International Refereed Journal of Engineering and Science (IRJES)

ISSN (Online) 2319-183X, (Print) 2319-1821

Volume 8, Issue 2 Ser I (Feb 2019), PP. 26-36

Driving Safely in Smart Cars in a Smart Road Environment

Keshav Luchoomun,
Ubiquitous Computing,

ABSTRACT :- Smart cars are promising application domain for ubiquitous computing. Context-awareness is
the key features of a smart car for safer and easier driving. Context recognition is important support for a smart
car to avoid accidents proactively. Regardless of many industrial innovations and academic progresses, there is
still today a lack of fully context-aware smart cars. This paper presents an overview of the related works on
different smart cars and smart road environment systems. The nature of context awareness in smart cars and
road environment is also presented. Moreover, the term paper proposes a hierarchical context model for
description of the complex driving environment. Smart cars continuously connectivity and sharing information
under intelligent mobility has been put forward. A smart car prototype including software platform and
hardware requirements is built to provide the running environment for the context model and application.
Performance metric shows the accuracy of context situation recognition.
KEYWORDS :- Context-aware, seamless connectivity, smart car, smart road, vehicle-to-vehicle communication
(VVC), vehicle-to-infrastructure communication (VIC).
----------------------------------------------------------------------------------------------------------------------------- ----------
Date of Submission: 11-06-2019 Date of acceptance: 28-06-2019
----------------------------------------------------------------------------------------------------------------------------- ----------

I. INTRODUCTION
Cars have become part of people‟s lives. These automobiles bring convenience and ease to our lives for
easy mobility from places to other places; go work, parents bring kids to school without getting late, in
emergency situations where a transport mean is needed to go urgently to hospital, for vacation trips and other
daily activities. However, many problems also arise, such as traffic congestions and accidents. A smart car aims
at assisting its driver with easier driving, less workload and less chance of getting injured [1]. Smart car must be
aware of the environment, which means sensing and recognizing the context around the car and the driver.
Context-awareness is the key feature of a smart car for safer and easier driving.
The tomorrow‟s vision of driving in a city which can interact, act and respond to the driver‟s needs
would be far from reality some 20 years back. Today, in the 21 st century, such smart road environments are
already being built around the world. Smart car must be able to sense, analyze, predict and react to the road
environment. In a near future, everything in a city, from the road directions, roads‟ lighting, traffic signaling will
be connected in a network. Smart cars or vehicles embedded with artificial intelligence will allow self-driving;
finding space parking on its own intelligence without driver maneuvering on the steering.
Obtaining such broad view of the environments of both roads and cars is only possible through
collaboration between the infrastructures and car‟s sensing abilities. The aims are to develop and test the
technology that will enable such cooperation, which will increase the amount of information available to driver.
Moreover, by combining quality data from roadside sensors and data sent directly from vehicles in vicinity,
advance knowledge can be gained about potential safety risks. This research paper focuses on how to build a
context-aware smart car.
The remainder of this paper is organized as follows. Section 2 introduces to related works on smart cars
and smart roads. The evaluation and problems of the existing works is given in Section 3. Nature of Context-
awareness in Smart Cars and Road environments is presented in Section 4. Section 5 shows the proposed system
of smart cars and Section 6 presents the evaluation performance of the smart cars system and Section 7 is about
experimentation of accuracy of context situation recognition. The conclusion is given in Section 8.

II. RELATED WORKS

In the past decades, lots of technologies have been developed in the field of intelligent transportation
system (ITS) (Wang et al.., 2006) [2] and the advanced driver assistant system (ADAS) (Küçükay and Bergholz,
2004) [3]. However, current smart cars are not in actual fact context-aware. Only a few types of the information
of road environments, referred as contexts, are utilized. The weaknesses are that most of current smart cars lack
complex reasoning which limit the smart car‟s ability of assisting the driving task efficiently and safely. The
following describes major progress, recent advances, ongoing researches, and research projects being carried out
in this field.

www.irjes.com 26 | Page
 Driving Safely in Smart Cars in a Smart Road Environment

