Serious Games Scenario Modeling for Non-experts
Yohan Duval
, David Panzoli
, Axel Reymonet
, Jean-Yves Plantec
, J
ome Thomas
and Jean-Pierre Jessel
IRIT, University of Toulouse, Toulouse, France
CUFR Jean-Franc¸ois Champollion, Albi, France
ACTIA Automotive, Toulouse, France
Serious Games, Authoring, Scenario Modeling, Game-based Learning.
The use of serious games and gamified softwares is a new and growing trend for training professionals in a
wide variety of disciplines where procedures and decision-making are key (automotive diagnostic, surgery,
etc). Serious Games are safer, less expensive and advocated to be more efficient. Unfortunately, there is a
lack of methodology and tools adapted for non-computing experts to develop their own gamified learning
scenarios. In this paper, we propose an approach allowing trainers to model professional activities in the form
of serious games scenarios. Trainers are enabled to express their expertise using a domain specific graphical
representation which will be implemented eventually in an easy-to-use authoring tool. The produced scenarios
describe both the actions necessary for completing the professional procedure and the associated pedagogy to
provide the learner with relevant educational feedback. The proposed approach specifies a model matching
those requirements, and is illustrated by an application example in the automotive context. We intend to
demonstrate that an appropriate model is likely to make scenario editing accessible to trainers who are not
necessarily familiar with computer modeling in the first place.
(Alvarez and Djaouti, 2011) defines a Serious Game
as a “computer application, for which the original
intention is to combine with consistency, both seri-
ous aspects [. . . ] with playful springs from the video
game”. Serious Games are becoming a new form
of well-appreciated training tools, because of their
safety, their low cost and their efficiency (Van Est
et al., 2011). Although trainers would benefit intro-
ducing Serious Games in their training, developing
such tools is too difficult for them, for obvious com-
plexity reasons, but also, we believe, in lack of an
adapted methodology.
Yet, involving the trainer in the process is essen-
tial for two reasons. Firstly, they have the knowledge
to describe professional activities in which they are
specialized, using their own natural language. Sec-
ondly, they are able to provide learners with relevant
pedagogical feedback during the execution of these
activities, such as the accomplishment of important
educational objectives or the validation of acquired
skills with Multiple Choice Questions (MCQ). How-
ever, trainers do not have the necessary computing ex-
pertise to transfer in a natural manner these two di-
mensions into a computer intelligible-formal format.
In this paper, we define a scenario as the descrip-
tion of the professional activities in terms of game in-
teractions associated with the description of the ped-
agogical feedback. Our objective is to propose an
approach that empowers the trainers with the abil-
ity to create serious game scenarios on their own
by expressing their expertise as naturally as possi-
ble. Throughout the article, we illustrate our approach
with a scenario from the Diag’Adventures project, set
in the automotive context. Particularly, the game sce-
nario we refer to describes the first steps of a car di-
agnostic operation, where activities are expected to
be carried out on the car as well as on the diagnos-
tic software. Figure 1 displays the virtual environ-
ment in which the game is set. In order to fulfill the
steps of the procedure, the user can switch at any time
between the garage view (a), where they can interact
mainly with a virtual car, and the diagnostic software
view (b). Whereas the garage is entirely simulated
in a 3D virtual environment, it is interesting to note
that the genuine diagnostic software is utilised. This
entails a more immersive simulation of the process,
Duval Y., Panzoli D., Reymonet A., Plantec J., Thomas J. and Jessel J..
Serious Games Scenario Modeling for Non-experts.
DOI: 10.5220/0005489904740479
In Proceedings of the 7th International Conference on Computer Supported Education (CSEDU-2015), pages 474-479
ISBN: 978-989-758-107-6
2015 SCITEPRESS (Science and Technology Publications, Lda.)
Figure 1: The game environment of Diag’Adventures fea-
tures two views : (a) the interactive car in 3D and (b) the
diagnostic software.
which should allow for a less steep learning curve
and a more efficient transfer into the real world of
the knowledge acquired virtually, since it eases the
back and forth transitions between the game and the
real-life activities. In the next section, we will briefly
review the related work. Then, we will point out the
coexistence of two separate dimensions within a sce-
nario: the description of the activity and the peda-
gogy. This will lead us to propose a model, which
we will detail and illustrate with the aforementioned
scenario. We conclude that our approach empowers
the trainers with the necessary tools to describe their
professional and pedagogical expertise in an expres-
sive yet formal fashion.
