Intelligent Agents for Supporting Driving Tasks: An Ontology-based

Alarms System

V. Zamora, O. Sipele, A. Ledezma and A. Sanchis

Carlos III University of Madrid, Avda. De la Universidad 30, 28911 Legan

es, Madrid, Spain

Keywords:

Ontology, Advanced Driver Assistance System, ADAS, Trafﬁc Scenarios, Multi-agent Systems, Alarms

Systems.

Abstract:

This paper presents a rule-based alarm system as part of an ADAS. This work is developed by using a multi-

agent framework, and it focuses on the driving safety, in particular, in urban environments. The main point of

the proposed system is that it takes decisions based on the fusion of the information from the driver, the vehicle

status and the state of the road ahead, and it is designed to alert the driver of the car (without taking control

of it) only when the system considers that it is necessary. Five dangerous scenarios are deﬁned, analysed and

studied, and a repository of rules is designed to help the driver in that situations. In order to represent the

concepts and its relation about the urban trafﬁc environment, the system uses an OWL Ontology based on a

previous research and extended in this work.

1 INTRODUCTION

Road trafﬁc safety is a subject that is of a great impor-

tance worldwide. Public administrations, especially

European Commission, are concerned about it. The

percentage of people who die on the road each year

has been growing in recent years. According to the

Global status report on road safety 2015 by World He-

alth Organisation (WHO, 2015), road trafﬁc injuries

claim more than 1.2 million lives each year, which

causes a negative social impact, and a negative eco-

nomic impact too. In addition, it is known that about

78% of crashes are due to driver distractions.

That is why both industrial and academic commu-

nities are interested in Advanced Driver Assistance

Systems (ADAS). ADAS are systems developed in

order to help the driver in the driving process, increa-

sing car safety and, more generally, road safety. No-

wadays, more and more vehicles incorporate ADAS,

not only in high-end cars, but also mid-range cars and

even few low-end cars.

These systems use information about the car,

and also contain information about the environment

and/or the driver himself. After a process of reasoning

with obtained data, an ADAS produces a response in

order to avoid or face, for example, dangerous trafﬁc

situations or driver distractions. ADAS responses can

be as a visual alarm, sound alarm, or even taking con-

trol over the car (e.g. turning the steering wheel, bra-

king, etc). However, researches about system warning

design show that an incorrect design can inﬂuence ne-

gatively on the driver’s behaviour, on one hand, incre-

asing his/her workload therefore decreasing his/her

situation awareness (Vahidi and Eskandarian, 2003)

and, on the other hand, ignoring the warning produ-

ced by the system (Lee et al., 2004).

In this work, a multi-agent system is developed

to integrate the information provided by a driving si-

mulator and to use this information in order to help

the driver on his driving task in an urban environment

using an ADAS. This support is given by triggering

and showing visual and sound alarms (depending on

the type of launched alarm) when dangerous trafﬁc

situations are occurred.

After this introduction, this paper is organised as

follows. The next section provides an overview of

the background and related work of ADAS and alert

systems. Section 3 describes the conceptual frame-

work in which this research is set. Section 4 details

the data modeling and the deﬁnition of trafﬁc scena-

rios, rules and HCI messages. Section 5 explains the

experimentation process, including materials, testing

process and results. And ﬁnally, Section 6 presents

the conclusions and suggests some future works gui-

delines.

Zamora, V., Sipele, O., Ledezma, A. and Sanchis, A.

Intelligent Agents for Supporting Driving Tasks: An Ontology-based Alarms System.

DOI: 10.5220/0006247601650172

In Proceedings of the 3rd International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2017), pages 165-172

ISBN: 978-989-758-242-4

165

2 STATE OF THE ART

As mentioned in Section 1, road trafﬁc safety is a seri-

ous area because of its social and economical impact.

