KR100586088B1 - Agent NPC's ability enhancing system and its method - Google Patents

Abstract

The present invention relates to an agent NPC capability improvement system and a method for improving an agent NPC based on a game style of a specific gamer to temporarily play a game style of a better gamer in a specific game state. The agent NPC capability improvement system according to the present invention includes a game state transition generation engine for generating game state transition odds information for each gamer composed of game state transition paths, odds and style vectors of individual gamers from game information progressed by test gamers; An optimal game style extraction engine configured to extract the highest win rate information for each game state transition based on the game state transition win rate information for each gamer, and extract the optimum game style information to be applied by comparing with the win rate information for the owner gamers; And an optimal game style application decision engine that extracts the optimal style information of the game state that matches the current game state progressed by the agent NPC from the application target optimal game style information, and transplants the style vector suggested by the optimal odds information. do. According to the present invention, the gamers learn the optimal game style for each game state transition path and present the optimal game style information in a part of the game state of the agent NPC to temporarily strengthen the agent NPC's ability of the master gamer.

Game, AI, Gamer, Optimal, Odds, Learning, Transfer, Style, Transplant, NPC

Description

Agent NPC's ability enhancing system and its method, in which the game style of a specific gamer is implanted

The following drawings attached to this specification are illustrative of preferred embodiments of the present invention, and together with the detailed description of the invention to serve to further understand the technical spirit of the present invention, the present invention is a matter described in such drawings It should not be construed as limited to

1 is a conceptual diagram of a learning model modeled on gamers constituting the prior art;

Figure 2 is an illustration of a style vector, weight vector, output vector model constituting the prior art.

3 is a relationship diagram between a style vector, a weight vector, and an output vector constituting the prior art;

4 is an environmental conceptual diagram of an agent NPC capability improvement system constituting the technique of the present invention;

5 is a conceptual diagram of a game state transition path constituting the technique of the present invention.

6 is a block diagram of an agent NPC capability improvement system according to an embodiment of the present invention.

7 is a block diagram illustrating an execution engine of the capability improving system according to an exemplary embodiment of the present invention.

8 is a flowchart of a method for improving agent NPC capability according to an embodiment of the present invention.

9 is a flowchart of a game state transition generation process according to an embodiment of the present invention.

10 is a flow chart of the optimal game style extraction process according to an embodiment of the present invention.

11 is a flowchart illustrating a process of applying an optimal game style according to an embodiment of the present invention.

12 is a detailed flowchart of an optimal game style application process according to an embodiment of the present invention.

Explanation of symbols on the main parts of the drawings

1: ability improvement system 2: game state transition generation engine

3: Optimal Game Style Extraction Engine 4: Optimal Game Style Extraction Engine

21: game state transition generation module 23: game state transition storage module

31: cumulative game state module 33: the optimal winning rate extraction

36: Optimal Game Style Extraction Module

41: memory loading module 42: game state extraction module

43: Optimal Game Style Utilization Module 100: Game Operating System

200: conventional game performance system

The present invention relates to a system and method for improving the capability of an agent NPC in which a game style of a specific game player is implanted. More specifically, the game style is temporarily introduced in a specific game state in which the agent NPC encounters a game. The present invention relates to a capability improvement system of an agent NPC and a method for improving the capability by improving the game performance.

In the early games, the game progress was simple, and when the gamers repeated the game, the game pattern was found to be repeated, the gamers learned the repetitive rules, predicted the progress of the game, and formulated game manipulation in the prediction situation. Played the game.

In recent years, the development of online games, mud games, RPG games, and the like, breaking away from the aforementioned monotonous game progress and providing various game characters and game situations, more dynamic games are being introduced.

More dynamic games require the control of a variety of characters and game situations, which are effectively handled by artificial intelligence processing techniques that perform highly complex decision making and present highly optimized solutions. Is implemented. In addition, beyond the traditional computer-to-gamer and gamer-to-gamer form, AI plays games with players on behalf of the computer, or NPCs (Agent NPCs) playing games with other players on behalf of the owner. The intelligent game of technology is evolving into a futuristic game.

The term NPC (non player character) used in the present invention refers to a character manipulated by a computer (game operating system). For example, an NPC controlled by the computer by designating that the opponent or the gamer of the gamer manipulated by the computer substitutes for itself. Among them, the characters manipulated by computers on behalf of the owner gamers are called agent NPCs.