In the domain of virtual environment modeling, pre-
vious works have already been carried out to allow
non-experts to perform this complex and multidis-
ciplinary task. Existing textual modeling languages
have been used to describe and orchestrate virtual en-
vironments. (Ishida, 2002) has extended the Scheme
language to describe interaction scenarios between
agents inside a 3D environment. The Q language
thus created introduces in particular cues and actions,
which refer respectively to observations and modifi-
cations in the environment. This allows an event trig-
gering system which is expressive enough to be used
by non-experts. In (Devillers and Donikian, 2003),
authors have chosen to define their own grammar to
design a textual language that allows the orchestra-
tion of agents in a 3D virtual world into multiple sub-
scenarios. This language provides a syntax close to
C++, and has the particularity of proposing instruc-
tions that can be represented as Finite State Machines.
It is also possible to use Extensible Markup Language
(XML) or JavaScript Object Notation (JSON) with
specific schemas to describe different objects in a vir-
tual environment (Tang et al., 2013) or to describe
Web-Based Multimodal Interaction Scenarios (Kat-
surada et al., 2003). These languages have the advan-
tage of being simpler to implement thanks to a large
amount of dedicated tools. However, the direct use of
these languages to describe activities in a virtual en-
vironment implies that the author of the scenario has
computer-expert skills, and therefore this approach
does not match our requirements.
Graphical notations can also be resorted to for the
design of scenarios in a virtual environment. For
instance, Unified Modeling Language (UML) class
and state diagrams are used in (Tang and Hanneghan,
2008). (Buche et al., 2010) designs procedures in a
virtual world using UML activity diagrams. (Pan-
zoli et al., 2014) uses the Business Process Model
and Notation (BPMN) language to model the activ-
ities and the interactions of several actors in a 3D
virtual operating room. Game scenarios can also
be described using Petri Nets (Ara
ujo and Roque,
2009). They proved to be powerful mathematical
analysis tools, but as such their usage is restricted
to an expert population. These last three represen-
tations have the advantage of being easier to use by
non-computing experts due to their graphical nature.
However, they still lack expressiveness: BPMN does
not support the definition of virtual world objects
attributes (Tang and Hanneghan, 2008), Petri Nets
have difficulties to model events in scenarios (Tang
and Hanneghan, 2008), and UML is too complex
since modeling every aspect of a scenario requires
the use of many types of diagrams (Tang and Han-
neghan, 2008)(Ara
ujo and Roque, 2009). To over-
come these problems, some researchers have used
their own representation. One common way to do so
is to use graphical blocks. (Van Est et al., 2011) uses
high-level configurable building blocks to describe
serious games scenarios. This representation allows
non-experts to naturally express professional activi-
ties over time with actions and events; however, it still
lacks support for the description of concurrent tasks
and for a clear description of the associated pedagogy.
Puzzle blocks, used in Scratch (Resnick et al., 2009)
and Alice (Kelleher and Pausch, 2006) for teaching
programming to children, allow the description of
various games scenario using a user-friendly repre-
sentation. However, these blocks are too low-level for
a non-computing expert to simply express high-level
The graphical approach seems to be the most rele-
vant choice for defining a higher-level framework and
methodology within the reach of non-programming
experts. Indeed, textual languages do not provide
them with enough expressiveness to describe in a for-
mal and simple representation both the professional
activities and the pedagogy. Existing traditional and
custom graphical notations do not provide every fea-
ture we need either. However, the principle of cues
and actions (Ishida, 2002), also used by (Van Est
et al., 2011) with events and actions, is interesting
since it allows non-experts to naturally express the
common distinction between observing and modi-
fying the virtual environment. We are also inter-
ested in using configurable building blocks (Van Est
et al., 2011) as the scenario description basis, com-
bined with BPMN (White, 2004) (Panzoli et al., 2014)
and activities diagram notations (Buche et al., 2010),
since it gives users the possibility to model high-level
tasks with traditional representations.