For years, the number of investments and investiga-

tions oriented to this topic has grown both privately

(by automotive companies) and academically. One

recent example of cutting-edge private technology is

IntelliSafe in Volvo. It consists in a number of ADAS

that are incorporated in Volvo cars in order to enhance

their safety such as the world-ﬁrst intersection bra-

king technology or the bird’s-eye view of the car and

its surroundings that allows driver to see obstacles all

around him (Volvo, 2016).

However, industrial community usually do not

communicate their research results apart from adver-

tises of released products or not speciﬁc enough re-

ports. On the contrary, academic community has

been very focused on this topic, aware of affordable

technology for mass-market vehicles.

Examples of ADAS that are already incorporated

in current vehicles are Lane Keeping Assist (LKA)

(National-Safety-Council, 2016), Autonomous Emer-

gency Braking (AEB) (Euro-NCAP, 2016) or Maxi-

mum Speed Permitted Assistant based on Automatic

Signal Detection (Tu and Fuh, 2016).

A recent publication about ADAS is a real time

driver distraction alert system for highway driving

for trucks where, taking account of the controls of

the vehicle, the system is capable to identify dis-

tractions on the driver and to notify him using alerts

(Dababneh, 2016). More recent publications are a

smartphone-based warning system in area of a con-

struction zone that notiﬁes drivers about upcoming

conditions within a construction zone (Qiao et al.,

2016) and a driver intention algorithm for pedestrian

protection and AEB systems (Diederichs et al., 2015).

The latter is an algorithm developed to detect driver’s

intention before initiating such transitions to automa-

ted driving as AEB in order to avoid annoyance, and

it is based on eye tracking of the driver and pedal acti-

vity of the vehicle. These systems are not included in

vehicles yet, but they represent new opportunities for

the future vehicles.

The research presented in this paper is based on

(Gutierrez et al., 2014). This paper presents an agent-

based system as part of the development of an ADAS

focused on urban driving safety. This system takes de-

cisions using information from the driver, the vehicle

status and the state of the environment. To represent

the concepts and its relation about the urban trafﬁc en-

vironment, an OWL Ontology is developed (Feld and

Mller, 2011).

Most of new researches and discoveries require

the reproduction of dangerous, extreme or even unu-

sual or impossible driving situations. Play this situ-

ations in real life would be sometimes unhealthy and

unethical, so that is why most of them use a driving si-

mulator (i.e. a truck driving simulator in (Dababneh,

2016)). This allows making any experiment, to recre-

ate any situation and to obtain all possible data of it.

Nowadays, there is a hard inter-brand competition

in ADAS and road safety in the car sector. Thus, most

of published research reports are too unspeciﬁc or un-

clear, so there is limited information about driving si-

mulators employed. Each brand usually has its own

driving simulator and doesn’t share much information

about it.

Examples that have been obtained of driving si-

mulators able to researchers are Carnetsoft research

driving simulator (Carnetsoft, 2016), VS500M car

driving simulator for educational and research activi-

ties (Simulation, 2016) or ST Software driving simu-

lator for research (Software, 2007).

In relation with these work, the realised works in

(Olmeda et al., 2013) (Pel

aez et al., 2012) (Musleh

et al., 2012) are focused on the development of ADAS

modules with the integration of new physic sensors on

vehicles, such as laser technology or cameras, and the

development of analysis methods that take the signal

produced by these sensors and obtain high level infor-

mation about the driving environment.

3 CONCEPTUAL FRAMEWORK

As it has been exposed before, there is a growing

trend of incorporating each time more ADAS with

safety features. Consequentially, SAE International

(Society of Automotive Engineers) deﬁnes a scale for

identifying an ADAS according to the automatisation

level that offers while a driving task is been perfor-

med (SAE, 2014). Focusing our attention on ADAS

without automatisation of driving controls and whose

purpose is only to warn about potentially risky situ-

ations, they don’t take account whether the driver is

aware about the risk or not. Therefore, to increase

this kind of systems on the instrumentation of the car,

keeping the independence between themselves and ig-

noring the driver’s perception could become counter-

productive, because the driver may be saturated with

a lot of irrelevant warning messages. Thus, the direct

consequence of this fact is that these systems could

lose their effectiveness and all the meaning for which

they were designed.