Communication and cooperation – At the Intelligent Transport System test site, Toyota is testing smart cars
that talk to each other and to the roads on the communication frequency of 700MHz band with the main goal of
reducing traffic accidents. According to Toyota, the site is equipped with a road-to-vehicles communication
system, consists of a vehicle detection system, a pedestrian detection system, a course monitoring system, traffic
signal and control devices [4]. The Vehicle-Infrastructure Cooperative Systems support driving, and aim to
avoid traffic accident by notifying drivers of the context captured through communication between the vehicles
and sensors installed on the road, or among vehicles. Since 2009, Toyota has commercialized some functions of
the system after participating in some providing tests held by the relevant governmental agencies and
automotive industries [5]. (a) The DSSS function, Driving Safely Support Systems, are designed to focus on
road environment, to convey information/context about vicinity vehicles, motorcycles, and pedestrians that are
outside the visual range of the drivers, as well as, traffic control information such as traffic signals, traffic state
(congestion or light traffic). The sage driving operation of the system started in operation in 2011. (b) Efforts for
communication System between Vehicles and Pedestrians or Among Vehicles. The new generation of vehicle
Infrastructure Cooperative System uses direct communication between vehicles and pedestrians or among
vehicles for continuous information exchange in order to prevent frontal collision accidents at intersections
difficult for drivers to see. Toyota participates actively in developing the systems to contribute to the
environment reform and to improvement of traffic flows [6]. To motivate the description of the communication
and cooperation system, an example of smart cars talking to each other and to the road is given as follow; smart
cars at Intelligent Transport System site capture real-time context from sensors and transmitters installed on
streets to reduce risk of accidents in situations such as missing a red traffic light, cars advancing from blind
spots and pedestrians crossing the street. The system also tests that smart cars talk to each other. In a test drive,
the presence of a pedestrian triggered a beeping sound in the car and a picture of a person popped up on the
screen in front of the driver. Picture of an arrow popped at intersection, indicating an approaching vehicle. The
system outputs a voice message “It‟s a red light”, if the driver was about to ignore a red light.
Driver assistant system – Many novel ideas have been implemented in newest series of concept cars by
automobile manufacturers. (a) BMW‟s ConnectedDrive includes BMW Assist, BMW Online and driver
assistance systems, supporting lane change warning and parking assistant (Hoch et al.., 2007) [7]. (b) Mercedes-
Benz is developing an intelligent driver assistance system that utilizes stereo cameras and radar sensors to
monitor surroundings around the car (Benz, 2007) [8]. (c) Advance pre-collision system, dynamic driving,
electronic brake assistance system and park-assistance systems for active safety are being installed by Lexus on
its LS-series (Lexus, 2007) [9]. (d) Furthermore, Toyota announced two newly developed safely system; an
intelligent Clearance Sonar and Drive-start Control, target to eliminate traffic problems (congestions and
accidents).
Collision avoidance system – With pedestrian fatalities taking an unexpected rise in recent years, automakers
are looking for ways to not only reduce the death toll but quite literally steer clear of pedestrians in the first
place [10]. The automobile safety system is designed to reduce the severity of an accident. Usually uses radar
and laser or camera sensors to detect an imminent crash, and provide an warning to the driver when there is an
imminent collision or take action autonomously without any driver input (by braking or steering or both). (a)
Ford‟s Collision Warning with brake Support was introduced in 2009 on the Lincoln MKS and MKT and Ford
Taurus [11]. This system provides a warning through a Head Up Display that visually resembles brake lamps. If
the driver does not react, the system pre-charges the brakes and increases the brake assist sensitivity to
maximize driver braking performance. Ultimately, the smart vehicle will automatically attempt to steer away
from the pedestrian [12]. Furthermore, a new Pre-collision system by Toyota includes collision avoidance
assistance that is effective in high-speed collisions and more than 90% of speed ranges in rear-end collisions
[13].
Vehicular ad hoc networks (VANETs) – VANETs have emerged as an application of mobile ad hoc networks
(MANETs), which use dedicated short-range communication (DSRC) to allow vehicles in close proximity to
communicate with each other or to communicate with roadside equipment. Applying wireless access technology
in vehicular environments has led to improvement of road safety and a reduction in the number of fatalities
caused by road accidents through development of road safety applications and ease of information sharing
between moving vehicles regarding the road.
Intelligent vehicular ad-hoc network (InVANET) integrates multiples networking technologies such as Wi-Fi
IEEE 802.11p, WAVE IEEE 1609, WiMAX IEEE 802.16, Bluetooth, IRA and ZigBee [14]. VANETs
are expected to implement technologies such as dedicated short-range communication which is a type of Wi-Fi.
Other candidate wireless technologies are cellular and satellite. As promoted in intelligent transportation
systems (ITS), vehicles talk with each other via inter-vehicle communication (IVC) as well as roadside base
stations via roadside-to vehicle communication (RVC). Providing IVC and RVC can considerably improve
traffic safety and comfort of driving and travelling. InVANET can be used as part of automotive electronics,

www.irjes.com 27 | Page
 Driving Safely in Smart Cars in a Smart Road Environment

