LOGGING, ALERT & EMERGENCY SYSTEM FOR ROAD
TRANSPORT VEHICLES
An Experimental eCall, Black-box and Driver Alerting System
Javier Fernández, Fernando Cantalapiedra, Mario Mata, Veronica Egido, Sergio Bemposta
Computer Architecture and automation dept., Universidad Europea de Madrid, Villaviciosa de Odón, Madrid, Spain
Keywords: Driver assistance, eCall, digital black-box, sign detection, alert system.
Abstract: This paper describes the experimental platform developed at UEM, mounted on a conventional vehicle. It
monitors most of the driver’s actions on the controls of the vehicle, logs the vehicle speed and position
using a GPS, detects and recognizes vertical traffic signs, and records the last seconds of the trip with a
panoramic video camera. If an accident occurs, the system calls emergency services (112 in Spain) sending
vehicle position information (via SMS) and opening a voice channel.
1 INTRODUCTION
The alert and control system developed in SACAT
project (Control and Alert in case of Accident
System for road Transport vehicles) is intended to be
used in freight and also in human transport by road.
It aims to increase security, specially trying to save
human lives by decreasing the time until assistance
arrives.
EU White Book on European Transport Policy for
2010 reveals worrying data. Taking into account all
costs, road transport represents approximately 1
billion Euros (above 10% EU gross domestic
product). Road transport carries 44% of total freight
and 79% of total people transportation. Between
1995 and 2003, road freight transport increased
19.4%, while rail freight transport decreased 43.5%
in the same period. However, road transport is the
most dangerous one, having a terrible cost of human
lives, and provoking expensive traffic jams (fig. 1).
EU proposes a two-lined action: agreement in
penalties, and the promotion of new technologies in
road security. It explicitly establishes that
technological evolution will also allow strengthening
the usual control and penalty means, using automatic
systems and driving-assistance on-board devices.
Related to this, EU proposes the use of black-boxes
in vehicles, for registering relevant parameters to
analyze the causes of an accident (like in other
transportation means).
Figure 1: Multiple accident and closed road.
In the same line, EU has adopted a directive, named
eCall (in the E_MERGE program) that will make
compulsory the installation of an alerting device in
every new vehicle marketed in the EU from 2009
on. This device should automatically alert
emergency services in case of accident using mobile
phones and GPS (EC,2003).
Another EU measure aimed to improve security in
freight vehicles is the obligation of installing digital
tachometers from August 2006 on (in Spain this
date has been moved to august 2005 for vehicles
above 3,5 tons or 9 passengers).
384
Fernández J., Cantalapiedra F., Mata M., Egido V. and Bemposta S. (2006).
LOGGING, ALERT & EMERGENCY SYSTEM FOR ROAD TRANSPORT VEHICLES - An Experimental eCall, Black-box and Driver Alerting System.
In Proceedings of the Third International Conference on Informatics in Control, Automation and Robotics, pages 384-389
DOI: 10.5220/0001204303840389
Copyright
c
SciTePress
The system described in this paper follows all these
lines of action in security matters, and puts them
together in a single prototype. Furthermore, it adds
some functionalities that will surely be included in
security directives in the future, such as automatic
recognition of traffic signs (Escalera,2003) and real-
time alerting to the driver. The whole system tries to
lower the risk of accident; and if it finally happens,
tries to minimize the time until victims are assisted,
and provides critical information for analyzing the
reasons.
The developed prototype is composed of two main
modules, with the following functionalities:
Black-box. Records the detected signs, vehicle data
(speed, position, driver actions) and a panoramic
video of the last minutes of the trip.
Alert System. If an accident takes place, the system
sends a text message (SMS) to emergency services
(112 in Spain). It includes the location of the vehicle
(GPS), kind of vehicle and load data (number of
passengers, dangerous loads, etc). This is an
essential matter, because every minute counts when
a person is wounded in a traffic accident. Recent
studies established that the probability of an accident
victim of surviving is doubled if he/she is assisted in
the first 20 minutes after the crash.
