Video Games and Artificial Intelligence - Research Paper Example

?Your Your 2 December Video Games and Artificial Intelligence Jones provides us with an excellent and explicit comparison between human intelligence and animal intelligence, he states that even though animals can perform some intelligence driven actions such as communicating and solving simpler problems but humans on the other hand can perform decision making for different scenarios by using many aspects of the intelligence endowed to them (2). He states that the human intelligence is far more equipped and complex and this concept can be utilized firmly in the presentation of an analogy between an artificially intelligent computer infrastructure and simple structure. He goes onto add that a powerful application based Chess game would definitely have an equally powerful chess database but would know nothing about any other board game, which is contrary to artificially intelligent gaming techniques (2). Artificial Intelligence (AI) and the gaming arena together, have been a sought after area of research since the discovery of AI. Even though, the main objective of the domain of AI which is of creating an interactive human like robot is far from the reality but the field has definitely revolutionized the world of video games, making them more life like and innovative ( Calero & Martin vii). Furthermore, the authors state that the video games provide an interactive real world environment to the player by depicting complex behavior, where perceptions are under total control, through the utilization of enhanced resolution of photo realization and state of the art multiprocessor architecture (vii). Before delving into the crux of AI and the components that drives this phenomenon, it would be a better idea to configure concept of what AI actually is and how it came about. Millington and Funge states that AI enables computers to perform task that human and animals are exclusively capable of for instances, tasks such as sorting, arithmetic etc are solved easily by a computer system but for it to perform tasks such as face recognition, decision making and language compatibilities needs AI to develop algorithms to give the instructions to the underlying hardware as to what to do next (4). They further add that these algorithms are developed using either using the philosophical notion by retracing the trail of human thoughts or by using the psychological technique, through the process of understanding the human brain and the processes therein. Thirdly, the algorithms can also be engineered by configuring them to perform human-like tasks (4). Millington and Fudge say that the AI in game is divided into three sections: Movement, Decision making and strategy (4). Shedding light into the movement segment of an AI model, gives us a clear indication that in most of the games except the combat ones, the movement is related to path finding. Jones explain that in most of the games integrating the path finding strategy, the main objective is to find a path between an arbitrary point A to another point B. In many game, multiple pathways exist between the two required points so constraints such as finding the shortest path or least cost exists (123). The authors illustrate this point by laying out a scenario in which multiple paths exists by considering two points separated by a hill, it is faster to take the path that round the hill that the one that goes up it, however it may prove to be an advantage for the player to go up the hill in terms of general progress of the game (123). Path finding, at certain times is actually searching for the shortest path that exists in a graph, in terms of games, a graph is a particular land route a gamer has to pass through. It consists of edges and nodes. Nodes are the points A and B that we discussed above and the path joining them are called edges, most of the time, these edges are weighted, giving them a measure of the distance that they cover. In a graph we have multiple nodes and edges and to find a shortest path between two particular nodes, algorithms have been developed. Algorithm such as Dijkstra, Depth first Search (DFS), Breadth First Search (BFS) and A* search algorithm are existent and each of these have different mechanism of iteration but amongst the ones mentioned above A* is considered to be optimal. On the contrary, Jones suggest that A* algorithm is very computer intensive and with the increasing number of agents and with their respective path finding processes occurring simultaneously, the iteration of the A* algorithm would be multiplied and since AI supports enhanced level of gaming strategies, the path finding component of the game needs to be optimized (123). Calero and Martin further places emphasis on the point that in AI supported video gaming, real time processing is taking place and