which identifies an optimally minimal path for navigation with minimal traffic intensity. The system can also
serve as a city guide to locate and identify landmarks in a new city.
Currently there is ongoing research in the field on InVANETs for several scenarios. The main interest is in
applications systems for traffic scenarios, mobile phone systems and sensor networks. Recent research has
focused on topology related problems such as range optimization, routing mechanisms, or address traceability.
New interest is emerging for green InVANETs, in other words, minimal power consumption for sensor
networks [15].
Driver behaviour recognition – The driver plays an important role in a smart car. VANETs also focus on
developing a novel and nonintrusive driver behavior system using a context-aware system, to detect abnormal
behaviours exhibited by drivers and to warn other vehicles on the road to prevent accident from happening [16].
Vigorous efforts are under way today to research and develop partially or fully automated driver behavior
systems, such as those for headway distance control of lane keeping control by making use of Intelligent
Transportation System (ITS) technologies [17]. Machine learning and dynamical graphical models, such as
HMM (Oliver and Pentland, 2000) [18], Gaussian Mixture Modeling (GMM) (Miyajima et al., 2007) [19] and
the Bayesian network (Kumagai and Akamatsu, 2006) [20], can be applied for modeling and recognizing drivers
behaviours. Several researches have examined the development or driver monitoring and detection systems
using these models above.
Most researches focused on building a context-aware smart car by developing a hierarchical model that is able
to collect contextual information (Behaviours: normal, drunk, fatigue, reckless and other uncertain context),
reason about and react, to contextual information about the driver, vehicle and environment, in order to provide
a safe and comfortable driving environment [21]. D. Sandberg and M. Wahde proposed a system for drowsy
driver detection in real time driving by collecting information about the driver‟s behaviour such as speed of the
vehicle, the vehicle‟s lateral position, yawing angle, steering wheel angle and the vehicle‟s lane position. Their
system uses artificial neural networks to combine different indications of drowsiness and to predict whether a
driver is drowsy and issue a warning if required [22]. H. Ueno, M. Kaneda, and M. Tsukino developed a non-
contact system to prevent driver drowsiness by detecting the eyes of the driver and checking whether they are
opened or closed using a CCD camera. The system is based on capturing the face of the driver and using image
processing techniques to check if the driver‟s eyes are closed for long intervals. If so, the system issues a
warning to driver [23].
Driver-Vehicle Interface – The Adaptive Integrated Driver-Vehicle Interface (AIDE) project tries to maximize
the efficiency and safety of advanced driver assistance systems, while minimizing the workload and distraction
imposed by in-vehicle information systems (Kutila et al., 2007) [24]. The Communication Multimedia Unit
Inside Car (COMUNICAR) project aims at designing an easy-to-use on-vehicles multimedia human-machine
interface. An information manager collects the feedback information and estimates the driver‟s workload
according to the current driving and environment situation (Bellotti et al., 2005) [25].
Smart Roads – In U.S, the Federal, State and local governments are investing about tens of billions dollars into
nation‟s infrastructure to make the roads, bridges and other assets much more intelligent. This vision known as
smart road or infrastructure promises to make the nation more productive and competitive, while helping the
environment and saving lives.
“The goal is not just funding projects for short-term job gains, it should be to create systems that are intelligent
and improve productivity in the long run”, says Paul Feenstra, vice president of government affairs at the
Intelligent Transportation Society of America, a group that promotes smart-road technologies. (a) Smart
Bridges, network of sensors at critical points is used for continuous electronic monitoring of bridge structures;
these devices can deliver data about the behaviour of the bridges under heavy traffic, in high winds or other
conditions. Also, hidden spots potentially serious problems such as hidden cracks, erosion after floods, long
before they might be apparent to a human inspector can be detected. “Automatic detection can make the
difference between a major disaster, a costly retrofit or a minor retrofit”, says Mohammed Ettouney, a principal
in Weidlinger Associates Inc., a New York engineering firm [26]. (b) Virginia Smart Road is a unique full-scale
research facility managed by the Virginia Tech Transportation Institute (VTTI), is currently working on lighting
system project. The Smart Road is equipped with variable lighting, multiple luminaire heads and poles mounted
for test simulations like crosswalks for pedestrians, vehicles approaching pole lighting as needed. To illustrate
this scenario; suppose a person driving at night on a winding road on a snowy weather condition and poor road
visibility, approaching a curve way, the person feels safe knowing the smart car is equipped with obstacle
detection sensors and lighting street detectors that will warn the driver to stop or provide additional safety
measures like making street lighting more brighter. The driver feels more confident and comfort [27].

