=Paper=
{{Paper
|id=Vol-1956/GHItaly17_paper_05
|storemode=property
|title=Monsters of Darwin: A Strategic Game Based on Artificial Intelligence and Genetic Algorithms
|pdfUrl=https://ceur-ws.org/Vol-1956/GHItaly17_paper_05.pdf
|volume=Vol-1956
|authors=Daniele Norton,Laura Anna Ripamonti,Mario Ornaghi,Davide Gadia,Dario Maggiorini
|dblpUrl=https://dblp.org/rec/conf/chitaly/NortonROGM17
}}
==Monsters of Darwin: A Strategic Game Based on Artificial Intelligence and Genetic Algorithms==
https://ceur-ws.org/Vol-1956/GHItaly17_paper_05.pdf
Monsters of Darwin: a strategic game based on Artificial
Intelligence and Genetic Algorithms
Daniele Laura Anna Mario Davide Dario
Norton Ripamonti Ornaghi Gadia Maggiorini
University of University of University of University of University of
Milan Milan Milan Milan Milan
Milan, Italy Milan, Italy Milan, Italy Milan, Italy Milan, Italy
ripamonti@di. ornaghi@di. gadia@di.unimi.it dario@di.unimi.it
unimi.it unimi.it
ABSTRACT The analysis and design of the overall user experience in a
The production of video games is a complex process, which video game is not a straightforward task, because several fac-
involves several disciplines, spanning from art to computer tors can play a role. Recently, there has been a growing inter-
science. The final goal is to keep entertained the players, by est in the analysis of a number of features that, even if consid-
continuously providing them novel and challenging contents. ered apparently of minor importance, could, nonetheless, im-
However, the availability of a large variety of pre-produced pact deeply on players’ user experience with the game [2, 4,
material is often not possible. A similar problem can be 14, 15]. In particular, in some video game genres, one of the
found in many single-player game genres, where the simu- problematic features is that after spending a certain amount of
lated behaviour generated by the Artificial Intelligence algo- time playing, players become well aware of the overall char-
rithms must be coherent, believable, but also adequately var- acteristics of the game and of its contents, and consequently
iegate to maintain a satisfactory user experience. To this aim, they begin to lose interest [13]. The solution would be to
there is a growing interest in the introduction of automatic or continuously provide novel and challenging contents, and to
semi-automatic techniques to produce and manage the video dynamically adapt the game behaviour, in order to maintain
game contents. In this paper, we present an example of strate- the game enjoyable and fun. To this aim, a larger number of
gic card battle video game based on the applications of Ar- video games are introducing Procedural Content Generation
tificial Intelligence and Genetic Algorithms, where the game (PCG) techniques, together with Artificial Intelligence (AI)
contents are dynamically adapted and produced during the methods specifically developed to dynamically change and
game sessions. adapt the game contents and behaviour.
Procedural Content Generation (PCG) is an approach with
ACM Classification Keywords long and established history in different fields of Computer
H.5.m. Information Interfaces and Presentation (e.g. HCI): Graphics, aimed at creating data algorithmically. Examples
Miscellaneous; I.2.1. Applications and Expert Systems: of procedurally generated contents are textures [5] , buildings
Games; K.8.0. General: Games and cities [27, 28]. In the context of video games, PCG can
be exploited to increase randomness of content and/or game-
Author Keywords play, with also the positive effects of reducing development
Strategic Games; Video games; User experience; Artificial time. For example, it can be used to automatically create a
Intelligence for video games; Genetic Algorithms game level for platform games to achieve a desired level of
complexity [26]. In other works, the application on PCG and
AI techniques has been applied to several specific aspects of
INTRODUCTION game optimization, such as: impact of Non Player Characters
The production of video games is a complex process, which (NPCs) ( [1, 18, 25, 31, 33]), and adaptive or personalized
involves several disciplines, spanning from art to computer content generation ( [10, 32]). Finally, there is a large litera-
science. Regardless of the specific game genre, the design ture related to the applications of Genetic Algorithms (GAs)
and development processes must have as final goal the pro- in video games. The proposed techniques are mainly used to
duction of a video game able to propose an enjoyable gaming generate or evolve the game environment [3, 7, 24, 29], or
experience even after long periods of time. for the evolution of the game agents behaviour, in order to
produce more challenging opponents to the players [6, 11,
12, 21, 22, 23].