a limited amount of per-frame CPU time is shared by multiple agents running the game and moreover, with each generation, the gaming environment is progressively moving towards greater complexity of simulation and environment. The authors introduce a much complex and compatible real time heuristic search algorithm, which comfortably replaces the A* search algorithm and simultaneously satisfy the maximum amount of planning per move independent of the size of the problem (1). They explain the mechanism behind the real time algorithm as one that computes a solution to a problem in a step-by-step manner contrary to other orthodox search algorithms, which configure an abstract solution even before the first action is attempted. The real time algorithm work in a way where it plans only a first few actions for the agent to take by carrying out a look-ahead depth search of the agents’ present state and along with the heuristic assistance, it allows the agent to take predetermined actions and then the planning-execution cycle repeats (2). However, in this algorithm, there is lies danger of the player ending up at a dead end since the goal state is not visible, hence in order to rectify this shortcoming, experts have formatted the working of the algorithm in a way that the desired response is achieved by using different attributes of the search mechanisms (2). The other portion of movement in AI endorsed gaming is the combat style as mentioned earlier. A fighting game is the simulation of real time, hand to hand combat style designed for two players but mostly only one player game is preferred as the computer takes up the role of the component. However, two player games can be played as well, but it is seen in some cases that user wants to play against the machine; this is due to the fact that the AI computing used in these fighting games is of a simple design e.g. Finite State Machines which is unable to interpret and adapt to the user’s pattern (Zhang 649). The author further clarifies that in order for the agent to be more purposeful and versatile, the experts have divided agents into three sections: Main Subagent (MSA), Executing Combo Subagent (ECSA) and Receiving Combo Subagent (RCSA), however it is notable to mention that all learning processes are delayed till one round of the game (650). Zhang elaborates the role of MSA as the component that executes simple attacks, blocks and gestures, at times, it even transmits the control to other subagents. This subagent records the environment, receives a stimulus called reward from it and bases it executes subsequent action on these rewards. This propels the subagent to behave in a way that is unpredictable to the user and retains their interest in the game (650). On the other hand, ECSA deals with the combos, their selection and their execution, combos are basically the combination of simple attacks that accrues significant damage. For the agent to exhibit the same level as the user, the combos that ECSA enact should be at the same levels as the user, and the techniques employed in the creation of such combos are: Ratios of user combos, indistinguishability of combos, and entropy of combo breakers. These mechanisms have their own exclusive working and mathematical iterations, which adds to the interactive ness of the game as a whole (651). Lastly Zhang explains the role of RCSA as a segment that mines the patterns of the combos executed by the user after each round and when called, it matches the mined patterns and the pattern of combo called up by the user to predict the future combos that would be needed (652). Shedding light into another aspect of AI in video games which is the strategy being implemented in it. Jones illustrates the fact that in the beginning of a real time strategy game, the focus is on maintaining the social and military infrastructure, which involves the procedure of creating new citizens who would develop the economy, and when it is fully developed, the society will be able to launch an attack and defeat the opponents (136). He further adds that these conditions are built-in in the internal structure of the game but for enhanced flexibility, it could be executed separately so that there remains a room for modification if need arises. To implement the above procedure Rule Based Learning (RBL) was used in the past (136). Tim explains RBL as an effective tool to embed the expert knowledge of game play into the gaming strategy which utilizes two sets of memory, one that stores the facts and the other that stores a set of rules, once a rule matching algorithm is implemented upon the facts, the rules that matches the facts are then stored in the conflict set. This conflict set is then scrutinized and one of the rules is chosen to fire through a process called conflict resolution, the rule is then executed and the working memory