We propose a system that acts as a human Co-

driver, providing to the driver of the car only the rele-

vant information for him/her on each moment. High-

VEHITS 2017 - 3rd International Conference on Vehicle Technology and Intelligent Transport Systems

166

level ADAS that manages all the other ADAS with sa-

fety features and identify whether the driver of the car

really needs the information provided through these

ADAS according to his/her attention level. All this

information will be uniﬁed under the same interface

and the interaction with the driver of the car will be

adapted to the environment requirements, where the

driver must pay attention on the driving task at most

of the time.

The Figure 1 shows an overview of the conceptual

framework for the development of the Intelligent Co-

Driver where are included the involved concepts for

designing this ADAS approach based on providing to

the driver relevant high-level information about dri-

ving task on real time.

Figure 1: Overall Conceptual Framework for development

intelligent Co-Driver.

The Intelligent Co-Driver development process

involves several disciplines and research ﬁelds such

as Electrical Engineering, Signal Theory, artiﬁcial in-

telligence including the computer vision and the pre-

dictive computational models.

How it can be seen in Figure 1, the Intelligent

Co-Driver development follows a continuous impro-

vement cycle composed by ﬁve activities basically.

At ﬁrst, this cycle starts by studying how an Intel-

ligent Co-Driver functionality can improve the road

safety and User Experience (uEx), analysing what are

the needs of a driver while he/she is performing a

driving task, studying its viability and evaluating its

acceptance level and the socio-economic impact that

would mean its implantation on the vehicles.

Secondly, it is performed the collecting informa-

tion activity that consists of the researching about

what information is required for the development of

Intelligent Co-Driver and what are technological re-

sources that provide that information.

Next, on the knowledge representation activity it

is studied the best way of modeling all the informa-

tion obtained on the previous activity through the on-

tology. The main goal is the standardisation of a data

model that should consider all the possible aspects in-

volved on a driving task.

Then, it is conducted the extracting of useful in-

formation activity that consists on a research process

that takes the data model as input for obtaining cogni-

tive model for a speciﬁc Co-Driver functionality that

provides relevant information for the driver.

After, the driver interaction activity studies what

is the better way for transmit this useful information

to the driver regardless of age, sex and his/her level of

familiarisation with the new technologies.

Finally, the Co-Driver is evaluated in terms of usa-

bility, reliability and performance. At the end of this

activity, the cycle comes back to its ﬁrst activity.

Each activity is feedback for its next activity,

going back if it will be necessary in order to accom-

plish the established requirements at beginning of the

development process.

4 SYSTEM DESCRIPTION

In this section, it is described the alarm system ba-

sed on rules that has been developed, by detailing the

data modeling and the trafﬁc scenarios, rules and HCI

messages that have been deﬁned.

4.1 Data Modeling

The alarm system works in the driving simulator en-

vironment, and it has to read, interpret and understand

all the information that is in there. That is why all the

data has to be modeled, structured and integrated.

Data modeling is done by designing an OWL/RDF

Ontology, which provides a formal deﬁnition that gi-

ves semantic structure to data. It represents the con-

cepts and its relation about the urban trafﬁc environ-

ment, and this concepts contain information that co-

mes from the driver, the elements outside the vehicle,

and from the vehicle itself. Therefore, each instance

of the ontology is a trafﬁc situation that is going to be

analysed, and making queries to the ontology allows

getting this instances.

As mentioned in Section 2, this research is based

on (Gutierrez et al., 2014). In that paper, it is designed

and described an OWL/RDF Ontology that represents

the starting point for data modeling in this work. It

can be observed that concepts of car, driver, car con-

text, pedestrian and pedestrian crossing the street are

deﬁned, and each one of these concepts has its own at-

tributes. Right after, it is described the discrete values

that have been used in this work for each attribute.

