No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Automatic generation and analysis of physics-based puzzle games
Abstract
In this paper we present a method for the automatic generation of content for the physics-based puzzle game Cut The Rope. An evolutionary game generator is implemented which evolves the design of levels based on a context-free grammar. We present various measures for analyzing the expressivity of the generator and visualizing the space of content covered. We further perform an experiment on evolving playable content of the game and present and analyze the results obtained.
... Manual content production is expensive and potentially not scalable (Iosup 2011). In contrast to manual content production, Procedural Content Generation (PCG) is the application of computer software to generate game content, specifically the algorithmic generation of game content with limited or no human contribution (Togelius et al. 2013). However, PCG is considered difficult as it not only incurs extensive computational overhead, but also requires the ability to compute the technical and cultural values of the generated instances (Hendrikx et al. 2013). ...
... In addition, it has been noted that the manual creation of game content has a range of other drawbacks that include a lack of flexibility and suggestions that manual approaches are inherently unscalable (Roden and Parberry 2004). The automatic creation of game content is not new, with examples dating back to the 1980s (Togelius et al. 2013), however there are continuing challenges in finding ways to reduce unwanted artefacts that can be encountered using simple random generation of content. To that end, a number of approaches have been considered with a growing emphasis on the use of computational intelligence techniques such as evolutionary computation. ...
... In general, most of the recent research in this area falls into one of the two categories presented above. The first being where the output of the process is evaluated in terms of whether it achieved the desired design goal (Merrick et al. 2013;Shaker et al. 2013) and the second being where some comparison is made, often between different representations or different tuning of the algorithms (Liapis et al. 2015). ...
This paper describes a study that examines the impact that procedurally generated content has on the quality of gaming experience. To that end, an experimental study has been undertaken where gamers play two versions of an otherwise identical game, the only difference being that in one version the game levels are designed by a human designer and in the second version they are procedurally generated. A game immersion questionnaire is used to capture the quality of the gameplay experience and the results across the two groups compared. Whilst there are observable differences in perceived total immersion, statistical analysis using one way ANOVA testing suggests that the difference is not statistically significant. Detailed analysis of the questionnaire responses identifies where variation between the two groups is statistically significant.
... Different techniques have been explored to automatically generate different aspects of content and some of them achieved remarkable results in commercial games [5], [6], [7], [8]. The automatic generation of various aspects of game content has been explored relatively extensively for different game genres such as car racing games [9], space shooter games [2], strategy games [10], platform games [3], [11] and physics-based games [12], [13], [14]. ...
... In what follows, we describe several expressivity measures that we defined to analyse our generator. Some of the measures are inspired by previous work on expressivity analysis [12]. ...
... To facilitate a more in-depth insight on the differences between the generated levels we converted them into one colour map [12]. The colour map is an image where each pixel is the average colour value of all pixels at the same position in the full set of levels generated. ...
Spelunky is a game that combines characteristics from 2D platform and rogue-like genres. In this paper, we propose an evolutionary search-based approach for the automatic generation of levels for such games. A genetic algorithm is used to generate new levels according to aesthetic and design requirements. A graph is used as a genetic representation in the evolution process to describe the structure of the levels and the connections between the rooms while an agent-based method is employed to specify the interior design of the rooms. The results show that endless variations of playable content satisfying predefined difficulty requirements can be efficiently generated. The results obtained are investigated through an expressivity analysis framework defined to provide thorough insights of the generator’s capabilities.
... Request permissions from Permissions@acm.org. [10, 11, 12] . This genre has recently become very popular , especially on mobile devices, and some of the main titles are Angry Birds, Cut the Rope and Tower of Goo. ...
... It is an interesting application for PCG, because it has several physics constraints to be considered when evaluating the quality of the content generated. Therefore, the playability evaluation is another issue once it needs to be done with a physics simu- lator [10]. This paper presents an estimation of distribution algorithm (EDA) for generating Angry Birds levels. ...
... Experiments were conducted aiming to indicate the strengths and weaknesses of proposed EDA by analysing its expressivity , i.e. the space of all levels that it can create [10]. It is done using three metrics based on those proposed in [6] and [10]: frequency, linearity and density. ...
This paper presents an estimation of distribution algorithm (EDA) to generate levels for physics-based puzzle games with the Angry Birds mechanics. The proposed EDA keeps three probability tables during its evolutionary process to sample new individuals that encode informations about the amount and placement of game objects inside the level. Sampled individuals are evaluated by a simulation-based fitness function, which considers the stability and the amount of the game objects inserted in a level. The best individual sampled from the probability tables is used to update them. Experiments indicated that the proposed EDA was capable of creating stable structures related to the Angry Bird gameplay.
... Request permissions from Permissions@acm.org. ACE '14, November 11 -14 2014 [10,11,12]. This genre has recently become very popular, especially on mobile devices, and some of the main titles are Angry Birds, Cut the Rope and Tower of Goo. ...