is updated (137). Jones further explains the method of effective decision-making in AI is done through explicit representation of knowledge so that the game engine makes sound decision by extracting this database of knowledge through its environment. Jones illustrates this fact by describing a scenario where user repeatedly tries to sign in to a secure address and this gets interpreted as a measure to hack the device and the fact that this activity can be monitored elsewhere as well (144). Semantic network is a method of knowledge representation widely popular in the domain of AI. It handles not only binary relations but also higher order relation making the network an extremely powerful tool (Bourbakis 517). Bourbakis reveals that the network handles knowledge with respect to concepts and hierarchical relationships between them; nodes are represented by concepts and relationships with edges. The framework of the network and the pathway in which knowledge travels is depicted below http://www.google.com.pk/imgres?q=semantic+networks&hl=en&biw=1280&bih=656&gbv=2&tbm=isch&tbnid=6TJA-SfioMov4M:&imgrefurl=http://marinemetadata.org/node/2086&docid=AlicCTmev-MZLM&imgurl=http://marinemetadata.org/files/mmi/images/semanticnetwork.png&w=488&h=304&ei=vpvbTubbJYTUsgbGz8yIDA&zoom=1&iact=hc&vpx=514&vpy=314&dur=6657&hovh=177&hovw=285&tx=201&ty=99&sig=112115215634239398747&page=1&tbnh=121&tbnw=195&start=0&ndsp=18&ved=1t:429,r:8,s:0 In this way, relationships between a large number of objects are represented at a wide scale (Jones 146) Another method of representation are frames which are a much more advanced version of semantic networks is utilized in which more object oriented abstraction takes place and a much more evolved structure is seem (Jones 146). Jones further elaborates frames as having two kinds: genetic frame and instance, the former define a class of objects while the latter is an instance of generic frames used to transfer knowledge and information between slots within the structure (146). Conventionally, Boolean logic is thought of as a way in which only two discreet outputs are exhibited, either its hold or cold, no values lying in between are considered to be true. However in the case of fuzzy logic, the values lying between the two extremes will also be accounted for in the measurements. Seemann and Bourg explain this fact in a translucent way; they describe something true to the degree one in fuzzy logic as being absolutely true whereas if its degree zeros in fuzzy logic then its absolutely false, also having values between 0 and 1 are declared legal by the object (189). They further add that fuzzy logics have a wide variety of applications including their effective integration in video games, fuzzy controls can also be used to navigate gaming units such as land vehicles, aircraft and ground routes, as well as in evading risky situations and in successfully maneuvering through land marks (190). The authors describes a gaming scenario in which the player had to to change his direction of travel and move towards a different direction, which may be enemy headquarters, some treasure or home, in this case fuzzy logic is utilized to bring about smooth steering. Fuzzy logic calculates the precise degree of turn and its directions; the angle as well as the appropriate axis diversion that is needed to provoke the change desired by the player (191). Threat assessment is also one aspect of gaming in which fuzzy logic is employed. Fuzzy systems are used to evaluate the level of threat that is posed by the enemy in terms of troop size and range of enemy’s forces when talking in terms of battle simulation gaming (Seemann and Bourg 191). Fuzzy logic are also a great help in classifying a player and a non player in terms of his/her strength, weapon proficiency, hit points and armor class or any other factors of your choice. The fuzzy system generates a numerical score representing rankings in an intuitive style with lesser rules and without the need of training the system again and again (Seemann and Bourg 192). Harris describes fuzzy logic as a technique that is not that directly implemented as its conventional programming counterparts are, in other word he describes fuzzy logic as a phenomna that implements different combinations if inputs, for instance if something happens then maybe something will follow it, if a player chooses to implement an action, then a set of another events might take place simultaneously (20). He reveals that in a Mario Bros game, when Mario kicks the Shellcreeper, fuzzy logic allows the player to kick it anywhere near Shellcreeper to flip it over. He also mention The Sims, a game in which players can create characters and are responsible for their welfare by making them eat, go to work and clean the house etc. In this game the AI programming is all about fuzzy logic for