Edited by: Tecnológico Superior Corporativo
Edwards Deming
January - March Vol. 6 - 2 - 2023
https://revista-edwardsdeming.com/index.php/es
e-ISSN: 2576-0971
Received: December 08, 2022
Approved: May 22, 2023
Page 42-53
Virtual reconstruction in traffic: a review of
techniques
Reconstrucción virtual en materia de tránsito: una revisión de
técnicas
María Fernanda Padilla Ortiz
*
Christian Esteban Rengifo Dávila
*
ABSTRACT
Virtual reconstruction of traffic scenes is an increasingly
useful tool for investigators and experts in the field. It
allows investigators to obtain a clearer picture of the
events surrounding a traffic accident, allowing for a more
accurate assessment of the data and a better understanding
of accident mechanisms. This qualitative research under a
systematic review presents the techniques and methods of
virtual reconstruction in traffic that have been developed
in recent years. As a result, eight articles published
between 2020 and 2023 that are part of high impact
repositories were obtained. It is concluded that virtual
reconstruction technology for the analysis of traffic
accident situations continues to make progress.
Keywords: Virtual reconstruction, forensic techniques,
accident analysis, 3d reconstructions, reconstruction
technology.
RESUMEN
La reconstrucción virtual de escenas de tránsito es una
herramienta cada vez más útil para los investigadores y
expertos en el campo. Permite a los investigadores
obtener una imagen más clara de los hechos que rodearon
un accidente de tránsito, lo que permite una evaluación
más precisa de los datos y una mejor comprensión de los
mecanismos de accidentes. Esta investigación de corte
cualitativo bajo una revisión sistemática presenta las
* Abogada de los tribunales y juzgados de la República. Universidad Nacional
de Chimborazo. mafer201215@gmail.com
https://orcid.org/0000-0003-3290-8374
Magíster en criminalística y ciencias forenses. Universidad Nacional de
Chimborazo. christianrengifo@hotmail.com
https://orcid.org/0000-0002-3367-8796
43
43
Edwards Deming Corporate Technology - July - September Vol. 6 - 2 - 2023 https://revista-edwardsdeming.com/index.php/es
e-ISSN: 2576-0971
técnicas y métodos de reconstrucción virtual en materia
de tránsito que se han desarrollado en los últimos años.
Como resultado se obtuvo ocho artículos publicados entre
el 2020 y 2023 que son parte de repositorios de alto
impacto. Se concluye que la tecnología de reconstrucción
virtual para el análisis de situaciones de accidentes de
tránsito sigue teniendo avances.
Palabras clave: Reconstrucción virtual, técnicas forenses,
análisis accidentes, reconstrucciones 3d, tecnología de
reconstrucción.
INTRODUCTION
Road traffic accidents are one of the leading causes of death and injury in the world.
Recent statistics show that more than 1.35 million people die each year in road traffic
crashes, and between 20 to 50 million are injured or disabled, moreover, this figure has
increased significantly in recent years. (World Health Organization (WHO), 2018).. In
Ecuador, in the year 2021, 2131 people were killed and 17 532 injured in traffic accidents,
which represents an increase of 33, 9 % and 33.8 % respectively compared to 2020. The
number of traffic accidents in Ecuador is worrisome, especially because of the number
of human lives lost as a result of these accidents. This situation has become a social
problem for various reasons (Saltos Salgado et al., 2005).(Saltos Salgado et al., 2020).
In the report published by INEC (2022) on traffic accidents details the most common
causes, where the driver's carelessness and recklessness with 43.5% is the main cause,
followed by disrespect for traffic signs with 21%, speeding with 14.3%, drunkenness or
drug use with 8%, pedestrian recklessness with 4.6% and other causes with 8.8%.
In addition, violators of the Organic Law of Land and Roads in Ecuador receive sanctions,
but this does not prevent the occurrence of these acts. This is due to the lack of
responsibility of those involved. As an example, Article 106 of the same Law determines
that "traffic infractions are those actions or omissions that, being able and should be
foreseen, but not wanted by the cause, are verified due to negligence, imprudence, or
inexperience, or due to non-observance of the laws, regulations, resolutions and other
traffic regulations" (Asamblea Nacional Constituyente, Asamblea Nacional
Constituyente, 2001). (National Constituent Assembly, 2008).
Therefore, in any traffic accident it is important to determine the cause and responsibility
for an accident, which is possible through forensic techniques because they allow
determining the speed of the vehicles involved, the direction of travel, the location of
the impact, any damage to the vehicle, evidence of reckless driving, and any other
relevant information to determine the cause of the accident.