III. EVALUATION OF EXISTING SYSTEMS

Most of the works listed above is not fully context-aware. The reasoning of contexts for further
analysis is not put enough emphasis on. Most current researches focus special practical technologies such as

www.irjes.com 28 | Page
 Driving Safely in Smart Cars in a Smart Road Environment

communication, sensing and driver assistance. These limit smart cars to be cars with certain accessories. There
is a lack of a common consensus and comprehensive understanding of smart cars in a holistic view.
Moreover, the driver behaviour detection system as presented in Section2 focus on the detection of
driver‟s status (drunk, affected by fatigue, drowsy) by monitoring the driver or the vehicle and issuing warning
messages to the driver to prevent road accidents. Whilst these systems have achieved good results in terms of
improving road safety, they are limited to alerting the driver or controlling the vehicle itself. The major concern
in term of context-awareness is these systems currently have not considered the behaviour of the driver as a high
level context (uncertain context).
This study attempts to build a smart car from the viewpoint of context-awareness. All the entities of a
context captured by the drivers and in driving environment will be preprocessed and defined. Reasoning plays
an important role in complex situation analysis. Leading context-awareness to a higher-level of reasoning will
allow development of different services and applications in smart cars without much modification to the current
architecture.

IV. NATURE OF CONTEXT-AWARENESS IN SMART CARS AND ROADS

Context recognition includes raw data collection and preprocessing, feature extracting and selection,
and recognition method. (a) Raw data from various sensors are captured and preprocessing, includes noise
removal, data calibration and transforming of data distributions. (b) Feature extraction and selection, features
extracted from sensor signals referred as context atoms are classified in hierarchical models enables high
performance. (c) Context recognition method, capturing context atoms from several sources at an instant of time
has similar aspects to sensor fusion. Bayesian classifier and neural network are commonly used methods.
Context is any information that can be utilized to characterize the situation of the smart car or driver or
the driving environment. Contexts information is collected when monitoring the roadway, the car and the driver.
To implement a context-aware smart car, the context analysis must be considered. Before recognition, the
system embedded in smart car must analyze the nature of context. Context data is separated in three layers
according to the degree of abstraction and semantics; sensor layer, context atom layer and context situation
layer. Figure 1 shows the three layers of context data. Firstly, sensor layer is the source of context data.
Secondly, context atom layer servers as an abstraction between the physical world and semantic world. Lastly,
context situation layer provides description of complex facts with fusion of context atoms.

Context Situation Layer Situation Situation

Context Atom Layer Atom Atom Atom Atom

Sensor Layer Sensor Sensor Sensor Sensor

Figure 1: The three-layers of context data

V. PROPOSED SYSTEM FOR SMART CARS

Hierarchical Context Model – Context situation recognition is a reasoning process and should be real time. In
order for the system to make good predictions and trigger events, a good classifier for context atoms is
necessary. The classifier will analyse the context atoms in two phases; the offline training phase and online
recognition phase. Hybrid classifier such as KNN (K-nearest neighbor) classifier or Bayesian and neural
network, most currently use, can be implemented to achieve the context recognition.
The offline training phase is used to learn the statistical relationship between context atoms and
situations and hence to generate the pattern of every single situation. Also, this can be used as a log of past
context atoms. The following illustrates an example how the offline training phase fits in context awareness
system A person aged around 65, frequently uses his/her smart car for daily activities. The context-aware system
has already captured the driver‟s behaviours and the latter driving styles, classified them as offline training
context data. So, when the same driver enters the smart car next time, through sensors camera, the system will
identify the driver, accesses its training context data and adapt according to the user. Hence, the smart car adapts
itself to the driver‟s profile using the offline training context data.
The online recognition phase is used to recognize the current situation according to its pattern in the running
time of a smart car.

www.irjes.com 29 | Page
 Driving Safely in Smart Cars in a Smart Road Environment

Figure 2: Context situation recognition architecture

Figure 2, defined eight trigger situations in the scenario, including three driver situations, three car
situations and two road situations. Collisions-1, and -2 are risk assessments and mean the danger levels. Since
assessments apply different features and have different targets, three statistical models have been trained and use
the corresponding classifiers to estimate the three kinds of situations. At any given time, the classifiers will
recognize the current situation if the situation stays unchanged for a continuous period of time. Only those
transitions that last for a period of time will be recognized as the transitions that really happen, while the
temporary situations will be ignored and not stored in the training storage.
Context atoms, Ontology definition – Each sensor captures a specific type of context atom. For each type of
context atom, a meaningful name must be assigned for application to use the contexts. Ontology is used to
define the name to guarantee the semantic understanding and sharing in smart cars. Three ontologies are used
are shown in Figure 3. (a) Ontology for car contexts include power system (engine status, accelerograph, fuel,
etc), the security system (safety of the car and driver such as airbag, safe belt, anti-lock braking system,
navigation system, etc) and conform system (entertainment devices, windows, air conditioner, etc). (b) Ontology
for driver contexts refer about the driver‟s physiological conditions like heart beat, blood pressure, density of
carbon dioxide, eyes close or open. The information is used to evaluate the health and mental status of the driver
for determining whether he/she is fit for driving. (c) Ontology for road contexts. The environmental contexts are
related to physical environments include information about road surface conditions, traffic density, signal lamps,
network status, road signs, weather conditions, etc.