... It is an interesting application for PCG, because it has several physics constraints to be considered when evaluating the quality of the content generated. Therefore, the playability evaluation is another issue once it needs to be done with a physics simulator [10]. ...
... Experiments were conducted aiming to indicate the strengths and weaknesses of proposed EDA by analysing its expressivity, i.e. the space of all levels that it can create [10]. It is done using three metrics based on those proposed in [6] and [10]: frequency, linearity and density. ...
This paper presents an estimation of distribution algorithm (EDA) to generate levels for physics-based puzzle games with the Angry Birds mechanics. The proposed EDA keeps three probability tables during its evolutionary process to sample new individuals that encode informations about the amount and placement of game objects inside the level. Sampled individuals are evaluated by a simulation-based fitness function, which considers the stability and the amount of the game objects inserted in a level. The best individual sampled from the probability tables is used to update them. Experiments indicated that the proposed EDA was capable of creating stable structures related to the Angry Bird gameplay.
... Despite their popularity, to the best of the authors' knowledge, this genre has not been deeply explored in the ield of PCG. There are only a few studies devoted to content generation for the Cut the Rope game [15], [16], [17]. ...
... This genre is an interesting application for PCG, because it has several physics constraints, which must be considered in the evaluation of the quality of the generated content. Evaluating playability is another issue in this genre since this needs to be done based on a physics simulator [15]. This paper presents a level generator for the Angry Birds game based on an evolutionary algorithm which represents a level with an array of columns. ...
... As described in [15], the expressivity of the generator is the space of all levels it can generate and a metric that indicates the generator strengths and weaknesses. This metric can be calculated by generating a large number of levels and evaluating their meaningful aspects. ...
This paper presents a genetic algorithm (GA) for the procedural generation of levels in the Angry Birds game. The GA evaluates the levels based on a simulation which measures the elements' movement during a period of time. The algorithm's objective is to minimize this metric to generate stable structures. The level evaluation also considers some restrictions, leading the levels to have certain characteristics. Since there is no open source code of the game, a game clone has been developed independently of our algorithm. This implementation can be used to support experiments with procedural content generation (PCG) methods for this game type. We performed experiments in order to evaluate the expressivity of the level generator and the results showed that the proposed algorithm could generate levels with interesting stable structures.
... Shaker et al. focused on the generation of levels for physicsbased puzzle games, using a clone of the mobile game Cut The Rope (2010) as a test ground [43]. The goal in Cut The Rope, shown in Fig. 16, is to make the candy drop in such a way that it reaches a frog monster placed at a fixed position. ...
... The game generator by Shaker et al. [43] evolves levels based on a context-free grammar, which is a set of recursive rewriting rules. Design grammars offer a concise way Levels, which are the phenotypes, are represented by lists of objects that can be placed anywhere in the map and may have some properties. ...
Procedural Content Generation (PCG) for games has existed since the 1980s and is becoming increasingly important for creating gameworlds, backstory and characters across many genres, in particular open-world games such as
and
. A particular challenge faced by such games is that the content and/or gameplay may become repetitive. Puzzles constitute an effective technique for improving gameplay by offering players interesting problems to solve, but the use of PCG for generating puzzles has been limited compared with its use for other game elements, and efforts have focused mainly on games that are strictly puzzle games, rather than creating puzzles to be incorporated into other genres. Nevertheless, a significant body of work exists, which allows puzzles of different types to be generated algorithmically, and there is scope for much more research into this area. This paper presents a detailed survey of existing work in PCG for puzzles, reviewing 32 methods within eleven categories of puzzles. For the purpose of analysis, the paper identifies a total of seven salient characteristics related to the methods, which are used to show commonalities and differences between techniques and to chart promising areas for future research.
... Most of the work in PLG for physics-based puzzle games has been conducted in the context of the Angry Birds game [10], [14], [15], [11], [12], however a few other works also used Cut the Rope as testbed [16], [17]. An Angry Birds level is composed of a stable pile of blocks which contains pigs inside a shown in Figure 1. ...
... There are few other works approaching the problem of PLG for physics-based puzzle games using the Cut the Rope game as testbed [16], [17]. In [17], a method based on Grammar Evolution generates levels using a fitness function that tries to find the best placement of game objects according to the rules of the game. ...
This paper presents Tanager, a level generator based on a genetic algorithm that is capable of producing feasible levels for the Angry Birds game. Evaluating playability in this game requires checking both the stability of the stacked blocks and the possibility of killing all the pigs with the given amount of birds. These two components are handled by the algorithm through a simulation. Three sets of experiments are conducted to evaluate Tanager. The first one measures the performance of the genetic algorithm underneath Tanager. The second one explores the expressivity of the generated levels considering their structural characteristics. The third one measures design quality of levels via an on-line user study. Results show that Tanager is capable of generating a considerable variety of feasible levels that are as engaging and enjoyable as those manually designed. However, the generated levels are less challenging than the hand-authored ones.