Differently from the other works on GAs in video games,
GHITALY17: 1st Workshop on Games-Human Interaction, April 18th, 2017,
the recently proposed GOLEM (Generator Of Life Embedded
Cagliari, Italy. into MMOs) algorithm [9] addresses explicitely the need to
GHITALY17:
Copyright © 2017 1st Workshop
for the on Games-Human
individual papers Interaction, September
by the papers' authors.18th, 2017,
Copying introduce more variety and unpredictability in the monsters
Cagliari,
permittedItaly.
for private and academic purposes. This volume is published and
Copyright
copyrighted © by
2017
its for the individual papers by the papers’ authors. Copying permitted
editors.
inside MMOs, in order to avoid that the players consider the
for private and academic purposes. This volume is published and copyrighted by its
editors.
game repetitive and less enjoyable after a long period of time The characterization of an AI algorithm for strategic video
spent playing. The main idea in GOLEM is to characterize game shares its terminology with the game theory. A game
each monster specie present in the game through its genome, is classified according to the number of players (usually two,
and to generate new species by recombining their chromo- even if it can be extended to a higher number of players), the
somes, which represent a set of physical characteristics and goal of the game, and the information each player has about
skills. Each monster in GOLEM is represented by a chromo- the game. With respect to the goal, two subclasses of strategic
some composed by 53 genes, and the recombination process games exist:
evaluates also the probability for the new monster to actually
survive in the habitat in which it is born (e.g., a marine-like • Zero-sum game: a player wins only if the other opponent
lose.
animal will unlikely survive in a desert). Moreover, other pa-
rameters are included in order to manage and, if needed, limit • Non zero-sum game: the focus of the player is on the win-
the population growth. ning, which may happen even without the opponent’s loss.
The concept of evolution of a population of creatures have Moreover, we can identify two cases also for the level of in-
been addressed also in a small number of commercial video formation the players have regarding the game:
games [8, 30], even if in a very different way than the ap-
proach proposed in GOLEM. • Perfect information: the player knows everything about the
state of the game, and about the possible options the oppo-
In this paper, we present a strategic card battle video game nent has for the next move.
called Monsters of Darwin (MOD). The game applies the GA
approach of GOLEM, but to a different game genre: in this • Imperfect information: there is some random element in
case, the monsters are depicted on virtual cards, and at each the game that make uncertain the possible evolution of state
turn, a new card with a new monster may be generated by the of the game.
combination of the monsters on the cards currently played. The most popular strategic video game AI technique is the
Moreover, being MOD a single-player video game, an AI minimax algorithm. The main concept is that, during a turn-
component has been implemented, to manage the game dy- based game, a player tries to play the best move possible to
namics and to act as Non Playable Character (NPC). achieve victory, while the opponent tries to use the best strat-
The paper is organized as follows: in the following sections egy to avoid this, by minimizing the player’s score. The mini-
we will recall the fundamental concepts related to AI and GA max algorithm is a recursive algorithm, which works on a tree
in video games, and then we will describe MOD design and data structure (the game tree) where each node represents a
implementation choices. Finally, we will draw conclusions board position, and each branch represents one possible move,
and discuss major future developments. leading from one board position to another. At each iteration,
the AI algorithm evaluates the current state, considers each
possible move and corresponding new board positions, and
ARTIFICIAL INTELLIGENCE IN VIDEO GAMES recursively calculates the score for each new possible situa-
Contemporary video games are often characterized by the use tion, in order to choose the most appropriate one. The choice
of advanced AI techniques. AI solutions for video games is performed using a heuristic called static evaluation func-
[17] are quite different from those typical of classic AI, be- tion [17].