Figure 3: Three ontologies for context atoms

www.irjes.com 30 | Page
 Driving Safely in Smart Cars in a Smart Road Environment

Scenario 1, Smart Car Adapting to Driver’s Profile - John Smith needs to drive his two grandsons at
school every morning located 10kms from home. Being a heart patient, he usually drives 40-50 km/h. His daily
routine activities in his smart car are as follows; (a) Starts engine car at 7h30. (b) Listens „Beetles and Queen‟
playlist songs. (c) Drives 10kms, drop kids at school. (d) Drives back home. On a particular day, John Smith
seats on his car at 7h50. The weather condition is bad, raining heavily, and windy. Road slippery and due to bad
weather condition, traffic flows on road is slow.
When John Smith enters the smart car, the camera sensor immediately captures the person by image
processing techniques. The smart car identifies the user as the old driver and through heart beat detector; it
captures health information from the driver. The smart car accesses its storage to check the calendar schedules,
times and days, and driver‟s identity who is purposed to be using the car. The smart car matches the collected
data from sensors to the context information in the offline training context data sets. Now, the smart car has full
knowledge about its driver and adapts itself according to the driver‟s need; adjust seat, starting playlist song
„Beetles and Queen‟, monitors the driving speed, predicts the destination in advance and time required.
The smart car is embedded with comprehensive system that detect normal and abnormal driving
behaviour using a context-aware system to collect and analyze contextual information about the driver, the
vehicle‟s state and environmental changes and to perform reasoning about certain and uncertain context. The
smart car notices a lateness of 20mins. It contacts the nearby base stations via wireless technology provided by
VANETs, establishes a shortcut path to reach on time.
Thanks to road sensors, monitoring traffic density and flow, road conditions and weather conditions,
the smart car is aware of the currents situation and triggers required actions. Due to bad weather conditions, the
driver needs assistance for driving in comfort and safely. So, street lights turn on as car approaches, voice alerts
on dashboard to notify the driver on the road sign and traffic signals due to poor visibility. Also, the smart car
continuously monitors other vehicles in vicinity and pedestrians through Vehicle Ad-Hoc Network to avoid
collisions.
However, halfway the meter gauge fuel indicates low and if changing direction to gas station will result
in lateness. The smart car serves the driver to assist him in several ways, the smart car is adapting in every
situation the driver is facing. The smart car talks to the vicinity cars to check their destinations and finds another
car with same destination. It sends the request to that car, that latter replies on its user‟s acknowledgement. The
children get in the other car and grandpa, John Smith, now can go to gas station and the kids can reach school on
time.

Connectivity and information sharing for intelligent mobility – Adopting hybrid schemes for combining
vehicle-to-vehicle communication (VVC) and vehicle-to-infrastructure communication (VIC) into a single
protocol can become advantageous and allow fast migration from VVC to VIC connectivity depending on the
operation context (high/low density, high/low velocity). This paper proposes the use of seamless connectivity in
heterogeneous wireless network environments, and in particular adopts them in VANETs, where VVC and VIC
represent the main communication protocols. To reach the goal of seamless connectivity, two major problems
have to be solved. Firstly, the context-aware system has to be developed on a basis that provides real-time data
of the road environments, driver‟s behaviours and the car itself and from fusion sensors. Secondly, the driver
should have easy access to services offered in the network.
The idea for seamless connectivity in VANET is a Vertical Handover technique based on vehicle
velocity, where a vehicle switches from a Serving Network (SN) to a Candidate Network (CN) if its speed is
below a fixed threshold level.
Seamless connectivity provides the ability for a vehicle to remain connected as it roams around
different types of networks. Vertical Handover allows vehicles to stay connected as they connect from one
network to a different type of network technology. As the vehicles move in a different network, the smart
systems in the vehicles must do auto-configuration in order to change networks without the drivers‟
intervention. The smart car to achieve seamless wireless connectivity, desires continuous sessions of wireless
connectivity. This is achievable by making the smart cars compatibles with heterogeneous existing network
technologies. Bus, private cars and heavy vehicles should work in Wi-Fi, VANETs, Wi-MAX, and Bluetooth.
The smart vehicles should have multiple interfaces to support multiple technologies.
On the other hand, the problem of a seamless connectivity becomes more challenging in VANETs,
because vehicle move across overlapping heterogeneous wireless environments. The complexity grows as there
are no restrictions in ubiquitous networks as there in present infrastructure networks. Also, cars with high speed
become more difficult to maintain good connectivity across overlapping network technologies. Hence, the
solution; VANETs handovers should be performed on basis of specific criteria such as vehicle mobility pattern
with considerable velocity and locality information.

www.irjes.com 31 | Page
 Driving Safely in Smart Cars in a Smart Road Environment

