The Research Of Non-player Characters Adaptive Behavior In Digital Game

Non-player character to the game is provided by the program controlled character. It is animportant part of the virtual game world, which has been consistently throughout the game plot.So it has always been the focus of attention of the players, their design greatly influence thesuccess of the game. In order to present a complete interactive narrative games, players for thenon-player characters evoke cognitive and emotional attitude, designers often use to control thebehavior of non-player characters to achieve information exchange mechanism between gamecharacters. Therefore, the study of the behavior of non-player characters by the attention thegame designer’s attention.In this research, through the retrieval and analysis of relevant literature, most of thenon-player character’s behavior is based on random decisions or predefined behavior traditionalpolicy-making decisions in digital games. It is possible to approximate the behavior of a clearunderstanding of the role of non-player mode, which greatly reduces the player’s participation, sothat players lose interest in continuing to experience the game. Therefore, this study combinedwith relevant theoretical research, the focus of the study, namely, whether the non-playercharacters have human thought patterns, making it a virtual game environment by sensingchanges in the details of their own state, the development of behavioral decision making andplanning guide their adaptive behavior, thereby increasing the real experience of the game, theplayer firmly attracted attention.In view of this, This research adopted literature analysis, case analysis and systemdevelopment method，used the existing game development tools, which design and developmenta non-player characters have adaptive behavior digital games. The main research work is asfollows.(1)After a review of the relevant literature and analysis of domestic and internationalresearch, relevant concepts and theories in this research involved a detailed study and elaborate,to lay a theoretical foundation for the development of the construction and subsequent instances of the game theory model.(2)Referring to the case of domestic and foreign game design and analysis, summed up theideas and principles of construction of the theoretical model, and proposed to build interrelatedelements of the theoretical model and between the various elements, and ultimately build a gamewith a non-player characters adaptive behavior overall theoretical model to refine the structure ofnon-player character design system adaptive behavior.(3) Combined with the specific subject knowledge, detailed instructional design. On thisbasis, the use of reinforcement learning techniques to achieve the adaptive behavior ofnon-player characters, develop a detailed idea of the functional modules. Including the role ofthe state space and the behavior of each space, said different states-Action for correction valuesand how those values based on a reasonable choice of behavior.(4)In view of the above proposed solution ideas, use of relevant game softwaredevelopment platform, designed to develop specific game cases. And operating results based onthe game, made after intensive study over the NPC with non-related experiments strengthenlearning and traditional learning and training to strengthen the effectiveness of the NPCintelligence.With the above research work gradually developed in this study achieved the followingthree significant research results.(1) This research construct a theoretical model of the whole game with non-playercharacters adaptive behavior, and combined junior high "Moral Education" of the disciplinesystem for detailed instructional design, on the basis of analysis of how the adaptive behavior ofnon-player characters in digital games, such as the representation of state space and action spaceof roles in the program, feedback incentive to develop signal different States-Action correctionvalues, how to perform in accordance with these values, and how to choose the appropriatebehavior of these behaviors.(2) This research use Eclipse software development platform to achieve the design ofperception subsystem, selection subsystem behavior and action execution subsystem. Eventuallyit developed a product called "Police Run" digitized simulation game, and presented in thedigital gaming site this carrier.(3) Analysis combined with non-player characters in the game instances running adaptive behavior effect, to make the findings of the experiment, compare the difference of reinforcementlearning over after NPC intelligent effect with non-reinforcement learning and traditionalreinforcement learning.Due to the limited time and technology development capability, the digital game design ofnon-player characters adaptive behavior can reflect the original intentions and ideas of theresearch, but there are still many problems and shortcomings. Future research will focus on howwe can be better the design and development of adaptive behavior of non-player characters.