• Car:

– Distance to car: far away, far, normal, close,

very close.

– State (of the vehicle): moving, stopped.

• Driver.

Intelligent Agents for Supporting Driving Tasks: An Ontology-based Alarms System

167

– Area of vision: front, front-right, front-left,

left, right, behind-left, behind-right, interior

rearview mirror, left exterior rearview mirror,

right exterior rearview mirror, speedometer, ra-

dio / air conditioning / down, roof / up.

– Attentive: yes, no.

– Horizontal/vertical direction of the eyes:

centre, left/up, right/down.

• CarContext.

– Existence of car in front/left/right: yes, no.

– Existence of pedestrian/pedestrian crossing:

yes, no.

• Pedestrian.

– Angle (relative to driver): front, front-left,

front-right.

– Distance: far away, far, normal, close, very

close.

– Trajectory: N, S, E, W, NW, NE, SE, SW.

• PedCrossing.

– Distance to pedestrian: far away, far, normal,

close, very close.

As mentioned before in this section, this ontology

represents a starting point. It represents basic con-

cepts, but it is incomplete. For this actual research,

the ontology is extended in order to encompass a wi-

der range of information and, thus, be able to address

more complex danger situations with the rule-based

decision-making system developed.

It is added to the car context the existence of

cars near the considered, in more positions: behind,

behind-left, behind-right, front-left and front-right.

Besides, it is included a new concept: the concept of

the driver’s vehicle (MyCar). This concept has the

following attributes:

• MyCar.

– Brake/Clutch/Throttle: full, none, medium.

– Gear: -1, 1, 2, 3, 4, 5, 6.

– Lat/long accel.: high, medium, low, null.

– Revolutions per minute: [0, maxRev].

– Speed: high, medium, low, null. (It is deﬁned

medium speed as approximate values relative to

maximum track speed).

– Steering wheel angle: left, centre-left, centre,

centre-right, right.

The proposed ontology diagram is shown in Fi-

gure 2.

4.2 Deﬁning Trafﬁc Scenarios

This research is oriented to urban driving. In addition

to the fact that most of researches are based on road

driving (Dababneh, 2016), in urban driving there are

many potential dangerous situations.

In this work, ﬁve dangerous trafﬁc scenarios

are deﬁned and analysed. To identify and deﬁne

these scenarios, an online survey was conducted

to drivers with different proﬁles (age, sex, expe-

rience, among others) about dangerous situations of

urban driving and the acceptance of an ADAS in

such cases (surveys and results can be found on

http://www.caos.inf.uc3m.es/adas-driver-modeling/).

Consequently, a previous work about identifying

risky driving situations and the survey results, the fol-

lowing ﬁve driving scenarios were deﬁned, conside-

ring them as representative for this ﬁrst approach. The

concepts that are included in the proposed ontology

are used in order to deﬁne them.

• Scenario 1: Risk of frontal collision. The dri-

ver is distracted on a road where there is medium-

intense trafﬁc, and the distance to the car that pre-

cedes the driver is reduced, becoming very short.

• Scenario 2: Risk of running over. While the dri-

ver is circulating through an urban environment, a

pedestrian crosses the road at a distance relatively

close to the vehicle.

• Scenario 3: Risk of rear collision. An overta-

king is going to take place, but the car behind the

driver’s vehicle is overtaking or initiates the pas-

sing too.

• Scenario 4: Risk of lateral collision. While

the driver is circulating through an urban envi-

ronment, a stopped car starts and initiates the mo-

vement in a relatively short distance.

• Scenario 5: Pedestrian not visualised. While

the driver is circulating through an urban environ-

ment, a pedestrian crosses the road from behind

a parked vehicle or there is an object that makes

the pedestrian not visible to the driver. The diffe-

rence between scenario 2 and this scenario is that

now the pedestrian is not always visible, so now

the situation is more critical.