Smart Car Prototype – To build a smart car prototype, the hardware infrastructure and software platform are
required. A software platform is developed to manage devices, support vehicle-to-vehicle communication and
vehicle-to-infrastructure communication and provide runtime environment for the context model and
applications.
Hardware Requirements – Intelligent devices, sensors and processors like electronic control units (ECU) need
to be embedded in the smart car. Table 1 shows the devices that can be used in the smart car prototype. Various
sensors were deployed in the smart car prototype as listed in Table 2.
For road context-sensors; the environment contexts are acquired by a Crossbow sensor board, which provides
sensing capabilities including road and weather condition and ultrasound sensors can be used to estimate the
distance between two adjacent cars.
The car contexts such as information about velocity, fuel consumption and wheel rotation can be
obtained from car CAN bus. Bluetooth GPS receiver determines the car location.
Finally, as for driver contexts, RFIDs (radio frequency identification) can be embedded in smart car to
recognize the identity of the driver and passengers. An alcohol sensor, detecting alcohol level of driver, can
prevent driver from taking the wheel if above the threshold level.
Cameras were installed to track the movement of eyes and head of the driver. Real-time face detecting
and eye tracking can be done by fusion of active sensors. We can still use the eye blink behaviours method and
image-based user recognition to compare the detected faces with registered users‟ face images, by (Pan et al.,
2007) and Fisherface algorithm (Belhumeur et al., 1997) respectively. A microphone sensor, to receive the vocal
command of driver; “music on”, “music off”, “weather update”, “parking availability” and other commands,
were offered to driver.

Table 1: Device list in the smart car prototype

Devices Types Purpose
Notebook PC SONY VAIO FJ Run the software platform and application.
PDA 400 MHz Intel, XScalPXA255 Conduct the cooperation with software platform in the smart car.
Virtual keyboard I-tech virtual laser keyboard Provide the passengers and driver with easier interaction.
Touch screen FA801-NP/C/T
Wireless router D-Link DIR-300 Provide users with information services in car.
Camera Logitech QuickCam Orbit Capture image of driver‟s mood and behaviours.
CAN analyser/ CANalsyt-II Necessary analyzing tool for messaging. CAN hub was used to
Connection CANHub-S5 combine several communication nodes in a car.

Table 2: Sensors list in the smart car prototype

Sensor Context
Road Weather Condition sensor Rainy, windy
Road Condition sensor Slippery
Ultrasound sensor Distance between 2 cars
Microphone Loudness
Traffic sensor Dense
Street lights sensors Off
Car Bluetooth GPS receiver Car-location,
Fuel level sensor Fuel consumption
Accelerometer Acceleration
Hall-effect sensor Wheel rotation, velocity
Driver Camera Head, gaze, identity
Microphone Vocal command
RFID Identity
Alcohol sensor Alcohol density

Software platform – A Context-aware platform for the smart car, as shown in Figure 4 is projected consisting
of four layers; network layer, broker layer, context layer infrastructure and services layer. (a) Network layer, the
smart car should support different network technologies. The smart vehicles can establish a serial-bus system to
communicate between vehicles‟ components (such as engine and the steering system). WLAN 802.11 a/b/g
networks support communication between vehicle-to-vehicle and vehicle-to-infrastructure. (b) Broker layer, as
the smart car enters a new environment, the sensor broker is responsible for discovery and registration of new
sensors added into the smart car. Sensors can transmit data via wireless network WLAN, serial port, Ethernet
and USB. The role of the broker is to assign a globally unique address or identity to a sensor, specify the
updating frequency and define the fashion for sensor to transmit data and for system to parse data. (c) Context
infrastructure has been implemented on the basis of a context toolkit (Daniel et al., 1999) and consists of three
parts. Firstly, the context wrapper transforms sensor data into semantic context atoms. Secondly, the context
reasoned trains and recognizes context situations. Thirdly, context storage is a repository for historical contexts
and provides the advances query services. (d) Service layer, smart cars goal is to create a safer, more efficient

www.irjes.com 32 | Page
 Driving Safely in Smart Cars in a Smart Road Environment

and more convenient driving environment for drivers, so specific services should be developed. Most services
such as slow down when the distance the front adjacent car is less than safety limits, sending signals for street
lights to turn on and off. Also, online services allows driver to get notification from social websites, check
messages which are all output to user in voice messages. Intelligent dashboard can display road conditions and
assist the driver in location guidance. The ECU role is t parse the message and sends control signals to a relay,
which will control the actuator to change its state. Moreover, other intelligent services that can be implemented
in the smart car such as self-driving lessons. Smart car with the software platform should be in full capacity to
give self-driving lessons to new learners. Roads indications, important information can be displayed via
dashboard and by voice alerts. To motivate the building of our smart car prototype, an example is shown in
Figure 5.