The present invention relates to a technique for controlling an agent NPC, and the present applicant has proposed a model for transplanting the game style of the owner gamer to the agent NPC and proceeding the game based on the owner's game style in which the agent NPC is implanted. .

As a prior art of the applicant, Patent Application No. 2004-27918 introduces a transplantation system and a transplantation method for temporarily applying a game style of a transplant target gamer (master gamer) to an agent NPC of a game operating system. Hereinafter, the basic technique of the prior art based on this invention is demonstrated.

The 'game style implantation system' constituting the prior art is referred to as 'game performance system' 200, which is a term more suitable for explaining the present invention, in accordance with the technical purpose of the present invention because it operates in conjunction with the technology of the present invention. do.

1 illustrates the concept of a learning model modeled on gamers of the prior art.

In the human world, gamers manipulate the game through keyboard or mouse manipulation in certain game states. Gamers have their own unique game styles. This game style is a factor influencing the game operation of the gamers. In the computer world, gamers' game styles are interpreted as human genetic information, divided into detailed categories of styles, and given values to express them as style vectors (S).

In addition, the process of operating a game in the human world is a change of game state over time. In the computer world, such a game state is defined as a weight vector (W), which is defined as the state component and the number of divisions that separate each state component.

In addition, in the human world, gamers input commands in a specific game state according to their game styles, and appear as game outputs in the computer world, and express each game output item as an output vector (O).

2 shows an example of a style vector, a weight vector, and an output vector model constituting the prior art.

The style vector S represents gamers' game habits or propensities, for example, aggression, defensiveness, input frequency, stability, and the like. Style vector (S) has a feature that can be found more detailed and intelligent game style gamers as the number of items that distinguish the user's game style.

The weight vector (W) defines a game state, and each game state is My Action (Action1), Opponent's Action (Action2), My Health (Power1), Opponent's Health (Power2), and Me and the opponent's distance (Distance) It is defined as a component of 5-dimensional item composed of. In addition, it is assumed that each component has a division number of 15, 15, 3, 3, and 10 levels. In the Dalian game played by gamers, the game state is determined by changing the above five game components according to the passage of time or the input operation of the gamers. When each component is divided by the number of divisions, my action (Action1) is divided into 15 ranges, and the remaining components are also divided by the number of divisions. Are distinguished. Therefore, the total game states generated by dividing the five-dimensional components, respectively, are 15 * 15 * 3 * 3 * 10 = 20,250, which has a total of 20,250 game states.

The output vector O is a set of game outputs that causes a transition from a particular game state to the next game state. This occurs by the player entering the best command through the input device among the possible input operations in a particular game state.

In the game performance system of the prior art, the progress of the game is defined as the relationship between the style vector (S), the weight vector (W) and the output vector (O) described above.

3 illustrates a relationship between a style vector, a weight vector, and an output vector constituting the prior art.

The style vector S is information obtained by dividing and quantifying the game progress of individual gamers into detailed items. It is like a set of unique user styles that determine the output of a game as if it were human genetic information. The style vector S is mapped to a specific output vector O through a calculation with a weight vector W in a specific game state. Expressed in terms of the human world, a gamer selects a particular operation from among game operations dependent on a specific game state (VS.weight vector) according to his game style (VS.style vector). It is to generate branch game output (VS. output vector).

A total of 20,250 game states consist of a two-dimensional weight vector (W) matrix that maps the style vector (S) to an arbitrary output vector (O). The k column in the weight vector W causes a specific output vector Ok. That is, the weight vector is information defining a relationship (correlation) between the style vector S and the output vector O in a specific game state.

The output vector O is a set of possible outputs obtained through the calculation of the style vector S and the weight vector W of a specific game state. The computed value of the possible output is related to the probability that the output will occur in the actual game. Therefore, the output vector value of the possible output having the highest value can be defined as the highest occurrence probability. This functional relationship is expressed as an expression as follows.

Style Vector (S): Vector S (n)

Output Vector (O): Vector O (m)

Weight vector (W): Vector W (n * m)

1 <= i <= n, 1 <= k <= m, 1 <= j <= m,

According to the defined n, m, the weight vector W exists in the two-dimensional matrix of n * m as many as the total number of state spaces (20,250). Therefore, the j column of the weight vector W stores the association information for determining the specific output vector Oj with respect to the style vector S.