... However, as far as we can tell, very little research has been done on this particular area of PCG. A small collection of studies have explored PCG for the physics-based game Cut the Rope [13], [14], as well as the popular mobile game Angry Birds [15], [16], [17]. ...
... This is typically expressed as a metric which indicates the generator's strengths and weaknesses in various capacities. In this paper we define four measures based on metrics used in previous research [14], [15], [25]: frequency, linearity, density and leniency. Frequency evaluates the number of times that a block occurs within a structure. ...
... Mantere et al. demonstrated how a genetic algorithm can generate challenging Sudoku puzzles [11]. Shaker et al. applied an evolutionary computation technique to a physicsbased puzzle game [12]. Fatemi et al. designed Sudoku levels with varying difficulty by solving a constraint-satisfaction problem that guarantees a unique solution [13]. ...
This study presents case studies using two wave function collapse (WFC) methods, graph-based WFC and simple tiled WFC, to create playable levels for two logic puzzle games: Strimko (Latin Squares) and Flow (connecting dots with pipes). We then evaluate the quality of the generated levels through extensive experiments. Our results indicate that WFC-generated levels are high quality, follow the graph structures' constraints, and are generated faster than levels generated by depth-first search and genetic algorithms. WFC methods can also adapt to new system specifications, common in puzzle games, by changing only the data instead of the code. This increases the stability of content production based on procedural content generation since it relies on data rather than procedures. Furthermore, WFC methods increase the efficiency of the manual process of creating in-game puzzle levels, allowing game designers to complete more tasks in the same amount of time and create a wider variety of assets.
... The physics constraints employed in these types of games create many problems for PLG and makes evaluating the quality of levels difficult. The playability/solvability of generated levels is particularly difficult to confirm, due to the exceptionally large state and action spaces (Shaker et al. 2013). This paper presents a procedural level generator for physics-based puzzle games similar to Angry Birds. ...
This paper presents a procedural generation algorithm for levels in physics-based puzzle games similar to Angry Birds. The proposed algorithm creates levels consisting of various self-contained structures placed throughout a 2D area. Each structure can be placed either on the ground or atop floating platforms within the available level space. These structures are created using a variety of different block types and do not require predefined substructures or composite elements. Target object locations are determined based on a combination of factors, including structural protection, occupancy estimation and overall dispersion. Experiments were performed in order to determine the ideal input parameters for generating desirable levels. The expressivity of the generator was also evaluated and the results show that the proposed method can generate a wide variety of interesting levels.
... We note that the overall idea of generating such a landscape over games ties closely with prior works on taxonomization of multiplayer games 38,85 . Moreover, 2D visualization of the expressitivity (i.e., style and diversity) and the overall space of of procedurally-generated games features have been also investigated in closely related work 86,87 . A recent line of related inquiry also investigates the automatic identification, and subsequent visualization of core mechanics in single-player games 88 . ...
Multiplayer games have long been used as testbeds in artificial intelligence research, aptly referred to as the Drosophila of artificial intelligence. Traditionally, researchers have focused on using well-known games to build strong agents. This progress, however, can be better informed by characterizing games and their topological landscape. Tackling this latter question can facilitate understanding of agents and help determine what game an agent should target next as part of its training. Here, we show how network measures applied to response graphs of large-scale games enable the creation of a landscape of games, quantifying relationships between games of varying sizes and characteristics. We illustrate our findings in domains ranging from canonical games to complex empirical games capturing the performance of trained agents pitted against one another. Our results culminate in a demonstration leveraging this information to generate new and interesting games, including mixtures of empirical games synthesized from real world games.
... There has already been some research in generating content for some puzzle games, but without keeping in account the learning aspect [4]. Some examples include: level generation for the Cut the Rope (ZeptoLab, 2010) puzzle game [17], generation of Sokoban levels [11], and generation of narrative puzzles for adventure games [7]. Moreover, the type of adaptive content generation that this paper describes can be seen as an instance of the experience-driven procedural content generation framework (EDPCG) [19], where the game adaptation mechanism generates puzzles with particular elements and challenges in response to the player's actions. ...
This paper describes a system to generate puzzles with a difficulty degree that adapts to the player. The puzzle is designed with the objective of being used by young pupils, and it is mainly a planning/sequencing task, which is considered one of the aspects of computational thinking. The system is powered by a constrained multi-objective algorithm (NSFI-2Pop) – which evolves the sequences of actions necessary to solve the puzzle – combined with a stochastic algorithm that translates the sequences in playable levels. We also present a pilot evaluation of the system, which seems to indicate that the levels presented to the player are perceived as having an increasing difficulty.
... A number of studies tackle solvability, such as those of Powley et al. [27], Shaker et al. [28] or Volkmar et al. [29] that aided the level design of (procedurally generated) games by assuring potential solutions are feasible. Schatten et al. [30] simulated large-scale dynamic agent systems to test quest solvability in MMORPGs. ...