Figure 4: Software platform for smart vehicles

Figure 5: Smart car prototype with built-in sensors and software interface.

www.irjes.com 33 | Page
 Driving Safely in Smart Cars in a Smart Road Environment

VI. EVALUATION PERFORMANCE

In order to assess the performance of the context-aware model and real-time implementation, the
architecture of the smart car is evaluated. The evaluation is done on the risk assessment by considering the
driver monitoring system. A smart car can monitor the driving environment, assess the possible risks and trigger
appropriate actions to avoid or to minimize the risk. The smart car architecture in Table 3 shows the risks
assessments.
Traffic monitoring system, various sensors can be used to identify the distance between the car and the
other road users. Active environment sensing in-and-out-car will be a general capability in the near future (Tang
et al., 2006). The radar sensor gives feedback about the position and velocity of the vicinity car. Multiple
cameras embedded in the road environments and cars eliminate blind spots, recognize obstacles and record the
surroundings. Also, via wireless technologies such as VANETs, the smart car can get information from the
internet or nearby cars.
Driver monitoring system, almost 95% of the accidents are due to human factors and in most cases
human behaviour is solely to blame. Smart cars present promising potentials to assist drivers in improving their
situational awareness and reducing errors. Physiological sensors can detect whether the driver is fit for driving.
With in-built cameras monitoring, the driver‟s mood and activity is known to the smart car and just the latter
adjust itself to the needs and activities of the driver.
Car monitoring system, context data like the dynamic of a car from engine, the brake system can be
read and transferred by controller area networks (CAN) to analyze where functionality of car.
Risk assessment verifies the risk of the driving task according to the situation of the traffic, driver and
car. Different levels of risks will lead to different responses, including notifying the driver through Human
Machine Interface (HMI), intelligent dashboard and triggering actions by car actuators.
HMI is a simple output in terms of voice alerts or dashboard‟s display that warns the driver of the
potential risks. For example, driver‟s behaviors are sleepy and cameras monitoring driver‟s eyes identifies eyes
closed. So, the tired driver is awakened by an acoustic alarm or seat vibrating.
Actuators, have the tasks to run specified control on the car without the driver‟s commands. The smart
car will adopt active measures in extreme cases where the reaction time of the driver has exceeded the safety
threshold. The active measures include stopping the car in case driver enable to react on time, or popping up
airbags if the system senses the driver is in danger of collision.

Table 3: Smart car risk assessment

Risk Assessment
HMI Driver Warning Seat vibrator, Hash noise, simple message
Car Actuator Passive protection, slow down, stop car, turn right/left.
Driver Monitoring Head camera, face camera, body optical camera, alcohol sensor
Car Monitoring Steering angle sensor, speed sensor, brake pressure sensor
Traffic Monitoring GPS, laser sensor, forward/side/rear camera

VII. EXPERIMENTATION - EVALUATION OF ACCURACY OF CONTEXT

SITUATION RECOGNITION
This experiment was carried out with 13 trigger situation of the smart car, including five driver
situations (Awake, abstracted, drunk, sick, drowsy), five car situations (Lack fuel, turn right, turn left,
accelerate, slow down) and three road situation (collision-1, collision-2, collision-3), means the levels of danger
of collision. As a matter of fact, the driver situation is the most complex one since it deals with face recognition
imaging processing. Hence, the driver recognition is taken as worst-case evaluation. From the Figure 6, we can
clearly see the increasing trend of accuracy as features (context information) grows. Eventually appropriate
contexts information will reduce the error probability. Another way to reduce the error rate is to introduce new
and independent context information (features) providing additional information. Even though increasing the
number of features will increase the computational cost of both the feature extraction and classification, it is
often reasonable to believe that performance will improve.

www.irjes.com 34 | Page
 Driving Safely in Smart Cars in a Smart Road Environment

Figure 6: Average accuracy of context situation recognition

VIII. CONCLUSION
Smart cars are a promising domain of ubiquitous computing, and have been subjected to
comprehensive researches and greater attention. This paper attempts to build a smart car from the view of
context-awareness. Our contributions are threefold: (a) modeling of context-aware the architecture of smart car.
(b) A three-layered context model is proposed to represent a complex driving environment. (c) The
implementation of a smart car prototype including its hardware requirements and software platform, to service
the context awareness system.
Our future work includes application of heuristic approaches on the classifiers and more innovative sensing
technologies to detect physiological behaviors of the driver. More focus on the driver prediction behavior will
be made.

