Thesis subject: Multi-agent
modeling of seismic crisis
Keywords: mobility, pedestrians, crisis, earthquake,
modeling, simulation,
multi-agents, serious games
Description:
In the field of seismic risk, the continued evolution of
construction codes has
led to a reduction in building vulnerability. However, urban
areas are becoming
increasingly populated. In France in 1950, the urbanization rate
was 55% of the
total population, this is predicted to rise to around 83% in
2030 (United
Nations, 2006). This growth coupled with the consequent increase
in infrastructure,
buildings, and activities, make urban areas particularly
vulnerable
environments. Faced with seismic risk, institutions and
companies can prepare in
three ways: by defining and implementing building codes and
planning policies
adapted to local seismicity; educating and preparing people to
deal with
crisis; and improving crisis management plans. However, in
countries with
moderate seismic risk, such as France, it is difficult to know
the real impact
of these measures given the rarity of such events. Therefore
modelling becomes
a useful tool. Immediately following an earthquake, people
evacuate mostly on
foot (Bertran Rojo, 2014) making them dangerously exposed to
falling
non-structural elements. This can be fatal, as was the case for
9 victims of the
Lorca earthquake (Martínez Moreno et al. 2012).
The objectives of this thesis
are: to develop a
multi-agent model to simulate pedestrian mobility following an
earthquake; to measure
evacuation effectiveness; to assess peoples risk of exposure and
security based
on their individual behaviours; and to evaluate current or
future crisis
management and information plans. This work represents a
significant advance
with what is currently possible in terms of modelling and
simulation of human
behaviour in crisis situations. With the innovative combination
of a BDI (belief,
desire intention) approach, GIS, multi-agent simulation and
serious gaming, the
goal is to develop a generic crisis model, which can be applied
to different
geographical contexts. Based on the work of Bratman (Bratman
1997; Rao and
Georgeff, 1995), the BDI approach will model the actions of
individuals taking
into account their underlying motivations, thus providing a
dynamic and
realistic model of human behaviour in crisis situations (Adam et
al. 2011;
Dugdale et al. 2010). To promote the transfer of the results of
the model to civil society, a serious game will be
developed.
The thesis addresses major social and
scientific
challenges: developing models that adequately represent the
complexity of
human behaviour, viewing crisis management as a complex
socio-technical system
and being attentive to emergent and self-organising phenomena,
investigating
techniques to support the emergence of mutual knowledge, etc.
To following
methodology will be used to achieve the goals of the thesis:
•
Step 1: Develop a model and a generic
multi-agent simulator
following an iterative approach:
1)
Undertake a detailed state of the art on mobility
modeling in crisis
situations;
2)
Data compliation and analysis of mobility
behaviours in seismic situations
(survey results analysis);
3)
Model design including definition of the physical
environment, underlying
rules, and agents’ characteristics and behaviours following a
BDI approach;
4)
Formalise the model using agent-UML, which is
especially suited for
multi-agent modeling;
5)
Implement the model in a simulator using the GAMA
platform, which offers
high-level primitives and the ability to integrate GIS spatial
data. The project
has close connections with the GAMA development community,
composed of a dozen
researchers from various research laboratories;
6)
Validate and calibrate the model and simulator,
following the companion
modelling approach to iteratively improve the model by involving
stakeholders (citizens
and decision makers);
7)
Perform simulations, after defining scenarios
(e.g. the introduction of new
information campaigns, defining an evacuation plan, shelter
places, etc.) and conducting
sensitivity analysis to test model and simulator robustness.
•
Step 2: Apply the simulator to a town
in the Rhone-Alpes
region: real geographic data will be collected and incorporated
into the model,
step 6) extended simulation; the actual municipality is not yet
finalized and
will be discussed with local stakeholders via the project
partner (PARN);
• Step 3: Create a
serious game
based on the model for the transferring results to civil society
and to facilitate
communication between actors. Game design will be done in
collaboration with
Christelle Gaïdatzis (Solidarity Association Images). The game
will be validated
and gaming sessions will be organized with the partners at the
end of the
thesis in collaboration with Christelle Gaïdatzis.