cause while the latter are usually aimed at optimizing solu-
tions for specific problems under no time or resources lim- GENETIC ALGORITHMS IN VIDEO GAMES
its, the former must provide a believable solution to complex GAs are a particular class of algorithms, mainly belonging to
decisional processes in (nearly) real-time with limited com- the AI field. They are applied to solve many classes of prob-
putational resources. AI in video games has a huge amount lems but, more in general, they are useful in heuristic search
of possible usages, mostly depending on the specific game processes and in optimization problems. They have been in-
genre. One example is represented by Non-Playing Charac- spired by Darwin’s evolution theory: a population is repre-
ters (NPCs) in First Person Shooter (FPS) games, where the sented by the chromosomes of a set of individuals, and a new
AI is responsible of the simulation of a believable behaviours generation in the population is produced by recombining the
for a group of enemies [16, 25]. genetic material according to specific rules. For each genera-
tion, a fitness function selects the most “suitable” parents and
In this paper, we consider a strategic video game. The video
iterates the reproduction process on them. To produce “chil-
games in this genre are characterized mainly by actions based dren”, the algorithm applies crossover (a genetic recombina-
on tactics and planning, in order to achieve victory. Player’s tion technique) and mutations on chromosomes, represented
decisions have a key role in the game, while chance is less
by bit sequences. The algorithm then evaluates, once the new
relevant than players’ ability. There are two sub-genres, turn-
generation has been created, if the population registers any
based or real-time, and the role of AI can vary relevantly de- improvement in any relevant feature: if this is the case, par-
pending on the nature of the strategic video game. Real-time
ents will be discarded and the new children will substitute
strategic games have usually a simpler AI, which is mainly
them in the reproduction process. Generally, these steps are
used as a strategic layer. On the contrary, turn-based games iterated until some optimal solution is reached [19, 20].
may have very sophisticated and complex AI, which may
even run on dedicated hardware, like in case of simulated Crossover is used by GAs to mix the genes of two elements
complex board games (e.g., Chess) [17]. of the population. Some different approaches can be applied:
1. Single point crossover: the chromosomes of the parents
are split into two parts in a randomly selected point. Two
new chromosomes are generated, by combining the first
and second parts of the original parents.
2. Two points crossover: the chromosomes of the parents are
split into three parts. The chromosome of one child is com-
posed by the first and third part of the first parent, merged
with the second part of the second parent. Another chromo-
some is created using the remaining parts.
3. Uniform crossover: this approach provides a higher genetic
variation, since each gene of the child is copied randomly
from one of the corresponding genes belonging to one of Figure 1. An example of Monster Card. In the bottom area it is indicated
the parents. The genes not chosen for the first child are the natural element of the monster. The attack, defense and vital forces of
used for the second one the monster are shown in the top area.
4. Arithmetic crossover: the offspring chromosomes are the
results of some arithmetic operation on the parents’ genes.
finally, the attack, defense and vital forces of the monster are
Finally, genetic mutations can be useful for inserting into the shown in the top area.
new generation’s chromosomes some characteristics not in-
heritable from parents, since not present in their genetic her- The four natural elements (air, earth, water and fire) are used
itage. Similarly to what happens in nature, mutations can to create a hierarchy between the Monster Cards, used during
introduce a new characteristic or modify/destroy an existing the Monster Duel or Monster Coupling stages. In particular:
one. • air rules over earth
MONSTERS OF DARWIN (MOD) • earth rules over water
Monsters of Darwin has been developed in order to investi-
gate novel approaches aimed at maximizing the user experi- • water rules over fire
ence in a particular game genre that, after long periods, may
• fire rules over air
suffer because of the repetition of game contents.