In addition, accidents include the collection and analysis of physical evidence, the
collection and analysis of tire tracks, pavement marks, vehicle damage, steering and
speed tests, alcohol or drug tests, and other relevant tests. Once the physical evidence
44
44
Edwards Deming Corporate Technology - July - September Vol. 6 - 2 - 2023 https://revista-edwardsdeming.com/index.php/es
e-ISSN: 2576-0971
is collected, forensic experts conduct an analysis to determine the cause of the accident.
This involves observation of the physical evidence, flight line of objects related to the
accident, accident reconstruction, and vehicle trajectory reconstruction. The analysis
also involves studies of the braking and acceleration patterns of the vehicles involved.
They also collect and analyze documentary evidence. This evidence contains police
reports, witness reports, statements, ticket tickets, vehicle maintenance records,
information on the condition of the vehicles before the accident, and other documents
related to the accident to determine cause and liability.(Montes Loaiza et al., 2013; Tian
et al., 2010).In the same sense, Rodriguez (2017), states that there are major deficiencies
in the technical documentation related to the investigation and reconstruction of traffic
accidents and other forms of violent death.
With the advancement of technology it has been possible for forensic science to take
steps for the benefit of the investigation and in order to present evidence based on
information and communication technologies. Thus, this research presents a systematic
review of ICT-based methodologies and tools in traffic forensics developed in the last
decade.
MATERIALS AND METHODS
This research was based on the report of (Kitchenham, 2004) to conduct a systematic
review of the literature on methodologies and ICT techniques used for traffic accident
reconstruction to identify, evaluate and interpret all available relevant documentation. A
source selection strategy with inclusion and exclusion criteria was carried out, as well
as a predefined search to process all research related to the research question. This
made it possible to obtain and evaluate the available evidence to arrive at a synthesized
result.
The following research question was posed: What are the methodologies or techniques
available for the reconstruction of traffic accidents based on virtual technologies?
The next step was to determine the sources of information search, considering the
importance of reliable and relevant research, the use of academic articles, academic
congress reports found in the repositories of Science direct and Scielo was determined.
In relation to the search strategy, the keywords and terms used were "traffic
reconstruction", "crash reconstruction", "virtual methods accidents", "digital techiques
accidents", "traffic accident reconstructions", "3d traffic accidents". The temporality of
the articles was delimited between publications and preprints from the year 2020, and
articles written in English, considering that this is the language in which the journals with
the greatest impact publish on the subject.
Once the documents were obtained, the abstract was reviewed for agreement with the
research objective and the title search string "accident or techniques" was applied to the
terms.
Inclusion criteria
Articles and conference reports published from 2020 to 2023.
Articles found in Scopus.
45
45
Edwards Deming Corporate Technology - July - September Vol. 6 - 2 - 2023 https://revista-edwardsdeming.com/index.php/es
e-ISSN: 2576-0971
Open access documents.
Research and review articles.
Exlusion criteria
Articles and short communications.
Gray literature.
An article was only selected if it was found in more than one database.
Slides and books.
Table 1: Number of studies with search criteria
No
filters
Refined
search
Search
string/items.
Aligned to the
objective/summary
traffic
reconstruction
2214
473
86
8
crash
reconstruction
455
133
45
7
virtual methods
traffic
1755
540
22
11
TOTAL
26
Source: Researchers, 2023
Twenty-six articles were obtained, which were subjected to a more extensive reading
by the researchers and finally eight studies were selected that provide relevant
information for this research. The results and conclusions of the documents were read
in order to discover how these researchers have approached the topic of investigation
and the solutions or points of view that they disseminate. Table 2 presents the articles
analyzed.
Table 2: Detail of selected articles
Title
Authors
Magazine or main base
Using the scanners and drone
for comparison of point cloud
accuracy at traffic accident
analysis.
(Kamnik et al.,
2020)
Accident Analysis &
Prevention
Method for automated
generation of road accident
scene sketch based on data
from mobile device camera
(Saveliev et al.,
2020).
Proceed to
46
46
Edwards Deming Corporate Technology - July - September Vol. 6 - 2 - 2023 https://revista-edwardsdeming.com/index.php/es
e-ISSN: 2576-0971
Usage of Digital Evidence in
the Technical Analysis of
Traffic Collisions
(Malinová et al.,
2021).
Transportation Research
Procedia
Application of Close-Range
Photogrammetry in
Documenting the Location of
an Accident
(Stehel et al.,
2021).