4.3 Deﬁning Rules

Once data have been integrated, modeled and structu-

red, and the scenarios have been designed, the rules

can be deﬁned. Speciﬁcally, seven rules are deﬁned

to identify the ﬁve dangerous scenarios established,

with the objective of helping and warning the driver

in order to avoid road accidents.

VEHITS 2017 - 3rd International Conference on Vehicle Technology and Intelligent Transport Systems

168

Figure 2: Proposed ontology diagram.

An expert system design process was carried out

for the rules deﬁnition. This process required gather-

ing information from several sources such as (1) con-

ducting interviews with driving instructors, (2) ana-

lysing the Spanish driving regulation and driving ma-

nual (DGT, 2017) that deﬁne how the driver must deal

with each driving situation and (3) obtaining the dri-

vers’ opinion with the conducted surveys.

Each scenario has one or several rules that detect

the potential danger and launch an alarm to the driver.

It is shown each rule/set of rules for each scenario

below. It is described the situation that makes the rule

to be activated, and, thus, the rule produces an output.

• Scenario 1: Risk of frontal collision.

– There is a car, close or very close, in front of

the vehicle.

– The driver is not attentive.

– The speed of the vehicle is medium or high.

• Scenario 2: Risk of running over (two rules, one

for pedestrians that cross to left and one for pede-

strians that cross to right).

– There is a pedestrian, close or very close, cros-

sing the street.

– The driver is not attentive.

– The speed of the vehicle is medium or high.

• Scenario 3: Risk of rear collision.

– There is a car, close or very close, in front of

the vehicle.

– This car is going to be overtaken by the vehicle.

– There is a car in movement behind the vehicle,

close or very close, that could want to pass the

car too.

• Scenario 4: Risk of lateral collision.

– There is a car parked on the right side of the

road that may start and join the road.

– This parked car is very close, close or medium

distance.

– The driver is not attentive.

• Scenario 5: Pedestrian not visualised (two rules,

one for pedestrians on the left side and one for

pedestrians on the right side).

– There is a parked car ahead of the vehicle.

– There is a pedestrian behind of the parked car.

– The trajectory described by the pedestrian’s

movement indicates he/she is going to cross the

road.

Since the system has only one output, a hierarchy

of alarms is needed and has been implemented. The

same process as the dangerous situations has been fol-

lowed in order to design the implemented hierarchy.

This hierarchy is shown below (Table 1).

Table 1: Alarm hierarchy (ordered from highest to lowest

priority).

Scenario Alarm Priority

Scenario 2 Risk of running over 1

Scenario 1 Risk of frontal collision 2

Scenario 5 Pedestrian not visualised 3

Scenario 4 Risk of lateral collision 4

Scenario 3 Risk of rear collision 5

Consequently, if two different alarms are laun-

ched, only the one with the highest priority will be

processed. The highest priority scenario is given by

the risk of running over, because human life is in dan-

ger. Next, there is the risk of frontal collision, follo-

wed by the hit of an unseen pedestrian. This order is

followed because a car cannot hit a pedestrian if the

vehicle has something in front that prevents the pas-

sage. Then, there is the risk of lateral collision and,

ﬁnally, the risk of rear collision, since an overtaking is

Intelligent Agents for Supporting Driving Tasks: An Ontology-based Alarms System

169

only going to be made if no other situation of danger

occurs.

4.4 Deﬁning HCI Messages

Now, the dangerous scenarios that have been deﬁned

are detected and alarms are generated by the rules.

These alarms are oriented to warn the driver, and are

designed to be clear but not annoying (an alarm is

launched only when it is truly necessary).

For each risk scenario rule/set of rules, there is

a different kind of visual alarm. Each alarm corre-

sponds with an image that is showed in the HCI inter-

face, as shown in Figure 3.

Figure 3: An alarm is showed in the HCI interface of the

driving simulator.

The images that correspond with the different

alarms for each scenario are shown below (Figure 4).