To allow the graphical representation of the new monsters
MOD is a zero-sum, imperfect information, turn-based single-
generated by the GA, each monster is composed by a set of
player strategic card battle video game. The player plays
2D patches, each having a texture of a different physical part
against an AI opponent in order to collect the highest possible
(head, body, legs, etc). With a similar approach, a new mon-
number of Monster Cards.
ster can be simply represented by assembling patches from
The game dynamics of MOD are quite simple. At each turn, different sets. Figure 2 shows a monster, and the patches com-
two cards are played, one from the player on turn (the “real” posing its graphical representation.
player or the AI component) and one from the opponent.
At the beginning of the game, the player has a card deck com-
Then, the player on turn must choose between two different
posed by 8 Monster Cards (2 for each natural element). In the
actions among the two cards: Monster Duel or Monster Cou-
following turns, the player will choose 8 cards among those
pling. Monster Duel follows rules similar to other card battle
available in her deck. The AI module creates its deck choos-
games, while Monster Coupling represents a procedural ap-
ing the Monster Cards randomly, even if balancing among the
proach aimed at lower the game contents repetitivity, by the
number of cards belonging to the different natural elements.
automatic generation of new Monster Cards depicting new
monsters. Monster Coupling is based on the application of a
GA, derived from the GOLEM algorithm [9]. A game ses- Monster Duel
sion ends when a player defeats all the cards of the opponent. During a Monster Duel, the damage a Monster Card can in-
The final winner is the first player winning 3 game sessions. flict to the vital force of the opponent’s card is established by
the difference between the corresponding attack and defense
In the following subsections, we provide further details on the force values (e.g., a card with an attack force of 8 will inflict
features and contents of MOD. a damage of 2 to a card with defense force 6). The damage
can be modified on the basis of their relation in the natural
Monster Cards elements hierarchy: the damage is raised of 1 unit in case the
The main contents of MOD are the Monster Cards, depict- element of the attacking card rules over the element of the
ing different kinds of monsters, using a graphic style inspired other card, otherwise it is lowered of 1 unit. All the Monster
by grimoires and medieval manuscripts. Figure 1 shows an Cards have an initial vital force equal to 6. When the value
example of Monster Card. Each card is composed by three reaches 0, the Monster Card is removed by the current game
areas: in the bottom area it is indicated the natural element of session, and it will become available only in the following
the monster; in the central area it is depicted the monster, and, games.
Characteristic Type Range
Head physical 8
Eyes physical 8
Arms physical 8
Body physical 8
Legs physical 8
Tail physical 8
Wings physical 8
Attack force 3
Defense force 3
Vital force 3
Figure 2. A monster of MOD, and the patches composing its graphical Element natural element 4
representation
Table 1. Characteristics and skills in a chromosome of a monster.
Monster Coupling and Genetic Algorithm in MOD
The GA implemented in MOD is applied during the Mon- During the selection of the Monster Card to play, if the AI is
ster Coupling stage. As in GOLEM [9], each monster is the player on turn, the choice of the card is completely ran-
described by a chromosome, which maps its characteristics dom. If, otherwise, the AI is responding to the player’s card
and skills. The characteristics are related to the physical as- choice, than the algorithm evaluates the played card, and per-
pect (e.g., number of legs, wings, etc.), to the force (i.e., the forms a selection of a “stronger” card among the available
attack, defense and vital forces), and to the natural element, ones in the deck. A card is “stronger” if its natural element
as illustrated in Table 1. rules over the element of the first card, and/or its attack force
is higher. During the choice between Monster Duel or Mon-
Each characteristic can assume a value in a certain range. For ster Coupling, the AI algorithm evaluation is based on the
the physical features, the range from 1 to 8 indicates one Mon- number of remaining Monster Cards in the deck, and on the
ster Card from the initial deck: therefore, a value of 2 for the difference of vital force and compatibility between the two
Head means that the new monster will be created using the Monster Cards.