14th International
scientific conference on
sustainable, modern and
safe transpor
Data-driven online traffic
reconstructions: Interactively
optimizing in virtual reality
(Wang et al., 2022)
Computers & Graphics
Using the iPhone's LiDAR
technology to capture 3D
forensic data at crime and
crash scenes
(Kottner et al., 2023).
Forensic Imaging
A feed-forwarded neural
network-based variational
Bayesian learning approach for
forensic analysis of traffic
accident.
(Xie et al., 2022).
Computer Methods in Applied
Mechanics and Engineering,
Circuits, Systems,
Communications and
Computers (CSCC),
International Conference on
(Sevcik et al., 2023).
Circuits, Systems,
Communications and
Computers (CSCC),
International Conference on
RESULTS
The eight selected papers were re-read taking into consideration how they answer the
research question. In the first paper that is part of this review it was determined that
the authors sought to test whether sketches obtained from different 3D point clouds
can be produced faster and more accurately than sketches obtained from classical police
measurements. To do so, they compared classical police work and sketching with
terrestrial laser scanners Riegl VZ-400i, Faro Focus S70, Geoslam ZebRevo 3D and data
collection from a Topcon Falcon 8 drone, all of which were used for 3D modeling and
sketching. They compared the accuracy using a graphical approach. For the subsequent
visual inspection of the traffic accident scene, the most suitable source is an orthophoto
obtained using the Surface from Motion (SfM) method, which is the result of processing
a series of georeferenced photographs taken from the UAV. They conclude that the
digital data provides the opportunity for a second look at the accident scene, whether
it is a quiet street or a busy road. The use of an accurate 3D representation of the scene,
such as a compilation of the crash data, can be helpful in addressing questions and doubts
that arise. This also provides the option to import directly into crash simulation tools,
such as PC-Crash or Analyser PR. (Kamnik et al., 2020)
47
47
Edwards Deming Corporate Technology - July - September Vol. 6 - 2 - 2023 https://revista-edwardsdeming.com/index.php/es
e-ISSN: 2576-0971
Thus, for the authors Saveliev et al. (2020) describe in their study that despite
government efforts to reduce the road fatality rate, road traffic accidents still occur
frequently and that accident scene sketches are performed manually in most countries,
which affects the quality and quantity of data collected. This poor quality of sketches
contributes to the return of cases for investigation. To improve this situation, the use
of an automatic RTA scene sketch generation system using mobile devices is proposed.
This will enable traffic officers to perform their work faster and more efficiently, increase
the accuracy and precision of traffic accident registration, and eliminate the subjectivity
of data presentation. The proposed method requires only a mobile device with a camera
to operate, which automatically generates sketches of road scenes. The process consists
of video recording vehicles and surrounding environments with a mobile device; building
maps and reconstructing the camera path using the SLAM method; reconstructing the
3D scene and classifying objects with convolutional neural networks; calibrating the scale
of the generated 3D scene and sizing the location of other objects; converting the 3D
model into a 2D RTA scene sketch. The developed system provides users with
recommendations on how to record video from different angles to capture all required
components, including the relative location of vehicles and static objects, vehicle damage
and surroundings within a 10 m radius. The video collected by the user will be
compressed on a mobile device and sent to a cloud for further processing. To generate
ATR scene sketches in an automated manner, we consider the basic stages of video
processing.
The ORB SLAM2 algorithm that is used offers faster initialization and operation, higher
model quality, and open source. It converts the video recorded by the user into a
sequence of frames, identifying features (key points) in each frame and positioning them
in a 3D space. The algorithm also generates an unoriented weighted visibility graph using
the keypoints and frames (Saveliev et al., 2020)
In the following research published by Malinová et al. (2021) they state that there are 3
ways to get data from the ECU: the first uses regular diagnostic tools, the second uses
the CDR (Crash Data Retrieval) system and the third uses the GIT Tool co. system.
Using diagnostic tools, maintenance and service related data can be read through the
OBD / OBD II socket. These records, known as freeze frame data, can be very useful
for accident analysis, although they are usually encrypted and only accessible to car
manufacturers. Thus, with this background, the authors state that the use of CDR (Crash
Data Retrieval) extracts and decodes the information stored in
The CDR system allows to extract and decode the information stored in the ACM
(Airbag Control Module). And that the data once downloaded can record up to 5
seconds before and after the accident, in addition to data on the speed difference, engine
speed, whether braking is taking place, the ignition cycle at the time of the collision, the
state of the seat belt and the state of the airbag. In addition, they specify that the data
must be stored securely to prevent tampering and be available to national authorities
for accident analysis. It must also be possible to accurately collect the model, version
and variation of the vehicle, as well as the active safety and accident prevention systems