Balancing the options available to players in a way that ensures rich variety and viability is a vital factor for the success of any video game, and particularly competitive multiplayer games. Traditionally, this balancing act requires extensive periods of expert analysis, play testing and debates. While automated gameplay is able to predict outcomes of parameter changes, current approaches mainly rely on heuristic or optimal strategies to generate agent behavior. In this paper, we demonstrate the use of deep player behavior models to represent a player population (n = 213) of the massively multiplayer online role-playing game Aion, which are used, in turn, to generate individual agent behaviors. Results demonstrate significant balance differences in opposing enemy encounters and show how these can be regulated. Moreover, the analytic methods proposed are applied to identify the balance relationships between classes when fighting against each other, reflecting the original developers' design.
... Procedural Content Generation in games [26,28] (PCG) comes in a variety of flavors. AI has been shown to excel in the automatic creation of levels [2,17], narrative [23], tutorials [4,6], levels for tutorials [5,15], puzzles [14,25], and even entire games [3,16]. The functionality and acceptable similarity of the content depends on the genre the AI generates for, but it is generally desired that high quality content can be generated rapidly on-demand. ...
This paper presents a method for generating floor plans for structures in Minecraft (Mojang 2009). Given a 3D space, it will auto-generate a building to fill that space using a combination of constrained growth and cellular automata. The result is a series of organic-looking buildings complete with rooms, windows, and doors connecting them. The method is applied to the Generative Design in Minecraft (GDMC) competition [24] to auto-generate buildings in Minecraft, and the results are discussed.
... Procedural Content Generation in games [26,28] (PCG) comes in a variety of flavors. AI has been shown to excel in the automatic creation of levels [2,17], narrative [23], tutorials [4,6], levels for tutorials [5,15], puzzles [14,25], and even entire games [3,16]. The functionality and acceptable similarity of the content depends on the genre the AI generates for, but it is generally desired that high quality content can be generated rapidly on-demand. ...
This paper presents a method for generating floor plans for structures in Minecraft (Mojang 2009). Given a 3D space, it will auto-generate a building to fill that space using a combination of constrained growth and cellular automata. The result is a series of organic-looking buildings complete with rooms, windows, and doors connecting them. The method is applied to the Generative Design in Minecraft (GDMC) competition to auto-generate buildings in Minecraft, and the results are discussed.
... L'objectif du General Game Playing (GGP) est de développer des programmes-joueurs capables de jouer à une grande variété de jeux [22]. De nombreux systèmes utilisés pour modéliser des jeux, communément appelés General Game Systems, existent actuellement pour différents types de jeux incluant : les jeux déterministes à information complète [11], les jeux combinatoires [2], les puzzles [28], les jeux de stratégies [17], les jeux de cartes [10] et les jeuxvidéos [25]. ...
Bien que les systèmes actuels de General Game Playing (GGP) facilitent la recherche en Intelligence Artificielle (IA) autour des jeux, ils sont souvent trop spécialisés et fournissent une capacité de calcul trop faible. Cet article décrit une première version du système ludémique de GGP dénommé LUDII qui apporte un outil efficace à la recherche en IA tout autant qu’aux concepteurs de jeux ou aux historiens mais aussi dans bien d’autres domaines. LUDII définit les jeux comme des arbres de ludèmes, correspondant à des concepts et mécanismes de jeu de haut niveau et facilement interprétables. Nous établissons les bases de LUDII en analysant ses principaux avantages : généralité, extensibilité, compréhensibilité et efficacité. Expérimentalement, LUDII surpasse l’un des plus performants raisonneurs décrit avec le General Game Description Language (GDL), basé sur un réseau de propositions (propnet), pour tous les jeux disponibles sur le serveur GGP.
... Automatic simulations of video game play have proven to be usable in situations where human testing is too tedious or not exhaustive enough for the purpose of finding bugs and glitches [4,15,38,45], parameter tuning [60], and assuring solvability [42]. Based on the insights and the potential of our previous work on a tool for completing and debugging adventure games [35]), we want to further extend the possibilities of autonomous game testing. ...
Due to a steady increase in popularity, player demands for (online) video game content are growing to an extent in which consistency and novelty in challenges is hard to attain. Challenges in balance and error-coping accumulate. We introduce the concept of deep player behavior models by applying machine learning techniques to individual, atomic decision-making strategies. We discuss their potential application fields in personalized challenges, autonomous game testing, human agent substitution, and online crime detection. Results from a pilot study that was carried out with the massively multiplayer online role-playing game Lineage II depict a benchmark between hidden markov models, decision trees, and deep learning. Data analysis and individual reports indicate that deep learning can be employed to provide adequate models of individual player behavior with high accuracy for predicting skill-use and a high correlation in recreating strategies from previously recorded data.
... Several papers have also explored the use of PLG for physicsbased puzzle games such as Cut the Rope [15], [16] and, more notably for this paper, Angry Birds [17], [18], [19], [20], [21], [22], [23]. The physics constraints employed in these types of games, along with the exceptionally large state and action spaces, create many problems for PLG [24]. ...