They are ordered from left to right (scenario 1 is the

ﬁrst image, scenario 2 is the second image, etc.).

Figure 4: Visual alarms that are produced by the rules.

In addition to the visual alarm, due to the level of

danger, in certain cases there is also a sound alarm.

These cases are the alarms associated with the scena-

rios 1 (risk of frontal collision), when the distance is

very short, and 5 (pedestrian not visualised). Again,

this sound alarm is designed to be clear, but not an-

noying to the driver.

5 EXPERIMENTATION

In this section, the process of experimentation is des-

cribed, by detailing the used material, the design of

the testing process and the obtained results.

5.1 Experimental Setup

For the performing of this work has been used several

resources that will been described as follows. The Fi-

gure 5 shows a simpliﬁed scheme about the testing

infrastructure used for this work where it can distin-

guish the interaction between three systems.

Figure 5: Testing infrastructure scheme.

Firstly, the Driving Simulator System with

the STISIM Drive Simulation Software (Sys-

tems Technology, 2013), driving devices and three

screens for 135 degrees vision that provide a more re-

alistic driving experience. This simulation software

includes several driving scenarios that reproduce real

driving situations and allows getting access and mana-

ging to all simulation parameters on real time through

a middleware layer. In addition, a Route Management

Module is developed in order to give the capacity to

the driving simulator of reproduce real routes for tes-

ting the system (Zamora et al., 2016). So, this system

provides the next aspects:

• Vehicle Dynamics. All the parameters related

with the driven car such as speed, gear, steering

wheel angle, pressure level of the pedals, among

others.

• Driving environment. The extraction of high-level

information about the driving environment is per-

formed by the Environment Agents described in

(Sipele et al., 2016).

• Interface Agent. It is a virtual Human-Computer

Interface (HCI) is deployed on the simulator

screen and it will raise the received alarms by the

Alarm System through visual and sound messages

to the driver’s car.

VEHITS 2017 - 3rd International Conference on Vehicle Technology and Intelligent Transport Systems

170

Secondly, the Monitoring System uses the Micro-

soft Kinect camera and a software implementation for

extracting useful information about the driver such as

the orientation of his/her head and a driver’s eye gaze

estimation.

Thirdly, the Alarm system has been implemented

using the Java Agent Development Environment Fra-

mework (JADE) (Italia, 2016) and its third-party Add-

on Web Service Integration Gateway (WSIG) (Bo-

ard, 2015). This implementation allows to establish

a communication system based on Service Oriented

Architecture (SOA). Thus, from the driving simula-

tion system is sent a Driving situation instance as a

web service request message to the Alarm System.

This request message is received by the WSIG Agent

where is processed and transferred to the implemen-

ted Alarm Agent. Finally, if the implemented Alarm

Agent produces an alarm, it will be communicated to

WSIG Agent that will send the alarm message to the

Interface Agent.

5.2 Testing Process Design

The objective of the testing process is to check if the

created repository of rules is able to detect the given

dangerous scenarios and if it is able to launch the cor-

responding alarm (visual or visual and sound).

The system is tested with black box tests in two

different ways. The ﬁrst way is directly in the de-

signed web environment, where each test is codiﬁed

in XML. Then, the XML is introduced to the system

as a request, and it generates a response indicating

whether a dangerous situation is detected or not, and

the corresponding alarm. These codiﬁed tests corre-

sponds with ontology instances, where each concept

and its attributes are described and take concrete va-

lues.

The second way of testing the system is perfor-

ming the tests on the simulator. While the driving task

is in process, the simulator creates instances of the on-

tology with values according with the actual driving

situation, and these instances are sent to the system

as a request. This way, when a dangerous scenario is

given, the corresponding visual (or visual and sound)

alarm is shown.