instance Harris describes that in the case when the Sim is not fed over a long period of time, there is not just a mere yes or no to this problem, the player has to figure out what has gone wrong (21). The following figure is an example of how varying the inputs of a fuzzy logic can be and how it interprets them. http://www.google.com.pk/imgres?q=fuzzy+=1&hl=en&gbv=2&biw=1280&bih=656&tbm=isch&tbnid=9OIagjGLhpFYeM:&imgrefurl=http://www.mathworks.com/products/fuzzylogic/index.html&docid=4ex3xUaHCmvbxM&imgurl=http://www.mathworks.com/cmsimages/40305_wl_fl_mainimage_wl_3248.gif&w=367&h=295&ei=213cTqOeHPH24QTLkOCBDg&zoom=1&iact=hc&vpx=189&vpy=186&dur=546&hovh=126&hovw=172&tx=119&ty=96&sig=112115215634239398747&page=1&tbnh=126&tbnw=172&start=0&ndsp=18&ved=1t:429,r:0,s:0 Now that the three core segments of an AI engine have been discussed, it will a wise idea to cast a glance as to how different components in an AI model are interfaced with each other to make it a functional device. Millington and Funge describes the model having exclusive infrastructures in two categories of the engine: a general mechanism of managing the AI behavior and a world-interfacing center, which will receive all the required inputs from the outside world (32). The authors further state that all the AI algorithms designed should be compatible with the AI model, also there must be an output module which transform the final output of the model into action and subsequently interfaces movement and animation controllers to the output module which for instance makes the player move in the x-plane steadily or deliver a punch to the opponent.It is an essential feature of the model (32). Millington and Funge further add that the algorithms configured for the game should be in perfect harmony with other components of the AI model, this would make sure that codes are being developed while the game is running, and the AI will be capable of replacing the placeholder behavior when these codes are ready to be executed (33). All this has to be thought of in advance of implementation, the authors also confirm that a good AI engine will aid in reducing reuse, debugging and development time, by formulating an exclusive AI for a specific character. This brings together different techniques, which are assembled in the right way for the given purpose (33). A crude representation of an AI model is represented below. http://www.google.com.pk/imgres?q=AI+model&hl=en&biw=1280&bih=656&gbv=2&tbm=isch&tbnid=JYu-LNTmRmYIJM:&imgrefurl=http://www.stonetrip.com/developer/doc/concept/controllers/ai&docid=nKSufsXyVZWjvM&imgurl=http://www.stonetrip.com/developer/doc/html/images/Concepts/AI1.jpg&w=608&h=446&ei=oWXcTsCoD-mk4gSE9PXVAw&zoom=1 Millington and Funge further elaborate the algorithms used in the AI engine by defining a central pool of algorithms, which are applicable on multiple characters. However, they state that these algorithms are written in script language rather than compiled codes that specify how a particular character is assembled and what techniques are required in this process. This data is written down using efficient toolchains, which makes certain that the experts can easily insert instructions devised for the game without manual help (33). The authors illustrate this affect by giving an example of steering behavior in which parameters are created and fed into the toolchain to bring about the desired action and does not change character wise (33). A sample toolchain is inserted below. http://www.google.com.pk/imgres?q=toolchains&hl=en&gbv=2&biw=1280&bih=656&tbm=isch&tbnid=95TcoIr3gksiiM:&imgrefurl=http://bytescrolls.blogspot.com/2008/12/account-on-open-source-summit-2008.html&docid=TcjyV9Ae2s60TM&imgurl=http://www.linuxjournal.com/files/linuxjournal.com/linuxjournal/articles/073/7355/7355f1.resized.png&w=1024&h=696&ei=PGvcTsnKMcP44QTy4uD9DQ&zoom=1&iact=hc&vpx=649&vpy=342&dur=6006&hovh=185&hovw=272&tx=155&ty=114&sig=112115215634239398747&page=6&tbnh=116&tbnw=171&start=89&ndsp=18&ved=1t:429,r:3,s:89 On the contrary, there are certain situations where characters have to implement special cases, in this case rule based learning that we have discussed above comes into play, where complicated matching rules are defined and when they are inserted in tool, they resemble program codes (Millington and Funge 34). The authors reveal that the level designers add these codes in the toolchain by transforming trigger, scripts and simple rule into programming language and the level editor change this data into one that can be fed into the output module of the AI engine (34). Millington and Funge then go onto explain how data travels from one part of the device to another and finally delivers the output, they explain that the data created in the tool, or in other words in the modeling and package design is prepared for use in the game. Once a level