Additionally, a bidirectional voice channel is opened
in an automatic manner, allowing direct
communication between 112 emergency services
and the passengers. It can also be manually launched
with an emergency button, if a passenger gets ill, or
the accident of a nearby vehicle is seem; the vehicle
position is also sent to the emergency service, so
there is no need to spend precious time trying to tell
where the incident is.
In the next sections, the system architecture, the two
main modules and the sign recognition subsystem
are detailed. Finally, in section 6, the present system
performance is described, and the opened lines of
work and improvement are discussed.
2 SYSTEM ARCHITECTURE
The system has been implemented on a conventional
vehicle Renault Express (figure 2). This model lacks
of communication bus, so all the instrumentation
and wiring associated to driving devices have been
added. The whole system uses the vehicle’s 12V
battery by means of a 600W dc/ac inverter, sourcing
220 AC volts. This allows using standard
equipments.
Figure 2: Test vehicle from SACAT.
The core of the application is formed by a
conventional PC and a PDA. The sensorial system
includes the following devices:
Bluetooth GPS module
Mobile phone with Bluetooth interface
Panoramic color CCD camera
Conventional color CCD camera
Encoder reading steering wheel movement
All/nothing sensors for pedals, direction
lights, hand brake and contact
The PC hosts the main system interface, feeds the
black-box (external USB drive), captures panoramic
and frontal video, and registers the data from the
driver’s action sensors. It also processes the frontal
camera images for detecting vertical traffic signs.
GPS
PDA
GPRS
PC
Color camera
Panoramic
camera
Vehicle
sensors
Figure 3: SACAT architecture.
LOGGING, ALERT & EMERGENCY SYSTEM FOR ROAD TRANSPORT VEHICLES - An Experimental eCall,
Black-box and Driver Alerting System
385
The GPS receiver is used for registering the speed
and coordinates of the vehicle; they are stored in the
black-box. The mobile phone allows communication
(SMS and voice channel) with emergency services
when a collision is detected (using the airbag firing
sensors) or when the driver uses the emergency
button, following eCall standards.
The PDA works as a wireless interface between the
GPS and mobile phone with the PC. It also is used
as input device for the system: data about the driver,
load, route, etc. is loaded in the central control of the
transport firm, and then plugged into the system.
Figure 4: Interface devices in SACAT.
3 LOGGING SUBSYSTEM
This subsystem is responsible of saving relevant
information from the system such as video, sensor
and GPS data, etc. This module is called “Black-
box” because of their similarity to the ones in
airplanes.
Panoramic video (fig. 5) is recorded by saving
periods of time in several consecutive files; the
oldest one is deleted each time a new file is saved, to
avoid running out of storage space in a long trip.
This way, the last images of the environment around
the vehicle before an accident are left available for
posterior analysis. This information facilitates
finding the causes of the accident (to avoid them in
the future), and clarifying responsibilities for it. The
video recording is done as a background process and
doesn’t interfere with the other processes of the
system.
Figure 5: Panoramic video.
Images are obtained using a 3D Sony CCD camera
mounted on the roof of the vehicle. When the
vehicle is stopped for a significant time during a trip
(to rest or have a break, for example), a separate
group of video files are saved, acting as a
surveillance device.
The sensor information is obtained from several
driving devices, such as pedals (clutch, brake, and
accelerator), steering wheel, indicators, hand brake,
and contact. The sensors are read using a data
acquisition card in the PC. The information gathered
is saved in the black-box’s database each time it
changes. A timestamp is added for each entry.
Position and Speed of the vehicle are obtained from
a bluetooth GPS device in the PDA (where a
navigation application can be run), and sent to the
PC via wireless. This information is also saved in
the black-box database.
When an accident occurs, the last GPS position (and
other relevant information, see next section) is sent
to the mobile phone, and then to emergency
services.
The aim of this black-box is to store a log of the trip,
and specially a log of the environment and the last
actions of the driver just before an accident.