head patch from the monster of the second card selected in
the deck, etc. For the force features, the value changes de-
pending on the subcategories: attack force can have values MOD implementation details
from 4 to 7, defense force can have values from 1 to 3, vital MOD has been implemented using the Unity 3D game en-
force can have values from 0 to 6. These values have been gine. For its characteristics, Unity 3D represents a powerful
chosen after a tuning and testing stage of the game. and flexible solution well suited for fast implementation of
The Monster Coupling stage begins evaluating the compati- different game genres, allowing the integration of different
bility between the two Monster Cards currently played. Two techniques and material, and the development for different
monsters are evaluated as compatible for the coupling if the gaming platforms.
natural element of the card of the player on turn rules over Figure 3 shows some screenshots of the Monster Duel and
the element of the opponent’s card, and if the difference in Monster Coupling stages.
vital force between the two monsters is equal of higher of 2.
If these constraints are not valid, then the Monster Card of the
player on turn (i.e., the one who has tried the Monster Cou- CONCLUSIONS AND FUTURE WORK
pling) will be inflicted a damage following the rules of the In this paper, we have presented a strategic card battle video
Monster Duel stage. Otherwise, the Crossover function of game called Monsters of Darwin (MOD). The game uses a
the GA algorithm is applied to the chromosomes of the two combination of AI and GAs in order to generate new Monster
monsters. To obtain the broadest diversity among the gener- Cards, by combining the characteristics of the cards played in
ated monsters, we have opted for the uniform crossover ap- each turn.
proach: for each characteristic in the chromosome, one value
The approach we are proposing with MOD is a preliminary
from the two parents is randomly selected and assigned to the
attempt to address one of the features that can impact on play-
new monster. Even if the crossover creates two new monsters,
ers’ user experience. In fact, repetition of contents and me-
only the first is considered, and it is added to the player’s per-
chanics in some game genres can lead to a loss of interest in
sonal deck, while the two original cards subject to the cou-
the game after some time. One possible solution to avoid this
pling are discarded from the game session.
issue is to introduce procedural and dynamic techniques to
continuously generate novel contents and to adapt the game
Artificial Intelligence in MOD
behaviour to the players’ actions, in order to maintain the
The main operations of the AI component in MOD are: the game enjoyable and fun.
selection of the Monster Card to play, and the choice between
Monster Duel or Monster Coupling actions. In both cases, the From the preliminary evaluation tests, the performances of
AI component uses the minimax algorithm for the evaluation. MOD are promising: the GA in MOD is able to produce a
Figure 3. Screenshots of the Monster Duel and Monster Coupling stages in MOD.
high number of new monsters even from a limited set of ba- Real Data. In eHealth 360°: International Summit on
sic Monster Cards, while the evaluation made by the AI algo- eHealth, Revised Selected Papers. Springer International
rithm is performed smoothly, limiting the waiting time to the Publishing, 110–118.
minimum.
5. David S. Ebert, F. Kenton Musgrave, Darwyn Peachey,
Future developments will consider the extension of the char- Ken Perlin, and Steve Worley. 2003. Texturing &
acteristics of the monsters chromosomes, and of the database Modeling: a procedural approach. Morgan Kaufmann.
of available monsters among which the players can choose
6. Anna I. Esparcia-Alcázar and Jaroslav Moravec. 2013.
the initial deck. Moreover, we aim at extending the AI com-
Fitness approximation for bot evolution in genetic
ponent, by introducing new possible actions during the game
programming. Soft Computing 17, 8 (2013), 1479–1487.
turns, like e.g., a defense action to oppose the attack from the
opponent player. 7. Miguel Frade, Francisco Fernandez de Vega, and Carlos
Cotta. 2012. Automatic evolution of programs for
REFERENCES procedural generation of terrains for video games. Soft
1. Gustavo Andrade, Geber Ramalho, Hugo Santana, and Computing 16, 11 (2012), 1893–1914.
Vincent Corruble. 2005. Automatic Computer Game
Balancing: A Reinforcement Learning Approach. In 8. Galactic Arms Race homepage. 2017. (2017).
