provided by a
battery pack. When an accident is detected, a HTML GET request is
prepared which contains the vehicle ID, latitude and longitude as
well as the type of vehicle(system can be expanded to be included
in four wheelers as well). The coordinates of the two-wheeler are
provided every second as a NMEA string. The NMEA string is then
decoded to reveal the latitude and longitude. The decoded GPS
information is then added to the HTTP GET request and sent to the
web server via the GSM module. The GSM module is only active
during data transmission and hence consumes very little battery.
The web server upon receiving the GET request stores the URL
parameters (VID, latitude, longitude, vehicle type) into a database.
Time of the accident is also added. Subsequently, an email (or other
form of notification) is generated to be sent to the emergency contact
and authorities which contains the vehicle ID, coordinates of the
accident, time of the accident and the vehicle type (can be expanded
to other types of vehicles) also.
The algorithm can be summarised into two parts-
a. Hardware notification
1. Get current GPS location from the GPS module.
2. Decode the GPS NMEA string to reveal the latitude and
longitude.
3. Incorporate the vehicle ID, location data and the vehicle
type into a GET request.
4. Connect to the web server using the GSM module.
5. Send the GET request
6. Disconnect from server.
b. Webserver notification
1. Receive the GET request parameters into variables.
2. Open connection to underlying database.
3. Write the accident details to the database including time.
4. Close connection to database.
5. Send notification to the emergency contact and the
authorities.
3. Plotting Accident data
 2nd IEEE International Conference on Computational Systems and Information Technology for Sustainable Solutions 2017

The geographic coordinates of the accidents are stored in a database
which allows for persistent storage of the accident coordinates.
The plotting of the accidents is done to identify the accident hotspots
in one glance which is useful in many ways. Ambulances can be
placed near accident prone areas for quicker emergency response.
Accident prone roads can be re-modelled to reduce the number of
accidents that occur due to improper road/intersection design.
The time taken to send the notification after the accident has been
detected is less than 10 seconds on an average. The web server
receives the correct coordinates and stores the accident data before
forwarding the accident coordinates to the emergency authority.
This notification is in the form of an email which is shown in Figure
5. This can be extended to SMS also.

Google provides an API for developers to plot a google map. The

API allows for custom road/satellite maps among others to be
created with custom markers. The system shows two kinds of maps
i.e., marker map and the heat map for the accidents to the authorities.
The map takes all the data points from the database via a script and
parses them to a google standard location format. Figure 5 Accident Notification to emergency authority
The Google maps API key can be generated for free by registering Figure 6 shows the distribution of accidents on the map as markers
yourself as a google developer and visiting the google developer site with each marker showing the type of vehicle, vehicle number and
and applying for a key. However, the API is free for use up to a the geographic location.
certain limit per day above which payment is required.
Figure 7 shows the heat map of accidents that have occurred. More
The marker map is a custom map which is created to show the the accidents are clustered in an area, darker is the colour. This
geographic coordinates of an accident. The marker shows the allows the authorities to identify the accident hotspots in a city and
vehicle number and the vehicle type when clicked on. The heat map can hence take precautionary and preventive measures to reduce the
represents the accident data points in the form of colours. Higher the number of accidents in that area in the future.
density of accidents in an area, darker the colour which signifies the
presence of an accident hotspot.

The algorithm can be summarised as,

1. Create a google custom map using the developer key

assigned.
2. Give the initial coordinates for the map to centre on.
3. Load all the accident coordinates from the database.
4. Bind each accident coordinate with the vehicle number in
the form of a map marker.
5. Display all the accident markers on the map.

IV. RESULTS
The system also contains a method to ensure that the rider wears his Figure 6 Marker map
helmet without which his vehicle would not start. An array of bump
switches are installed in the helmet which give a signal to the micro
controller indicating that the rider has worn his helmet. If no signal
is received by the microcontroller, then it does not allow the engine
ignition to take place or switches off the engine in case the rider
removes his helmet while riding.

Results show that the helmet was able to correctly identify the
accident impacts 275 times out of the 290 times in which 15 were
false positives i.e. an accuracy of 94.82% and sends the notification
with the correct and current coordinates 96.72% (266/275) of the
time. This accident detection result is based on the threshold level
set at 20,000 units of acceleration change in a time span of 50ms.
For testing purposes, the change in acceleration threshold has been Figure 7 Heat map
set appropriately. The helmet does not beep when it is
accelerated/decelerated gradually as would be the case in real life. It V. CONCLUSIONS AND FUTURE WORK
beeps and sends a notification with the geographic coordinates when
The smart helmet system is designed to be a low cost solution to
a sudden acceleration is detected in a short span of time or the two-
report and monitor the two wheeler accidents. It is a cheap and
wheeler slips and slides.
viable solution which is easy to operate and in line with the Digital
India vision of the government. The system also offers additional

978-1-5386-2044-1/17/$31.00 ©2017 IEEE 34

 2nd IEEE International Conference on Computational Systems and Information Technology for Sustainable Solutions 2017

benefits which are not present in the other systems. Most of today’s
accident detection and prevention systems are built for four-
wheelers. The smart helmet could prove to be a life saver for the
numerous two-wheeler riders in the country and the world by
ensuring that help reaches them when they meet with an accident
without any delay.

Additionally, various add-ons can be made in order to scale-up the

system and make it compatible for all types of vehicles on the road.
An additional sensor can also be incorporated to detect if the vehicle
operator is under the influence of alcohol. Additional analytics can
be done on the accident data to view hidden accident trends which
were not visible in the first glance. The system can also be made to
function as a remote immobilizer in case the vehicle is stolen and
the location of the vehicle can also viewed immediately.

REFERENCES

[1] S. Chandran, S. Chandrashekhar, E. Elizabeth N, “Konnect: An

Internet of Things (IoT) based Smart Helmet for Accident Detection
and Notification”, India Conference (INDICON), 2016 IEEE
Annual.

[2] Ministry of Road Transport and Highways, Transport Research

wing, “Road Accidents in India”, 2015.

[3] M. Fogue, P. Garrido, F. J. Martinez, J. C. Cano, C. T. Calafate,

P. Manzoni, “A System for Automatic Notification and Severity
Estimation of Automotive Accidents”, IEEE Transactions on
Mobile Computing, March 2013.

[4] J.White, C. Thompson, H. Turner, B. Dougherty and D. C.

Schmidt. “WreckWatch: Automatic Traffic Accident Detection and
Notification with Smartphones”, Mobile Networks and
Applications, June 2011.

[5] S. Hari Sankar, K. Jayadev, B. Suraj, P. Aparna, “A

comprehensive solution to road traffic accident detection and
ambulance management”, International Conference on Advances in
Electrical, Electronic and Systems Engineering (ICAEES), 2016.

[6] E. Nasr, E. Kfoury, D. Khoury, “An IoT approach to vehicle

accident detection, reporting, and navigation”, Multidisciplinary
Conference on Engineering Technology (IMCET), IEEE
International, 2016.

[7] Amit Meena, Srikrishna Iyer, Monika Nimje, Saket Joglekar,

Sachin Jagtap, Mujeeb Rahman, “Automatic Accident Detection
and reporting framework for two wheelers”, Advanced
Communication Control and Computing Technologies (ICACCCT),
2014.

[8] Bruno Fernandes, Vitor Gomes, Joaquim Ferreira, Arnaldo

Oliveira, “Mobile Application for Automatic Accident Detection
and Multimodal Alert”, IEEE Vehicular Technology Conference
(VTC Spring), 2015.

35 978-1-5386-2044-1/17/$31.00 ©2017 IEEE