4 ECALL SUBSYSTEM
In this section, the alerting module of the system is
described. It is designed to be adapted to the rules
established by the European Commission (EC), the
ICINCO 2006 - ROBOTICS AND AUTOMATION
386
European Association of Vehicle Manufacturers
(ACEA) and the multi-sector public/private-
partnership for the implementation of Intelligent
Transport Systems and Services (ERTICO). These
entities work for the introduction of an emergency
call system that will be automatically launched in
case of accident, named eCall. This system will
automatically alert emergency systems and provide
them with critical information about the accident.
eCall has therefore the potential to greatly reduce the
number of fatalities, the severity of injuries and the
stress in post-crash situations, by speeding up the
response of the emergency services and allowing to
choose the adequate material and human resources
needed for each accident.
This emergency call module incorporates this
functionality, and some others, according to the
configuration shown in figure 6.
Figure 6: Emergency call module.
Figure 6 shows the hardware elements involved in
the emergency module, and the steps given after an
accident occurs. Hardware architecture is composed
of a GPS receiver, the main PC, a PDA and a mobile
phone with Bluetooth connectivity.
If an accident occurs, the PC receives the airbag
signal, and then activates the emergency module.
This module sends an SMS via the mobile phone to
the corresponding PSAP (Public Safety Answering
Point), 112 in Spain. The message sent to the PSAP
should contain the so-called MDS (Minimum
DataSet). This minimum set of data consists of the
following information: “When”, “Where” and
“Who” (E-merge,2004).
This information is considered the minimum needed
for speeding up emergency services’ response with
the adequate resources. The message sent contains
the instant of the transmission ‘hh:mm:ss’ (“When”),
the latitude, longitude and movement direction of
the accident (“Where”), and information about the
driver and the load of the vehicle (“Who”). With the
standardization of the e-Call information, messages
will be adapted to the same model for all countries.
The system implemented in this project could be
easily adapted to send the so-called FDS (Full
DataSet) that would contain additional data
(enterprise information, insurance data, etc.). The
idea promoted in the European program is to send a
MDS to the PSAP and more detailed information to
a private service provider (PSP).
Considering the importance of the emergency
message, after sending it for the first time, the PC
sends a signal and the information to the PDA, that
will be in charge of two different tasks: sending
again the message via mobile phone, and
establishing a voice call to make possible to the
emergency services to talk with the vehicle
occupants if conscious. This call can also be
manually established.
5 SIGNAL RECOGNITION
SUBSYSTEM
Detection of vertical traffic signs is a classic
application for computer vision researchers, that still
remains unsolved (except in particular and
controlled situations). In the last decades, promising
results have been obtained thanks to new
computational techniques ((Bahlmann,2005),
(Escalera,2001), among others) but surprisingly very
few systems has been integrated in a real vehicle
with some success, as for example (Priese,1994) in
their vehicle VITA II. The system developed in our
project SACAT, although far from being
commercial, opens real and promising
experimentation and improvement ways.
A
u
t
o
m
a
t
i
c
c
a
l
l
Accident
MDS information is sent via
wireless
4
º
5
º
GPS
information
A
u
t
o
m
a
t
i
c
c
a
l
l
Accident
MDS information is sent via
wireless
4
º
5
º
GPS
information
1
s
t
2
n
d
3
r
d
4
th
5
th
LOGGING, ALERT & EMERGENCY SYSTEM FOR ROAD TRANSPORT VEHICLES - An Experimental eCall,
Black-box and Driver Alerting System
387
The traffic signs recognition module integrated in
the system tries to detect and then read the vertical
signs present along the road. This is done aiming at:
- Logging the signs in the Black-box’s
database.
- Alerting the driver in dangerous situations,
such as:
o Over speeding
o Overtaking prohibitions
o Crossroads, Stops, Give way signs
o Curves, slopes, road works
This module uses a color CCD camera with
adjustable zoom, mounted inside the vehicle and
aiming at the movement direction. Each image is
processed following the these steps:
Step 1: Sign detection.
First, objects candidate to be signs are detected and
isolated. A color segmentation, followed by a shape
analysis, are used for this purpose.
Color segmentation is done in HLS space, using a
parametrization learned off-line with a genetic
algorithm. It results in a series of objects that are
subjected to a shape analysis (figure 7).
Figure 7: Color segmentation and shape analysis.
The shape analysis is done by first filtering by
geometric restrictions, followed by a probabilistic
adjustment of straight and curve lines to the shapes
(Blake,2005), (Gonzalez,2002). Objects surviving
both processes have a high probability of being
traffic signs, because at least they have the right
colors and shapes.