REFERENCES
[1]. Moite, S, 1992, How Smart Can a Car Be. Proc. Intell. Vehicles Symp., p.277-279.
[2]. Wang, F.Y., 2006. Driving into the future with ITS. IEEE Intell. Syst., 21(3): 94-95. Wang, F.Y., Zeng, D., Yang, L.Q., 2006.
Smart cars on smart roads: an IEEE intelligent transportation systems society update, IEEE Perv. Comput., 5(4):68-69.
[3]. Kücükay, F., Bergholz, J., 2004. Driver Assistant Systems. Int. Conf. on Automotive Technologies. Istanbul, Turkey.
[4]. Toyota Testing Smart Cars, http://www.networkworld.com/community/blog/toyota-testing-smart-cartalk-each-other-and-road
[5]. Innovation, Intelligent Transport System, http://www.toyotaglobal.com/innovation/intelligent_transport_systems/infrastructure
[6]. Hoch, S., Althoff, F., Rigoll, G., 2007. The Connected Drive Context Server-flexible Software Architecture for a Context Aware
Vehicle. Advanced Microsystems for Autmotive Applications, Springer, Berlin, Heidelberg.
[7]. Benz, 2007. Mercedes-Benz to Present Driver Assistance Systems of Tomorrow at the ITS World Congress 2008. Available from:
http://e-carzone.com/news/auto_5911.shtml
[8]. Lexus, 2007. Lexus Advanced Active Safety Technologies. Available from: http:// www.lexushelp.com/
models/2008/LSh/Description.htm.
[9]. Available from: http://www.thedetroitbureau.com/2013/10/new-toyota-ford-systems-car-steer-clear-of-pedestrians/
[10]. Vehicular Ad Hoc Network. Available from: http://en.wikipedia.org/wiki/Vehicular_ad_hoc_network
[11]. Ozan, K.Tonguz, Gianluigi Ferrari (May 2006), John Wiley & Sons., ed. Ad Hoc Wireless Networks: A Communication-Theoretic
Perspective. Intelligent ad hoc vehicular network papers (Overview): Arunkumar Thangavelu, Sivanandam S.N (February 2007).
“Location Identification and Vehicular Tracking for Vehicular Ad Hoc Wireless Networks”. IEEE Explorer 1 (2): 112-116.
[12]. Vehicular Technology, IEEE Transactions on (Volume: 62, Issue: 9), Nov 2013, IEEE Vehicular Technology Society.
[13]. Fancher, P.S., et al., A Field Operational Test of Adaptive Cruise Control, SAE Paper 980852 2. .kegaya, M. et. Al., Development
of a Lane Following Assistant System, AVEC 98, 9837102
[14]. Oliver, N., Pentland, A.P., 2000. Graphical Models for Driver Behavior Recognition in a Smart Car. Proc. IEEE Intell. Vehicles
Symp., p.7-12.
[15]. Miyajima, C., Nishiwaki, Y., Ozawa, K., Wakita, T., Itou, K., Takeda, K., Itakura, F., 2007. Driver modeling based on driving
behavior and its evaluation in driver identification. Proc. IEEE, 95(2): 427-437.
[16]. Kumagai, T., Akamatsu, M., 2006 Prediction of human driving behavior using dynamic bayesian networks. IEICE-Trans. Inf. Syst.
E89-D(2): 857-860.
[17]. M. Sakairi and M. Togami, “Use of water cluster detector for preventing drunk and drowsy driving”, in Proc. IEEE Sensors, Kona,
HI, Nov. 2010, pp. 141-144.
[18]. D. Sandberg and M. Wahde, “Particle swarm optimization of feedforward neural networks for the detection of drowsy driving”. In
Proc. IEEE IJCNN, Hong Kong, China, Jun. 2008, pp. 788-793
[19]. H. Ueno, M. Kaneda, and M. Tsukino, “Development of drowsiness detection systems,” in Proc. IEEE Veh. Navig. Info. Sys.
Conf., Yokohama, Japan, Aug, 1994, pp. 15-20.

www.irjes.com 35 | Page
 Driving Safely in Smart Cars in a Smart Road Environment
[20]. Kutila, M.H., Jokela, M., Kinen, T.M., Viitanen, J., Markkula. G., Victor, T.W., 2007. Driver cognitive distraction detection:
feature estimation and implementation. Proc. IMechE J. Autom. Eng. 221(9): 1027-1040.
[21]. Bellotti, F., de Gloria, A., Montanari, R., Dosio, N., Morreale, D., 2005. COMUNICAR: designing a multimedia, con text-aware
human-machine interface for cars. Cogn. Technol. Work, 7(1):36-45.
[22]. Available from http://online.wsj.com/news/articles/SB123447510631779255
[23]. Available from http://www.virginiador.org/projects/constsal-smartrdoverview.asp

Keshav Luchoomun" Driving Safely in Smart Cars in a Smart Road Environment" International
Refereed Journal of Engineering and Science (IRJES), vol. 08, no. 02, 2019, pp 26-36

www.irjes.com 36 | Page