Proceedings of the Fourth International Joint http://galacticarmsrace.blogspot.it.
Conference on Autonomous Agents and Multiagent 9. Andrea Guarneri, Dario Maggiorini, Laura Anna
Systems (AAMAS ’05). ACM, New York, NY, USA, Ripamonti, and Marco Trubian. 2013. GOLEM:
1111–1112. Generator Of Life Embedded into MMOs. In
2. Richard Bartle. 2003. Designing Virtual Worlds. New Proceedings of the Twelfth European Conference on the
Riders Games. Synthesis and Simulation of Living Systems: Advances
in Artificial Life, ECAL. 585–592.
3. Leonardo F. B. Silva de Carvalho, Helio C. Silva Neto,
Roberta V. V. Lopes, and Fábio Paraguaçu. 2010. An 10. Erin J. Hastings, Ratan K. Guha, and Kenneth O.
application of genetic algorithm based on abstract data Stanley. 2009. Automatic Content Generation in the
type for the problem of generation of scenarios for Galactic Arms Race Video Game. IEEE Transactions on
electronic games. In 2010 IEEE International Computational Intelligence and AI in Games 1, 4 (Dec
Conference on Intelligent Computing and Intelligent 2009), 245–263.
Systems, Vol. 2. 526–530. 11. Yuan Hong and Zhen Liu. 2010. A First Study on
4. Daniele De Felice, Marco Granato, Laura Anna Genetic Algorithms Based-Evolvable Motivation Model
Ripamonti, Marco Trubian, Davide Gadia, and Dario for Virtual Agents. In 2010 International Conference on
Maggiorini. 2017. Effect of Different Looting Systems Multimedia Technology. 1–4.
on the Behavior of Players in a MMOG: Simulation with
12. Johannes Inführ and Günther R. Raidl. 2012. Automatic 23. Antonio M. Mora, Mária A. Moreno, Juan J. Merelo,
Generation of 2-AntWars Players with Genetic Pedro A. Castillo, Maribel G. Arenas, and Juan L. J.
Programming. In Computer Aided Systems Theory – Laredo. 2010b. Evolving the cooperative behaviour in
EUROCAST 2011: 13th International Conference, Unreal™ bots. In Proceedings of the 2010 IEEE
Revised Selected Papers, Part I. Springer Berlin Conference on Computational Intelligence and Games.
Heidelberg, Berlin, Heidelberg, 248–255. 241–248.
13. Raph Koster and Will Wright. 2004. A Theory of Fun for 24. Fausto Mourato, Manuel Próspero dos Santos, and
Game Design. Paraglyph Press. Fernando Birra. 2011. Automatic Level Generation for
14. Dario Maggiorini, Alessandro Nigro, Laura Anna Platform Videogames Using Genetic Algorithms. In
Ripamonti, and Marco Trubian. 2012a. Loot Proceedings of the 8th International Conference on
Distribution in Massive Online Games: foreseeing Advances in Computer Entertainment Technology (ACE
impacts on the player base. In 8th International ’11). ACM, New York, NY, USA, Article 8, 8 pages.
Workshop on Networking Issues in Multimedia 25. Laura Anna Ripamonti, Silvio Gratani, Dario
Entertainment (NIME’12) Int. Conf. on Computer Maggiorini, Davide Gadia, and Armir Bujari. 2017a.
Communication Networks ICCCN. Believable group behaviours for NPCs in FPS games. In
15. Dario Maggiorini, Alessandro Nigro, Laura Anna Proceedings of IEEE Digital Entertainment, Networked
Ripamonti, and Marco Trubian. 2012b. The Perfect Virtual Environments, and Creative Technology
Looting System: Looking for a Phoenix?. In 2012 IEEE Workshop (DENVECT 2017).