This paper presents an overview of the second AIBIRDS level generation competition, held jointly at the 2017 IEEE Conference on Computational Intelligence and Games, and the 26th International Joint Conference on Artificial Intelligence. This competition tasked entrants with developing a level generator for the physics-based puzzle game Angry Birds. Submitted generators were required to deal with many physical reasoning constraints caused by the realistic nature of the game's environment, in addition to ensuring that the created levels were fun, challenging and solvable. This year's competition was a significant improvement over the previous year, with a greater number of participants and more advanced generators. Within this paper we describe the framework, rules, submitted generators and results for this competition. We also provide some background information on related research and other video game AI competitions, as well as discussing what can be learned from this year's competition. There are several game and real-world applications for this type of research, and we provide some examples of the types of levels we would like future competition entries to generate.
... The physics constraints employed in these types of games create many problems for PLG and make evaluating the quality of levels difficult. The playability/solvability of generated levels is particularly difficult to confirm, due to the exceptionally large state and action spaces [12]. Although the proposed generator is designed specifically for the Angry Birds elements and environment, the techniques used can be applied to many other games which share similar mechanics and level designs. ...
... In this case, treating the game character as a point was an overapproximation that simplified search without introducing too many false paths. Shaker et al.'s editor for Cut the Rope levels used polyhedral over-approximations to show which parts of the stage were influenced by particular puzzle elements (essentially finding closed-form solutions to aspects of the puzzle) [9]. Smith et al.'s tools for the Refraction educational game [10], [11] use a tight over-abstraction of the game's core rules where the order in which puzzle pieces are placed does not matter, whereas the qualitative spatial relations between those remains important; this is key to keeping the encodings of puzzles and solutions small. ...
While general game playing is an active field of research, the learning of game design has tended to be either a secondary goal of such research or it has been solely the domain of humans. We propose a field of research, Automated Game Design Learning (AGDL), with the direct purpose of learning game designs directly through interaction with games in the mode that most people experience games: via play. We detail existing work that touches the edges of this field, describe current successful projects in AGDL and the theoretical foundations that enable them, point to promising applications enabled by AGDL, and discuss next steps for this exciting area of study. The key moves of AGDL are to use game programs as the ultimate source of truth about their own design, and to make these design properties available to other systems and avenues of inquiry.
... In spite of the research interest of PbSGs, generating content for physicsbased simulation games is an area that has been explored timidly and, as far as we know, only [6] uses Grammatical Evolution to automatically generate levels for a clone of Cut the Rope, a commercial physics-based puzzle game. ...
This paper presents a spatially-structured evolutionary algorithm (EA) to procedurally generate game maps of different levels of difficulty to be solved, in Gravityvolve!, a physics-based simulation videogame that we have implemented and which is inspired by the n-body problem, a classical problem in the field of physics and mathematics. The proposal consists of a steady-state EA whose population is partitioned into three groups according to the difficulty of the generated content (hard, medium or easy) which can be easily adapted to handle the automatic creation of content of diverse nature in other games. In addition, we present three fitness functions, based on multiple criteria (i.e., intersections, gravitational acceleration and simulations), that were used experimentally to conduct the search process for creating a database of maps with different difficulty in Gravityvolve!.
We present a survey of mixed-initiative methods for the creation of content for video games. We also propose a definition of what mixed-initiative implies, as the term lacks a clear specification. The survey includes works not directly aimed at video games but which create content that can potentially be used in games, such as art programs utilizing mixed-initiative. Furthermore, we highlight research areas that overlap wholly or partly with mixed-initiative, such as casual creators, explainable AI, or interactive evolutionary computation. We examine these and several other topics in the context of mixed-initiative. Finally, we provide a catalogue of typical techniques and challenges connected with mixed-initiative before considering future directions.
Building Artificial Intelligence (AI) that can successfully interact with the physical world in a comprehensive and human-like way is a big challenge. Physics simulation games, i.e., video games where the game world simulates real-world physics, offer a simplified and controlled environment for developing and testing Artificial Intelligence. It allows AI researchers to integrate different areas of AI, such as computer vision, machine learning, knowledge representation and reasoning, or automated planning in a realistic setting and to solve various problems that occur in the real world without having to consider all of its complexity at once. This chapter first outlines the main categories of physics simulation games, some of which have become increasingly popular in recent years with the widespread availability of handheld touchscreen devices. It then discusses the motivation and rationale for conducting Artificial Intelligence research on these games and highlights the main research goals. Some of the underlying AI problems and recent advances are discussed and exemplified using a popular physics simulation game. Finally, an overview of current research in related areas is given.