5.3 Results

As mentioned in Section 5.2, the system is veriﬁed

following two different processes: directly in the de-

signed web environment and using the driving simu-

lator. The ﬁrst of the described processes is used du-

ring the rule design phase, since it is possible to test

each rule separately. In this way, it could be checked

that all rules work correctly while they are being de-

signed, one by one. However, this testing process is

not enough, because the used instances are ”artiﬁcial”

(implemented manually).

The second of the described processes is used to

test the complete repository of rules. The progress of

the testing process is similar to driving a real car. Ex-

amples of alarms launched in the HCI of the driving

simulator are shown in Figure 3.

This testing process is more complete than the pre-

vious one, because the system is tested as if it were

implemented in a real-world environment by using the

driving simulator. It is checked that all the deﬁned

dangerous scenarios are detected and the correspon-

ding alarms are launched.

As far as we know, there are no other related

works about this speciﬁed research line focused on

establish a hierarchical alarm system based on urban

driving situation analysis that works as a co-driver,

only warning the driver when it is necessary. There-

fore, a comparison cannot be showed.

6 CONCLUSIONS AND FUTURE

WORK

In this paper, it is presented a rule-based alarm system

based on a multi-agent system previously proposed

(Gutierrez et al., 2014).

This work is developed to integrate the informa-

tion provided by a driving simulator and to use this in-

formation to help the driver on his driver task by trig-

gering and showing visual or visual and sound alarms

when dangerous trafﬁc situations happen. It is focu-

sed in urban environments, where there are many po-

tential dangerous situations caused, in most cases, by

distractions of the driver.

The entire project represents an approach to an

ADAS, so the ﬁnal objective is to be embed in real

vehicles and to use real data provided by real sensors.

Consequently, as a future work there is the extension

of the system in order to take account of more dan-

gerous situations, by obtaining more information and

by deﬁning more rules, and in order to complete the

existing scenarios.

One limitation of the system correspond with the

value of the attributes of the concepts, since they are

categorical. The system would be more precise if data

were numerical, so it is a future work to be accomplis-

hed.

Since the system only can give one output, it is

necessary to implement an alarm hierarchy. If the

number of scenarios is extended, this hierarchy would

be much more complex. That is why it is a future

Intelligent Agents for Supporting Driving Tasks: An Ontology-based Alarms System

171

work the implementation of this system using fuzzy

logic, which allows considering all the rules at the

same time and, this way, dispensing the need of the

hierarchy of alarms.

Finally, taking account the human factors invol-

ved on the driving warning systems, aspects such as

time reaction, situation awareness, divided attention,

among others, will be studied for improving of the

designed alarm system.

ACKNOWLEDGEMENTS

This work has been supported by the Spanish Mi-

nistry of Economy, Industry and Competitiveness,

(TRA2015-63708-R) and (TRA2016-78886-C3-1-R)

Projects.

REFERENCES

Board, J. (2015). JADE Web Services Integration Gateway

(WSIG) Guide.

Carnetsoft (2016). The carnetsoft driving simula-

tor. URL: http://www.carnetsoft.com/. (Last access:

20/11/2016).

Dababneh, L. F. (2016). Development and validation of

a driver distraction alert system for highway driving

using a truck driving simulator. In Electronic Theses

and Dissertations. UOIT.

DGT (2017). Normativa y legislaci

on. URL:

https://goo.gl/4gDo6C. (Last access: 17/02/2017).

Diederichs, F., Schttke, T., and Spath, D. (2015). Driver

intention algorithm for pedestrian protection and au-

tomated emergency braking systems. In 2015 IEEE

18th International Conference on Intelligent Trans-

portation Systems, pages 1049–1054.

Euro-NCAP (2016). Autonomous emergency braking.

URL: https://goo.gl/yj2rkJ. (Last access: 22/09/2016).

Feld, M. and Mller, C. (2011). The automotive ontology:

Managing knowledge inside the vehicle and sharing it

between cars. In Proceedings of the 3rd International

Conference on Automotive User Interfaces and Inte-

ractive Vehicular Applications, Automotive UI 2011,

pages 79–86.