Step 2: Sign reading and interpretation.
Objects surviving first step filtering are passed to the
reading phase. The content of the objects is matched
with a data base with the legal traffic signs. This is
based on correlation matching techniques with
dynamic deformation.
Figure 8: Reading and interpretation.
The final result is an interpreted sign (figure 8), that
is logged into the black-box’s database, and can be
used for driver alerting policies. If the interior of the
object is not recognized, it is discarded.
6 CONCLUSSIONS AND FUTURE
WORK
The prototype described in this article, developed
within the SACAT project, has been driven in tests
for more than 5000 km. Its current state and
performance is as follows.
The Black-box module is fully operative, logging
data of the vehicle and trip every second. The output
of this module can be adapted to any of the
upcoming UE standards for this kind of systems.
Within this module operates the signal recognition
subsystem, that currently works with vertical danger
and prohibition signs. This subsystem is currently in
a full experimentation and improvement phase,
having promising results:
- Right recognition in 83% of danger and
prohibition signs
- 5% confusion between similar signs at long
distances (2% short distances)
- Detection distance between 10 and 50 m
ICINCO 2006 - ROBOTICS AND AUTOMATION
388
- One image per second processing
Alert module is also fully operative. Results
highlights in crash simulations (alert messages were
sent to a private mobile phone) are:
- SMS with the MSD successfully received in
destiny: 98%. A few SMS were lost or not
sent in absence of coverage.
- Voice channel successfully established
when coverage available (95%).
Future improvements we are currently working in
are:
- Enhancing traffic sign recognition
subsystem by:
o Improving system robustness.
o Including informative panels to the
set of signals handled.
- Detecting road lines and lane markings for
preventing involuntary lane changes
- Using optical flow techniques to estimate
distances to other vehicles and to the road
limits (active radar). This helps the driver
to keep secure distances.
- Improving communication with emergency
services using UMTS (3G) technology.
This allows to establish a video call as well
as a voice call, allowing to evaluate the
state of the vehicle and its occupants.
REFERENCES
European Comission (DG Enterprise and DG Information
Society), eSafety forum: Summary report 2003, March
2003.
A. de la Escalera; J. Mª Armingol; M. Mata. “Traffic Sign
Recognition and Analysis for Intelligent Vehicles”.
Image and Vision Computing 21, pp 247-258, 2003 .
E-merge final report June 2004 IST-2001-34061.
http://www.gstforum.org/en/subprojects/rescue/about_
gst_rescue/introduction/e-merge.htm
Claus Bahlmann, Ying Zhu, Visvanathan Armes, Martin
Pellkofer, Thorsten Koehler: “A System for Traffic
Sign Detection, Tracking, and Recognition Using
Color, Shape, and Motion Information”. IEEE
Intelligent Vehicles Symposium (IV 2005). 2005.
A. de la Escalera, J. M. Armingol,, M.A. Salichs. “Traffic
Sign Detection for Driver Support Systems”. 3rd
International Conference on Field and Service
Robotics, pp 141-146.Espoo, Finlandia. June 11-13,
2001
Priese, Lutz and Klieber, Jens and Lakmann, Raimund and
Rehrmann, Volker and Schian, Rainer. “New Results
on Traffic Sign Recognition”. Proceedings "Intelligent
Vehicles Symposium '94".pp. 249-254. 1994.
Andrew Blake and Michale Isard. “Active Contours. The
Application of Techniques from Graphics, Vision,
Control Theory and Statistics to Visual Tracking of
Sahpes in Motion”. Springer-Verlag Berlin
Heidelberg. New York. ISBN 3-540-76217-5
E. González, V. Feliú, A. Adán, L. Sánchez. “Descriptores
de Fourier para identificación de objetos en entornos
3D”. http://www.isa-
cr.uclm.es/xxvjornadas/ConfMan_1.7/SUBMISSIONS
/140-liezgarthg.pdf
LOGGING, ALERT & EMERGENCY SYSTEM FOR ROAD TRANSPORT VEHICLES - An Experimental eCall,
Black-box and Driver Alerting System
389