Artificial intelligence in digital games has developed in the last 40 years. It has a long and deep history with digital games. AI techniques in digital games evolved independently and differently from the academic AI research of science and engineering which require functionality in the real world. Digital games have complex and large-scale virtual 2D/3D worlds where game characters live in, recognize, make decisions, and design their motions to fitting their environment. The digital world of games is larger, more complex, and more detailed than any other virtual world. It is the most suitable experimental field to study and evaluate AI technologies in the virtual world. The AI for a game character is called “Character AI” and its function is for characters to make decisions. This is much different from functional AI in academic research, and making a character means to create one whole intelligence. The other unique AIs are “Navigation AI” which analyzes and recognizes the environment of game world and “Meta AI” which dynamically controls and changes the progress, situation, and drama of the game. These three AIs, namely, Character AI, Navigation AI, and Meta AI, cooperate with each other and develop one unified system to form a dynamic user experience. In addition, recently learning and evolution approaches have been introduced into AI for digital games. In this chapter such current status of AI in digital games is described.
Building Artificial Intelligence (AI) that can successfully interact with the physical world in a comprehensive and human-like way is a big challenge. Physics simulation games, i.e., video games where the game world simulates real-world physics, offer a simplified and controlled environment for developing and testing Artificial Intelligence. It allows AI researchers to integrate different areas of AI, such as computer vision, machine learning, knowledge representation and reasoning, or automated planning in a realistic setting and to solve various problems that occur in the real world without having to consider all of its complexity at once. This chapter first outlines the main categories of physics simulation games, some of which have become increasingly popular in recent years with the widespread availability of handheld touchscreen devices. It then discusses the motivation and rationale for conducting Artificial Intelligence research on these games and highlights the main research goals. Some of the underlying AI problems and recent advances are discussed and exemplified using a popular physics simulation game. Finally, an overview of current research in related areas is given.
While there exist a variety of game description languages (GDLs) for modeling various classes of games, these are aimed at game playing rather than the more particular needs of game design. This paper describes a new approach to general game modeling that arose from this need. A class grammar is automatically generated from a given library of source code, from the constructors and associated parameters found along its class hierarchy, to give a context-free grammar that provides access to the underlying code while hiding its implementation details.
Algorithms can generate game content, but so can humans. And while PCG algorithms can generate some kinds of game content remarkably well and extremely quickly, some other types (and aspects) of game content are still best made by humans. Can we combine the advantages of procedural generation and human creation somehow? This chapter discusses mixed-initiative systems for PCG, where both humans and software have agency and co-create content. A small taxonomy is presented of different ways in which humans and algorithms can collaborate, and then three mixed-initiative PCG systems are discussed in some detail: Tanagra, Sentient Sketchbook, and Ropossum.
This paper introduces an authoring tool for physics-based puzzle games that supports game designers through providing visual feedback about the space of interactions. The underlying algorithm accounts for the type and physical properties of the different game components. An area of influence, which identifies the possible space of interaction, is identified for each component. The influence areas of all components in a given design are then merged considering the components’ type and the context information. The tool can be used offline where complete designs are analyzed and the final interactive space is projected, and online where edits in the interactive space are projected on the canvas in realtime permitting continuous assistance for game designers and providing informative feedback about playability.
PCG approaches are commonly categorised as constructive, generate-and-test or search-based. Each of these approaches has itsdistinctive advantages and drawbacks. In this paper, we propose an approach to Content Generation (CG) – in particular level generation – that combines the advantages of constructive and search-based approaches thus providing a fast, flexible and reliable way of generating diverse content of high quality. In our framework, CG is seen from a new perspective which differentiates between two main aspects of the gameplay experience, namely the order of the in-game interactions and the associated level design. The framework first generates timelines following the search-based paradigm. Timelines are game-independent and they reflect the rhythmic feel of the levels. A progressive, constructive-based approach is then implemented to evaluate timelines by mapping them into level designs. The framework is applied for the generation of puzzles for the Cut the Rope game and the results in terms of performance, expressivity and controllability are characterised and discussed.
In this paper, we present an approach for automated evaluation and generation of videogames made with PuzzleScript, a description-based scripting language for authoring games, which was created by game designer Stephen Lavelle [1]. We have developed a system that automatically discovers solutions for a multitude of videogames that each possess different game mechanics, rules, level designs, and win conditions. In our approach, we first developed a set of general level state heuristics, which estimates how close a given game level is to being solved. It is used to adapt the best-first search algorithm to implement a general evaluation approach for PuzzleScript games called GEBestFS. Next, we developed an evolutionary framework that automatically generates novel game mechanics from scratch by evolving game design rulesets and evaluating them using GEBestFS. This was achieved by developing a set of general ruleset heuristics to assess the playability of a game based on its game mechanics. From the results of our approach, we showcase that a description-based language enables the development of general methods for automatically evaluating games authored with it. Additionally, we illustrate how an evolutionary approach can be used together with these methods to to automatically design alternate or novel game mechanics for authored games.
Evolutionary Design by Computers is a collection of essays that describe recent research into "evolutionary" computing where computers mimic the strategies of biological evolution to solve problems in architecture, engineering, art and artificial life. Peter Bentley's excellent introduction to the current state of evolutionary design quickly directs readers to the field. Bentley shows that no matter how various practitioners identify themselves they are united in applying the principles of Darwinism to computer algorithms.