is loaded, the AI behavior is created for a character and loaded onto the game engine, during the gameplay, the game then calls the game engine which updates the AI behavior it has received in accordance with the information it has loaded through the world interface and finally gives the output to the game (34). Taking the discussion inside the gaming world, it is obvious that the games of the new generation in terms of simulations and narratives are much more complex and visuospatial (Tavinor 74). Tavinor further describes the internal ambience of the game to be both fictional and non-fictional, he discusses that the presence of a virtual camera in the game is a part of the nonfiction access to the fictional gaming arena. The first person camera style utilized by games such as Wolfenstein 3D and Doom places the camera where the players’ eye is ought to be in the digital world, in this case, the field of vision is restricted (75). However, third camera person is employed in combat and adventure games where the camera is fixed in the environment and moves with the player (75). AI has also played a significant role in narratives in video games. Neilson, Smith and Tosca discusses the fact that AI gives the storytelling a digital platform in which computer driven attributes such as procedural, participatory, spatial and encyclopedic nature comes into play (191). The following is the schematics of the underlying mechanism that is at work in narrative games. http://www.google.com.pk/imgres?q=AI+video+game+schematics&hl=en&gbv=2&biw=1280&bih=699&tbm=isch&tbnid=z6iMHqgmMrLYIM:&imgrefurl=http://altugi.wordpress.com/&docid=VncQYzq7Rwo6pM&imgurl=http://altugi.files.wordpress.com/2010/12/play2.jpg&w=1060&h=817&ei=7abcTok0jJQ6-cOVsw4&zoom=1&iact=hc&vpx=692&vpy=308&dur=581&hovh=197&hovw=256&tx=176&ty=45&sig=112115215634239398747&page=1&tbnh=142&tbnw=184&start=0&ndsp=17&ved=1t:429,r:14,s:0 Harris reveals that the games that we play today originated sometime during the 1970s with the early video games such as Pong, Galaga, Donkey Kong and Pac-Man etc. However the level of AI used in the games became more advanced and complex as the world stepped into the 21st century and a multi-million industry took shape out of the video gaming buzz (18). Harris further elaborates that as this industry evolved, the AI was not only used to configure game algorithms but also special affects were born out of the same phenomena by a technique known as computerized generated imagery (CGI) (19). CGI is a powerful tool for engineering special effects, background and characters. Harris explains that the animators started creating characters through the use of frames and then they subsequently added texture, color and movement to it, he also mentions that the gaming industry and Hollywood played an integral part in the development of AI (19). Nowadays the human-computer interaction (HCI) in the games domain have increasingly started to resemble the natural human to human interaction in terms of speech, vision and haptics etc and hence in order to give rise to accuracy and precision and to create value added gaming, additional inputs such as biosensors, motion tracking equipment and other sensors are being integrated into the gaming world (Lliadus, Vlahavas and Bramer 440). However, the authors point out in that in this vicious cycle of development, the process of introducing new games genre and innovative gaming opportunities is facing a detrimental setback (440). In order to undermine these shortcomings, the authors have talked about a platform known as PlayMancer, which is a work in progress and will be able to bridge the existing 3D games with the innovative modes and gaming interaction with the help of advanced configuration tools (440). This new platform has embedded in itself a series of open systems such as gaming engine, spoken dialog management system and open spoken interface systems which support multimodality, adaptability to contextual changes, to user’s need and to gaming tasks (Lliadus, Vlahavas and Bramer 442). The authors propose that the platform has the dynamic ability of generating task related data, interfacing of dialog and interactive 3D graphics object at the design phase. They state that the architecture of the platform consists of three major segments: Game Engine, Application Manager and the Multimodal Dialogue Manager. http://www.playmancer.com/index.php/games-gallery The picture inserted above is the prototype of a game that puts PlayMancer to use, this game provides a view form within the ship where the player begins the game and also has a dynamic view of all the achievements and awards he has gathered so far. (www.playmancer.com). Ciancarini, Nakatsu and Rauterberg argues that even though the AI endorsed games are becoming increasingly more developed and complex, but even today, the AI of almost