Supervision: Julie Dugdale (Grenoble
Informatics Laboratory, LIG, MAGMA
research team) and Elise Beck (Public Policy, Political Action,
Territories, main
research theme "Risk, Environment, Society").
Associated
research laboratories:
the student will be linked to two
laboratories:
• Grenoble Informatics
Laboratory (LIG).
https://www.liglab.fr/
• Public
Policy, Political Action,
Territories (PACTE) http://www.pacte-grenoble.fr/
Partners: The
doctoral research is
based on a partnership involving the above mentioned
laboratories, an
organisation devoted to scientific research and transferring
results to the
community (PARN), and an Association for scientific mediation
(Images
Solidaires) for the design of the non-computer based serious
game, and its use
with different actors (decision makers, citizens).
Candidate profile: the student should possess a M2 diploma (or
equivalent) in computer
science, cognition, geography or geomatics (good grades are
essential).
Required competences:
-
Good programming skills in an object oriented
language
-
Knowledge of modeling and multi-agent simulation
(NetLogo, GAMA) would be
advantageous
-
Interest in the issue of natural hazards and
multidisciplinary approaches
-
Spoken and written English is recommended. Spoken
and written Spanish is
useful but not essential.
-
Intellectual curiosity, inventiveness and
dedication.
Important note: Once accepted, the successful candidate will be
required to register at
the MSTII Doctoral school (Mathematics,
Information science and technology, and Computer science). As a
consequence the
awarded PhD will be in the area of computer science.
Application procedure: Candidates should send an email with their CV, examination
grades
and a cover letter
to Julie Dugdale
([log in to unmask]) and Elise Beck
([log in to unmask]) by the 21st June 2015
at the latest.
Interviews will be held in early July.
Start date: 1st October 2016
Funding: Rhône-Alpes Region, ARC 7 ("Innovations,
Mobility, Territories and
Urban Dynamics") doctoral allowance. Monthly salary, approximate
1,300
Euros, Net.
References:
Adam
C., Gaudou B.,
Hickmott S., Scerri D. (2011) - BDI agents and social
simulations - In
RIA 25(1). Special issue on Agent-based simulation. F. Amblard
and J. Dugdale
(Eds) - January-February
2011.
Bertran Rojo, M. (2014).
Correr entre los escombros - Courir
entre les débris. La mobilité individuelle en période de crise
sismique:
facteur d’exposition humaine dans le cas du séisme de Lorca
(Espagne 2011).
Thèse de doctorat, Université de Grenoble. 254 p.
Bratman, M. E. (1999)
[1987]. Intention, Plans, and
Practical Reason. CSLI Publications. ISBN 1-57586-192-5.
Dugdale,
J., Bellamine-Ben
Saoud, N., Pavard, B. and Pallamin, N. (2010). Simulation and
Emergency
Management. In Van de Walle, B., Turoff, M. and Hiltz, R.H.
(eds) Information
Systems for Emergency Management. Series: Advances in Management
Information
Systems. Sharp.
Martínez Moreno, F., A.
Salazar Ortuño, J. Martínez Díaz, J.
A. López Martín, R. Terrer Miras, et A. Hernández Sapena. 2012.
« EsLorca: Una
Iniciativa Para La Educación Y Concienciación Sobre El Riesgo
Sísmico ».
BOLETÍN GEOLÓGICO Y MINERO. 123 (4): 575‑588.
Rao A.S. and M. P. Georgeff.
BDI-agents: From Theory to Practice. In Proceedings of the First
International
Conference on Multiagent Systems (ICMAS'95), San Francisco,
1995.
United Nations, Department of
Economic and Social Affairs, Population Division. 2006: World
Urbanization
Prospects: The 2005 Revision. Urban and Rural Areas Dataset
(POP/DB/WUP/Rev.2005/1/Table A.2), dataset in digital form. http://esa.un.org/unup/.
New York: United
Nations.