The collection includes several theoretical essays that discuss the relationship of computer-driven design to human innovation. A section on evolutionary designs features several case studies on real applications of these techniques--specifically engineering problems for designing satellite booms, flywheels and a reliability measurement for networks.
Among the contributions are essays on computer-art packages that make use of evolutionary algorithms. Programs such as Mutator and Forms show how anyone can use these evolutionary techniques. This discussion includes a survey of today's evolutionary art (evoart) packages-- including several that are available on the Web.
The last part of the book covers artificial life It showcases a programme that evolves simple block-like creatures that walk, swim, jump and even compete with each other--a programme that has obvious applications for robotics. The final sections examine additional real-world applications of evolutionary design techniques for architecture (for designing tables and hospital floor plans) and electrical engineering (analogue circuits in particular).
All in all Evolutionary Design by Computers provides an excellent introduction to one of today's most promising areas of computer-science research for both specialists and general readers alike.
We are delighted to announce the release of GEVA [1], an open source software implementation of Grammatical Evolution (GE) in Java. Grammatical Evolution in Java (GEVA) was developed at UCD's Natural Computing Research & Applications group (http://ncra.ucd.ie).
This paper presents the use of design grammars to evolve playable 2D platform levels through grammatical evolution (GE). Representing levels using design grammars allows simple encoding of important level design constraints, and allows remarkably compact descriptions of large spaces of levels. The expressive range of the GE-based level generator is analyzed and quantitatively compared to other feature-based and the original level generators by means of aesthetic and similarity based measures. The analysis reveals strengths and shortcomings of each generator and provides a general framework for comparing content generated by different generators. The approach presented can be used as an assistive tool by game designers to compare and analyze generators' capabilities within the same game genre.
Adapting game content to a particular player's needs and expertise constitutes an important aspect in game design. Most research in this direction has focused on adapting game difficulty to keep the player engaged in the game. Dynamic difficulty adjustment, however, focuses on one aspect of the gameplay experience by adjusting the content to increase or decrease perceived challenge. In this paper, we introduce a method for automatic level generation for the platform game Super Mario Bros using grammatical evolution. The grammatical evolution-based level generator is used to generate player-adapted content by employing an adaptation mechanism as a fitness function in grammatical evolution to optimize the player experience of three emotional states: engagement, frustration and challenge. The fitness functions used are models of player experience constructed in our previous work from crowd-sourced gameplay data collected from over 1500 game sessions. Copyright © 2012, Association for the Advancement of Artificial Intelligence.
This paper explores a method for analyzing the expressive range of a procedural level generator, and applies this method to Launchpad, a level generator for 2D platformers. Instead of focusing on the number of levels that can be created or the amount of time it takes to create them, we instead examine the variety of generated levels and the impact of changing input parameters. With the rise in the popularity of PCG, it is important to be able to fairly evaluate and compare different generation techniques within similar domains. We have found that such analysis can also expose unexpected bi-ases in the generation algorithm and holes in the expressive range that drive future work.
This study evolves and categorises a population of conceptual designs by their ability to handle physical constraints. The
design process involves a trade-off between form and function. The aesthetic considerations of the designer are constrained
by physical considerations and material cost. In previous work, we developed a design grammar capable of evolving aesthetically
pleasing designs through the use of an interactive evolutionary algorithm. This work implements a fitness function capable
of applying engineering objectives to automatically evaluate designs and, in turn, reduce the search space that is presented
to the user.
This paper presents a first attempt at evolving the rules for a game. In contrast to almost every other paper that applies computational intelligence techniques to games, we are not generating behaviours, strategies or environments for any particular game; we are starting without a game and generating the game itself. We explain the rationale for doing this and survey the theories of entertainment and curiosity that underly our fitness function, and present the details of a simple proof-of-concept experiment.
It is easy to create new combinatorial games but more difficult to predict those that will interest human players. We examine the concept of game quality, its automated measurement through self-play simulations, and its use in the evolutionary search for new high-quality games. A general game system called Ludi is described and experiments conducted to test its ability to synthesize and evaluate new games. Results demonstrate the validity of the approach through the automated creation of novel, interesting, and publishable games.
Video game content includes the levels, models, items, weapons, and other objects encountered and wielded by players during the game. In most modern video games, the set of content shipped with the game is static and unchanging, or at best, randomized within a narrow set of parameters. However, ideally, if game content could be constantly renewed, players would remain engaged longer in the evolving stream of novel content. To realize this ambition, this paper introduces the content-generating neuroevolution of augmenting topologies (cgNEAT) algorithm, which automatically evolves game content based on player preferences, as the game is played. To demonstrate this approach, the galactic arms race (GAR) video game is also introduced. In GAR, players pilot space ships and fight enemies to acquire unique particle system weapons that are evolved by the game. As shown in this paper, players can discover a wide variety of content that is not only novel, but also based on and extended from previous content that they preferred in the past. The implication is that it is now possible to create games that generate their own content to satisfy players, potentially significantly reducing the cost of content creation and increasing the replay value of games.