As we pointed out in the introduction, trainers have
the knowledge to describe both professional activities
and associated pedagogical feedback in a natural lan-
guage. On the other hand, they lack the computing
expertise to transfer this knowledge into a computer-
intelligible format, to finally create scenarios fitting
their need and directly interpretable by a Serious
Game Engine. Most of the methods previously dis-
cussed describe activities in an actor-centered point
of view. Our approach will propose a method that de-
scribes these activities from the virtual environment
point of view. As a matter of fact, it allows trainers
to focus on the expected result instead of what the
player should do in order to reach this result, which
is more natural and straightforward to them. Besides,
we did not notice any custom graphical modeling no-
tation (blocks modeling) that uses existing and ex-
pressive representations (UML, BPMN) to model ac-
tivities. Therefore, we will endeavor to specify a pre-
cise graphical syntax allowing non-computing experts
to naturally express learning scenarios in a virtual en-
vironment. Furthermore, none of the previous meth-
ods takes into account the description of pedagogical
feedback during the player’s activities. Regarding our
approach, we want to give the trainer the possibility
to add educational information in the scenario, so the
learner can get direct feedback during their learning
sessions. Two dimensions in the process of describing
serious game scenarios distinctly arise. On the one
hand, the professional activities need to be described
as sequences of tasks. On the other hand, pedagogi-
cal objectives and feedback (advice messages, MCQ,
etc.) need to be modeled and associated to the ex-
ecution of the activities. These two dimensions are
orthogonal as they cannot be naturally described us-
ing the same representation. However, they share a
common entity: the activity block, which simply rep-
resents an elementary action in the virtual environ-
ment. Those two dimensions, along with their respec-
tive properties, are detailed in the next sections.
3.1 Scenario Modeling: Activities
The first dimension consists in describing how the
learners activities should be modeled. The goal we
seek to achieve is to allow a non-computing expert
trainer to make use of their expertise to describe these
activities as naturally as possible using a graphical
representation. Basically, the graphical notation is
expected to mediate between i) the description in a
natural language of one or several professional activi-
ties, and ii) the equivalent description in a formal lan-
guage that can be used by the game engine to run the
scenario. BPMN is a good candidate, since its pri-
mary goal is to provide a notation that is readily un-
derstandable by all business users, from the business
analysts to the technical developers to the business
people (White, 2004). However, this representation
does not support the configuration of each individual
task. This is why the configurable building blocks ap-
proach introduced in (Van Est et al., 2011) is interest-
ing: it allows the division of the global activity into
a set of individual configurable tasks. For the rest of
this paper, we will call them activity blocks.
Some activity blocks can be generic enough to be
reused in whatever scenarios, like logic blocks or vari-
ables. However, a large majority of blocks will be
context-specific. Anyway, we classify them into 3
categories: events, actions, and observations. Each
of these blocks represent a different notion, but they
all aim at describing activities from the objective per-
spective of the virtual environment rather than the
player’s subjective point of view. It allows to describe
the expected result without having to specify how to
reach this result.
The first notion a trainer would want to describe
is the need to wait for a specific change in the virtual
environment to resolve the next task. This is exactly
why we define event blocks. In most cases, they will
be used to wait for the player to do an action, like
When the door is opened, then . . . ”. But it can also be
used to wait for a particular state of the environment
induced by a previous action from the player, for ex-
ample “When the car speed is above 60 km/h, then
. . . ”. We represent Events as circles.
The second notion the trainer would need to de-
scribe is the fact that at some point during the ex-
ecution of a given activity, they want to modify the
virtual environment to place the learner in a specific
situation. We use Actions blocks to express this idea.
For example, it could be “Move the car from A to B”,
or “Change the Camera View to the inside of the car”.
We choose to represent Actions as rectangles.
Finally, the trainer may want to check one or sev-
eral attributes of the virtual environment, to perform
different actions depending on the observed result.
We define observation blocks to represent this notion.
It helps to represent cases like “If the lights are on,
then . . . else . . . ”. We represent Observations as dia-
mond shapes.
We define each of these activity blocks to be able
to possess a set number of arguments, so they can
be reused in different contexts. For example, for the
event “When the car speed is above 60 km/h, then
. . . ”, the speed limit could be a variable argument in
the block. Additionally, we want to allow the scenario
author to reuse sequences of activity blocks in future
scenarios or in different sections of a same scenario,
without them having to describe the sequence twice.
In this perspective, we let the trainer define composite
blocks, which are basically the merging of a sequence
of elementary activity blocks into one higher-level
block. They also have a set number of arguments, so
it is possible to quickly modify some of the parame-
ters of the internal elementary blocks. These compos-
ite blocks are stored in a library accessible from any
Finally, the trainer may want to describe concur-
rent activities, to express the fact that two different
tasks can be performed in parallel. To represent this
idea, we use the BPMN notion of pools. A scenario
has at least one pool, which contains the main activ-
ities description. It starts with a special event block
named “Start” and ends with a another named “End”.