all the games, are based on a finite sequence of events/reactions that are easily predicted by expert players and soon become uninteresting for them (1). They are of the opinion that machine-learning techniques should be employed to learn the player’s behavior and subsequently adapt to the game’s AI, in this way the game would become more thrilling and the fun would last longer. They state that the ideal game here would have limited information of the game’s current situation (1). For instance, Ghost is a simple board game that is played by two opponents, to them the position of the game pieces on the board is visible, but they cannot see the type of opponent ones, given this information, different strategies can be adopted, either the player can run away from the opponent’s piece or attack it, hence inorder to win, the player must extract information from the opponents’ behavior with keeping in mind the fact that different players adopt different strategies (2). Lastly to infer, AI have become a source of great global development when it comes to graphical user interface and have brought the animations of the virtual in close proximity with the natural existent environment. The 3D technology today could not even be dreamt of in the earlier bits of the 20th century, the accuracy and the precision of mimicry of the natural environment leaves the person awe-struck and ready for more. The gaming industry has transformed into multi-million dollar venture due to the incorporation of AI and its components into the gaming features. This has not only evolved the gaming world but also Hollywood whose movies now represent the miracles of AI and the incredible human-computer interface it provides. Works Cited Bourg, David and Glenn Seemann. Al for game developers. USA: O’Reilly Media, Inc. , 2004. Print. Bourbakis, Nicholas. Artificial Intelligence methods and applications. USA: World Scientific, 1992.Print Calero, Pedro Antonio and Marco Antonio Gomez Martin. Artificial Intelligence for Computer Games. New York: Springer, 2011. Print. Harris, Micheal. Artificial Intelligence. New York: Marshall Cavendish, 2010. Print. Jones, Tim. Artificial Intelligence: A Systems Approach. Canada: Jones and Bartlett Learning, 2008.Print. Lliadis, Lazaros, Vlahavas Ioannis and Max Bramer. Artificial Intelligence Applications and Innovations: Proceedings of the 5th IFIP Conference on Artificial Intelligence Applications and Innovations (AIAI'2009), April 23-25, 2009, Thessaloniki, Greece. New York: Springer, 2009. Print. Millington, Ian and John Funge. Artificial Intelligence for games. USA: Morgan Kaugmann, 2009. Print. Nielson, Simon, Smith, Heide and Pajares Tosca. Understanding video games: the essential introduction. New York: Taylor and Francis, 2008. Print. Tavinor, Grant. The art of video games. Singapore: John Wiley and Sons, 2009. Print Zhang, Byoung-Tak. PRICAI 2010: Trends in Artificial Intelligence: 11th Pacific Rim International Conference on Artificial Intelligence, Daegu, Korea, August 30- September 2, 2010. Proceedings. Germany: Springer, 2010.Print. Google Images. Diagram of semantic network. Available at: http://www.google.com.pk/imgres?q=semantic+networks&hl=en&biw=1280&bih=656&gbv=2&tbm=isch&tbnid=6TJA-SfioMov4M:&imgrefurl=http://marinemetadata.org/node/2086&docid=AlicCTmev-MZLM&imgurl=http://marinemetadata.org/files/mmi/images/semanticnetwork [accessed on 5th December 2011]. Google Images. Fuzzy Logic Tool Box. Available at: http://www.google.com.pk/imgres?q=fuzzy+logic&um=1&hl=en&gbv=2&biw=1280&bih=656&tbm=isch&tbnid=9OIagjGLhpFYeM:&imgrefurl=http://www.mathworks.com/products/fuzzy-logic/index.html&docid=4ex3xUaHCmvbxM&imgurl=http://www.mathworks.com/cmsimages/403 [accessed on 5th December 2011. Google Images. AI Model. Available at http://www.google.com.pk/imgres?q=AI+model&hl=en&biw=1280&bih=656&gbv=2&tbm=isch&tbnid=JYu-LNTmRmYIJM:&imgrefurl=http://www.stonetrip.com/developer/doc/concept/controllers/ai&docid=nKSufsXyVZWjvM&imgurl=http://www.stonetrip.com/developer/doc/ht [accessed on 5th December 2011]. Google Images. Toolchains. Available at: http://www.google.com.pk/imgres?q=toolchains&hl=en&gbv=2&biw=1280&bih=656&tbm=isch&tbnid=95TcoIr3gksiiM:&imgrefurl=http://bytescrolls.blogspot.com/2008/12/account-on-open-source-summit-2008.html&docid=TcjyV9Ae2s60TM&imgurl=http://www.linuxjourn [accessed on 5th December 2011]. Google Images. The video game medium frames. Available at: http://www.google.com.pk/imgres?q=AI+video+game+schematics&hl=en&gbv=2&biw=1280&bih=699&tbm=isch&tbnid=z6iMHqgmMrLYIM:&imgrefurl=http://altugi.wordpress.com/&docid=VncQYzq7Rwo6pM&imgurl=http://altugi.files.wordpress.com/2010/12/play2.jpg&w=1060 [accessed on 5th December 2011]. www.playmancer.com. Prototype of games. Available at: http://www.playmancer.com/index.php/games-gallery [accessed on 5th December 2011]. Read More