We describe the first steps in the adoption of Shape Grammars with Grammatical Evolution for application in Evolutionary Design. Combining the concepts of Shape Grammars and Genetic Programming opens up the exciting possibility of truly generative design assist tools. In this initial study we provide some background on the adoption of grammar-based Genetic Programming for Evolutionary Design, describe Shape Grammars, and give a brief overview of Grammatical Evolution before detailing how Grammatical Evolution used Shape Grammars to successfully rediscover some benchmark target structures.
This paper addresses the problem of automatically constructing tracks tailor-made to maximize the enjoyment of individual players in a simple car racing game. To this end, some approaches to player modeling are investigated, and a method of using evolutionary algorithms to construct racing tracks is presented. A simple player-dependent metric of entertainment is proposed and used as the fitness function when evolving tracks. We conclude that accurate player modeling poses some significant challenges, but track evolution works well given the right track representation.
One of the applications of evolutionary algorithms is the automatic creation of designs. For evolutionary techniques to scale to the complexities necessary for actual engineering problems, it has been argued that generative systems, where the genotype is an algorithm for constructing the final design, should be used as the encoding. We describe a system for creating generative specifications by combining Lindenmayer systems with evolutionary algorithms and apply it to the problem of generating table designs. Designs evolved by our system reach an order of magnitude more parts than previous generative systems. Comparing it against a non-generative encoding we find that the generative system produces designs with higher fitness and is faster than the non-generative system. Finally, we demonstrate the ability of our system to go from design to manufacture by constructing evolved table designs using rapid prototyping equipment. 1 Introduction Evolutionary algorithms (EAs) have been succe...
We present a system for generating complete game designs by evolving rulesets, character layouts and terrain maps in an orchestrated way. In contrast to existing approaches to generate such game components in isolation, our ANGELINA system develops game components in unison with an appreciation for their interrelatedness. We describe this multi-faceted evolutionary approach, and give some results from a first round of experimentation.
This chapter describes Grammatical Evolution (GE) in detail (Ryan et al., 1998; O’Neill and Ryan, 2001; O’Neill, 2001). We show that it is an evolutionary algorithm (EA) that can evolve complete programs in an arbitrary language using a variable-length binary string. The binary genome determines which production rules in a Backus Naur Form (BNF) grammar definition are used in a genotype-to-phenotype mapping process to a program. GE is set up such that the evolutionary algorithm is independent of the output programs by virtue of the genotype-phenotype mapping, allowing GE to take advantage of advances in EA research. The BNF grammar, like the EA, is a plug-in component of the system that determines the syntax and language of the output code, hence, it is possible to evolve programs in an arbitrary language.
Grammatical Evolution (GE) is a grammar based GA to generate computer programs which has been shown to be comparable with GP when applied to a diverse array of problems. GE has the distinction that its input is a BNF, which permits it to generate programs in any language, of arbitrary complexity, including loops, multiple line functions etc. Part of the power of GE is that it is closer to natural DNA than GP, and thus can benefit from natural phenomena such as a separation of search and solution spaces through a genotype to phenotype mapping, and a genetic code degeneracy which can give rise to silent mutations (Mutations that have no effect on the phenotype). We have previously shown how runs of GE are competitive with GP, and in this paper we analyse characteristics such as genotypic diversity, and individual genotypic length, in an attempt to shed light on the power of the system. Results indicate that GE can use certain features of the system to its benefit ...
Genetic Programming IV
- J Koza
- M Keane
- M Streeter
- W Mydlowec
- J Yu
- G Lanza
J. Koza, M. Keane, M. Streeter, W. Mydlowec, J. Yu, and G. Lanza,
Genetic programming IV. Kluwer Academic Publishers, 2003.
- Maxis
Maxis, 2008, spore, Electronic Arts.
Making racing fun through player modeling and track evolution
- J Togelius
- R D Nardi
- S M Lucas
J. Togelius, R. D. Nardi, and S. M. Lucas, "Making racing fun through
player modeling and track evolution," in Proceedings of the SAB'06
Workshop on Adaptive Approaches for Optimizing Player Satisfaction
in Computer and Physical Games, 2006.
The advantages of generative grammatical encodings for physical design
- G Hornby
- J Pollack
G. Hornby and J. Pollack, "The advantages of generative grammatical
encodings for physical design," in Proceedings of the 2001 Congress
on Evolutionary Computation, vol. 1. IEEE, 2001, pp. 600-607.
Analyzing the expressive range of a level generator
- G Smith
- J Whitehead
G. Smith and J. Whitehead, "Analyzing the expressive range of a
level generator," in Proceedings of the 2010 Workshop on Procedural
Content Generation in Games. ACM, 2010, p. 4.
Spore, Electronic Arts
- Maxis