From this single pool, we can add as much as nec-
essary, each of them containing a series of activities,
starting with a global event to specify when the asso-
ciated concurrent activities is being started.
3.2 Scenario Modeling: Pedagogy
The second dimension consists in defining how
should be modeled the pedagogical feedback the
learner will receive in the virtual environment while
they play a training session. We introduce pedagogy
blocks to describe these feedback.
A learning scenario has pedagogical objec-
tives (Alvarez and Djaouti, 2011) (Garris et al., 2002)
that need to be specified so the learner knows their
progress along their activity and what they have
learned so far. To let the non-expert express this no-
tion, we define objective blocks. The writer can in-
stantiate as many of these blocks as necessary.
We also introduce a tree view representation so the
trainer can hierarchically order these objectives. In
practice, completing one (high-level) objective could
depend on completing multiple sub-objectives, and so
on in a recursive fashion. Starting from the root which
represents the global objective of the scenario, the au-
thor will be able to specify a hierarchy between ob-
jectives and sub-objectives.
3.3 Associating the Activity and the
Having described both the activities and the pedagogy
of a scenario, by means of their respective graphical
notation, interconnecting those two dimensions hap-
pens at the level of the activity block. The associa-
tion is performed by the author. It consists, for each
elementary objective (a leaf on the objective tree),
in defining two notions: which task starts the objec-
tive? and which task completes the objective? This is
achieved by connecting each objective to a composite
block (i.e. a series of block as detailed in section 3.1),
the first block within which being related to starting
the objective and the last block being related to com-
pleting the objective.
In the objectives description, the author is given
the possibility to define two sequences of feedback
blocks per objective: one for when the the objective
is started, and another for when it is completed. They
can then express logic like “When the sub-objective 1
is started, display the advice. . . or “When the objec-
tive 2 is completed, launch a MCQ to check acquired
3.4 Application Example
To illustrate the proposed approach, we have come up
with the Figure 2 presenting a scenario description ex-
ample applied to the context of car diagnosis. In this
scenario, the trainee is merely expected to establish
Main Activity: connect the diagnostic
software to the car
Concurrent Activity 1: plug the VCI to the
Concurrent Activity 2: plug the VCI to the
USB port
Test Variable
(Activity 1 AND
2 OK )
Move Camera To
Move Camera To
Connect VCI to car
Connect VCI to
Access Vehicle
Select Vehicle
Software to car
Connect VCI to
Navigate through
the Diagnostic
Access Vehicle
Select Vehicle
Connect VCI to
(Select Vehicle)
(Auto Detect)
(Auto Detect)
Figure 2: A toy-example scenario in Diag’Adventures. At the top (green), the pedagogy tree. At the bottom (red), the
activities. Learning objectives and activities are connected by the dotted green lines.
the connection between the diagnostic software and
the (virtual) car. The process is described in the pro-
fessional activities description view. The software is
mainly a succession of screens, in which the user can
interact with a number of GUI elements, to progress
in their diagnostic activity. We could just have de-
scribed the succession of screens that would complete
the main objective, if we did not care how the learner
accessed them. However, in this particular scenario,
we want the user to follow an optimal path; this is why
we also describe the buttons that lead to this optimal
use of the software. The two concurrent activities de-
scribe interactions in the 3D virtual environment that
can be done at any time during the scenario.
In the pedagogy description, we defined the hi-
erarchy between the scenario objectives and sub-
objectives. Each leaf of the tree is then associated
with a sequence of activity blocks in the activities de-
scription view. Pedagogical feedback are represented
by the icons on the objective nodes/leaves. As an
example, when the “Navigate through the diagnotic
software” objective is started, we change the support-
ing message to indicate to the learner what they have
to do. Then, when this objective is completed, we
present a MCQ to the user to check if they have un-
derstood the process.
Through this example, we see the two advantages
of the graphical representation and the methodology
we propose. Firstly, it makes a clear division between
the descriptions of the professional activities and the
pedagogy. This allows trainers to simply express their
dual-expertise into two orthogonal dimensions, and
to establish the link between them only afterwards.