Conference on Computational Intelligence and Games 26. Laura Anna Ripamonti, Mattia Mannalà, Davide Gadia,
(CIG). 371–378. and Dario Maggiorini. 2017b. Procedural content
16. Dario Maggiorini, Laura Anna Ripamonti, and Samuele generation for platformers: designing and testing FUN
Panzeri. 2013. Follow the Leader: a Scalable Approach PLEdGE. Multimedia Tools and Applications 76, 4
for Realistic Group Behavior of Roaming NPCs in (2017), 5001–5050.
MMO Games. In Proceedings of the Twelfth European 27. Manlio Scalabrin, Laura Anna Ripamonti, Dario
Conference on the Synthesis and Simulation of Living Maggiorini, and Davide Gadia. 2016. Stereoscopy-based
Systems: Advances in Artificial Life, ECAL 2013, Sicily, procedural generation of virtual environments. In
Italy, September 2-6, 2013. 706–712. Proceedings of IS&T’s Stereoscopic Displays and
17. Ian Millington and John Funge. 2009. Artificial Applications XXVII (28th Symposium on Electronic
Intelligence for Games, Second Edition. Morgan Imaging : Science and Technology). 1–7.
Kaufmann Publishers Inc., San Francisco, CA, USA. 28. Michael Schwarz and Pascal Müller. 2015. Advanced
18. Olana Missura and Thomas Gärtner. 2009. Player Procedural Modeling of Architecture. ACM Trans.
Modeling for Intelligent Difficulty Adjustment. In Graph. 34, 4, Article 107 (July 2015), 12 pages.
Discovery Science: 12th International Conference. 29. Nathan Sorenson and Philippe Pasquier. 2010. Towards
Springer Berlin Heidelberg, Berlin, Heidelberg, a Generic Framework for Automated Video Game Level
197–211. Creation. In Applications of Evolutionary Computation:
19. Melanie Mitchell. 1998. An Introduction to Genetic EvoApplicatons 2010. Springer Berlin Heidelberg,
Algorithms. The MIT Press. Berlin, Heidelberg, 131–140.
20. Charles. J. Mode and Candace K. Sleeman. 2012. 30. Spore homepage. 2017. (2017). http://www.spore.com.
Stochastic Processes in Genetic and Evolution: 31. Pieter Spronck, Marc Ponsen, Ida Sprinkhuizen-Kuyper,
Computer Experiments in the Quantification of Mutation and Eric Postma. 2006. Adaptive Game AI with
and Selection. World Scientific Publishing Company. Dynamic Scripting. Mach. Learn. 63, 3 (2006),
21. Antonio M. Mora, Antonio Fernández-Ares, Juan J. 217–248.
Merelo, Pablo García-Sánchez, and Carlos M. 32. Julian Togelius, Renzo De Nardi, and Simon M. Lucas.
Fernandes. 2012. Effect of Noisy Fitness in Real-Time 2007. Towards automatic personalised content creation
Strategy Games Player Behaviour Optimisation Using in racing games. In Proceedings of the IEEE Symposium
Evolutionary Algorithms. Journal of Computer Science on Computational Intelligence and Games. Togelius is
and Technology 27, 5 (2012), 1007–1023. working at IDSIA on SNF grant 21-113364 to J.
22. Antonio M. Mora, Ramón Montoya, Juan J. Merelo, Schmidhuber.
Pablo G. Sánchez, Pedro Á. Castillo, Juan L. J. Laredo, 33. Georgios N. Yannakakis and John Hallam. 2009.
Ana I. Martínez, and Anna Espacia. 2010a. Evolving Real-Time Game Adaptation for Optimizing Player
Bot AI in Unreal™. In Applications of Evolutionary Satisfaction. IEEE Trans. Comput. Intellig. and AI in
Computation: EvoApplicatons 2010. Springer Berlin Games 1, 2 (2009), 121–133.
Heidelberg, Berlin, Heidelberg, 171–180.