Gutierrez, G., Iglesias, J. A., Ordo

nez, F. J., Ledezma, A.,

and Sanchis, A. (2014). Architecture and ontological

modelling for assisted driving and interaction. In In-

formation Fusion (FUSION), 2014 17th International

Conference on, pages 1–8. IEEE.

Italia, T. (2016). Jade. URL: http://jade.tilab.com. (Last

access: 22/11/2016).

Lee, J. D., Hoffman, J. D., and Hayes, E. (2004). Collision

Warning Design to Mitigate Driver Distraction. Chi,

6(1):65–72.

Musleh, B., de la Escalera, A., and Armingol, J. (2012).

Detecci

on de obst

aculos y espacios transitables en en-

tornos urbanos para sistemas de ayuda a la conducci

basados en algoritmos de visi

o est

ereo implementados

en gpu. Revista Iberoamericana de Autom

atica e In-

form

atica Industrial, 9:462–473.

National-Safety-Council (2016). Lane keeping assist. URL:

https://goo.gl/U2yZPu. (Last access: 22/09/2016).

Olmeda, D., Premebida, C., Nunes, U., Armingol, J., and

de la Escalera, A. (2013). Pedestrian detection in

far infrared images. Integrated Computer-Aided En-

gineering, 20:347–360.

Pel

aez, G., Romero, M., Armingol, J., de la Escalera,

A., Mu

noz, J., van Bijsterveld, W., and Bola

no,

J. A. (2012). Detection and classiﬁcation of road

signs for automatic inventory systems using compu-

ter vision. Integrated Computer-Aided Engineering,

19:285–298.

Qiao, F., Rahman, R., Li, Q., and Yu, L. (2016). Safe and

environment-friendly forward collision warning mes-

sages in the advance warning area of a construction

zone. In International Journal of Intelligent Trans-

portation Systems Research.

SAE (2014). Automated driving. sae international standard

j3016. URL: https://goo.gl/MNBvoh. (Last access:

21/11/2016).

Simulation, V. (2016). Virage simulaion driving simulator

systems. URL: https://goo.gl/Qhz3nP. (Last access:

20/11/2016).

Sipele, O., Zamora, V., Ledezma, A., and Sanchis, A.

(2016). Testing adas though simulated driving situ-

ations analysis: environment conﬁguration. In Pro-

ceedings of the First Symposium SEGVAUTO-TRIES-

CM. Technologies for a Safe, Accessible and Sustaina-

ble Mobility. R&D+I in Automotive: RESULTS, pages

23–26.

Software, S. (2007). Driving simulators for research. URL:

http://www.stsoftware.nl/simulator.html. (Last access:

20/11/2016).

Systems Technology, I. (2013). Stisim drive: Car

driving simulator & simulation software. URL:

https://goo.gl/tOB8Zu. (Last access: 14/09/2016).

Tu, K.-H. and Fuh, C.-S. (2016). The speed-limit sign de-

tection and recognition system.

Vahidi, a. and Eskandarian, a. (2003). Research advances

in intelligent collision avoidance and adaptive cruise

control. IEEE Transactions on Intelligent Transporta-

tion Systems, 4(3):143–153.

Volvo, C. C. (2016). Intellisafe. URL: https://goo.gl/zIqZIf.

(Last access: 20/09/2016).

WHO (2015). Global status report on road safety.

In Violence and Injury Prevention. WHO Library

Cataloguing-in-Publication Data.

Zamora, V., Sipele, O., Ledezma, A., and Sanchis, A.

(2016). Integrating real driving routes in a simula-

ted environment. In Proceedings of the First Sym-

posium SEGVAUTO-TRIES-CM. Technologies for a

Safe, Accessible and Sustainable Mobility. R&D+I in

Automotive: RESULTS, pages 27–30.

VEHITS 2017 - 3rd International Conference on Vehicle Technology and Intelligent Transport Systems

172