Secondly, the common entity shared by these two di-
mensions, the activity block, uses a specific syntax
which gives enough expressiveness for the user to de-
scribe a large amount of professional activities. That
being said, it must be noted that the expressiveness is
restricted to a specific context and therefore the ap-
proach is more suited for describing procedural ac-
tivities as compared to activities entailing an absolute
freedom of the learner inside the environment. As a
matter of fact, it is easier to monitor the player’s ac-
tions in a procedural scenario where we describe the
optimal path, than in a free scenario where we could
merely describe checkpoints. Besides, the pedagogy
feedback could not be described as accurately as in
a procedure. In the future, we plan to work on this
aspect in order to cover a larger number of scenarios.
In this paper we have described a methodology and
the associated graphical modeling approach allowing
the description of serious games scenarios. The main
idea is to split a scenario into two dimensions differ-
ent by nature yet connected: the pedagogy and the
procedure. It is advocated that a non-computing ex-
pert can then express his expertise in the most natural
Future work will address the design and imple-
mentation of an authoring tool to validate this ap-
proach. We will define and model a number of var-
ious representative scenarios using our methodology.
We intend to measure the expressiveness of our repre-
sentation, and assess the proportion of the situations
likely to be imagined by a trainer that can be mod-
eled. Among these scenarios, we will have a partic-
ular interest in the description of collaborative tasks
like in (Panzoli et al., 2014) or (Buche et al., 2010).
Diag’Adventures is supported by the Midi-Pyr
region and collaboratively developed by ACTIA Au-
tomotive, OPERANTIS, and the University JF Cham-
Alvarez, J. and Djaouti, D. (2011). An introduction to seri-
ous game definitions and concepts. Serious Games &
Simulation for Risks Management, page 11.
ujo, M. and Roque, L. (2009). Modeling games with
petri nets. Breaking New Ground: Innovation in
Games, Play, Practice and Theory. DIGRA2009. Lon-
dres, Royaume Uni.
Buche, C., Bossard, C., Querrec, R., and Chevaillier, P.
(2010). Pegase: A generic and adaptable intelligent
system for virtual reality learning environments. In-
ternational Journal of Virtual Reality, 9(2):73–85.
Devillers, F. and Donikian, S. (2003). A scenario language
to orchestrate virtual world evolution. In Proceedings
of the 2003 ACM SIGGRAPH/Eurographics sympo-
sium on Computer animation, pages 265–275. Euro-
graphics Association.
Garris, R., Ahlers, R., and Driskell, J. E. (2002). Games,
motivation, and learning: A research and practice
model. Simulation & gaming, 33(4):441–467.
Ishida, T. (2002). Q: A scenario description language for
interactive agents. Computer, 35(11):42–47.
Katsurada, K., Nakamura, Y., Yamada, H., and Nitta, T.
(2003). Xisl: a language for describing multimodal
interaction scenarios. In Proceedings of the 5th inter-
national conference on Multimodal interfaces, pages
281–284. ACM.
Kelleher, C. and Pausch, R. (2006). Lessons learned from
designing a programming system to support mid-
dle school girls creating animated stories. In Visual
Languages and Human-Centric Computing, 2006.
VL/HCC 2006. IEEE Symposium on, pages 165–172.
Panzoli, D., Sanselone, M., Sanchez, S., Sanza, C.,
Lelardeux, C., Duthen, Y., and Lagarrigue, P. (2014).
Introducing a design methodology for multi-character
collaboration in immersive learning games. Sixth In-
ternational Conference on Virtual Worlds and Games
for Serious Applications: VS-Games 2014.
Resnick, M., Maloney, J., Monroy-Hern
andez, A., Rusk,
N., Eastmond, E., Brennan, K., Millner, A., Rosen-
baum, E., Silver, J., Silverman, B., et al. (2009).
Scratch: programming for all. Communications of the
ACM, 52(11):60–67.
Tang, S. and Hanneghan, M. (2008). Towards a domain spe-
cific modelling language for serious game design. In
6th International Game Design and Technology Work-
shop, Liverpool, UK.
Tang, S., Hanneghan, M., and Carter, C. (2013). A platform
independent game technology model for model driven
serious games development. Electronic Journal of e-
Learning, 11(1):61–79.
Van Est, C., Poelman, R., and Bidarra, R. (2011). High-
level scenario editing for serious games. In GRAPP,
pages 339–346.
White, S. A. (2004). Introduction to bpmn. IBM Corpora-