In the prior art, a style vector (S) is set to define individual game progress styles for a plurality of test gamers, and the style vector (S) and output are analyzed by analyzing which inputs test gamers generate game outputs in individual game states. The weight vector W for each game state having generalized association information between the vectors O is trained. When the weight vector W is learned, the conventional game performance system 200 has the association weight vector W information on which game output a certain game style generates for all game states.

Next, if the game style of the transplanted gamers (owner gamers) is not known, the game performance system 200 may be inferred by inferring and learning the style vector S from the weight vector W learned from the actual game output of the transplanted gamers. Build). That is, the game performance system 200 enables style transplantation to the agent NPC by completing the style vector and the generalized weight vector of the gamer to be transplanted.

The control of the agent NPC in which the style of the specific gamer is transplanted is controlled by the game execution system 200 from the game operating system 100, and the style vector S of the gamer to be transplanted based on Equation 1 described above. And calculate the weight vector (W) of the current game state to predict the output vector (O), and transmits the game output matching the output vector (O) to the game operating system (100).

In the conventional game performance system 200, the game output matching the predicted output vector (O) is transmitted to the agent NPC, so that the agent NPC in which the game style of a specific gamer is implanted proceeds on behalf of the game according to the game style of the gamer. Implemented.

In the above-described conventional game performance system 200, since the game style of the owner gamer can be ported to the agent NPC, there is an advantage of creating a more dynamic game through the creation of various characters.

On the other hand, as a result of observing the game operating method of the agent NPC by applying the conventional game style transplant system 100 to the game operating system 100, the agent NPC operates the game according to the game style of the owner gamers, Consistency was determined by winning and losing game style.

For example, the owner gamer may not be able to overcome a certain game state during the actual game progression, and may sometimes lose. The agent NPC, which transplants the game style of the owner gamers, follows all the advantages and disadvantages of the owner and proceeds with the game. Therefore, the agent NPC does not overcome the difficulties in the game state where the owner meets in the actual game progression. There was a problem with losing the game.

In this case, the gamer repeats the game until it finds a way to win by starting the game again, reaching a certain game state, and playing the game in a new style. Or get the game information progressed by the gamers around you and find out how to win by applying it to the game state you are in.

The present invention, based on the game style of the owner gamers, the agent NPC in the game is in the game state of the game is in the hurry of win or lose during the game, disadvantageous game state during the game progress, or any game state, only for a while The game style value of the other gamers with high odds is not the owner's game style value, so that the game is out of the game state.

That is, in the prior art, the game style of the master gamers is transplanted, so that the game is consistently performed according to the game style learned by the agent NPC regardless of the win or lose, but in the present invention, the agent NPC based on the game style of the host gamers is briefly selected. While the game style of the gamers, not the game style of the master gamers, the game style optimized by the gamers temporarily adopts the game style of the specific gamers to implement the agent NPC of improving the ability to play the game.

The present invention was devised to solve the above problems, and based on the odds information of individual test gamers, the transition from a specific game state to a game state through the game log of test gamers affects the actual win or loss. A system and method for improving agent NPC's ability to inject a game style of a specific gamer that infers and learns game styles, and temporarily transfers the optimal game styles to agent NPCs to support agent NPCs to play games with high win rates. The purpose is to provide.

In order to achieve the above object, the capability improvement system of the agent NPC in which the game style of a specific gamer is implanted according to the present invention is an agent of a game operating system by learning the optimal game style for each game state transition based on the odds from the test gamers. In the agent NPC capability improvement system that is temporarily applied to the NPC, the game state by gamers including the ID, the game state transition path, the win / loss / parity wins, and the style vector of the individual test gamers from the game information progressed by the test gamers. A game state transition generation engine for generating transition odds information and storing the game state transition odds DB for each gamer; Extracting the highest odds information for each game state transition from the game state transition odds information for each gamer generated by the game state transition generation engine, and the game win rate information and the highest odds information recorded by the owner gamers in a specific game state transition An optimal game style extraction engine that extracts the optimum game style information to be applied and stores the optimal game style DB to be applied; And extracting game style information of a specific gamer that matches the current game state of an agent NPC from the application target optimal game style information extracted by the optimal game style extraction engine, and applying game style information of the specific gamer to the agent NPC. It is characterized in that it comprises a determination engine for applying the optimal game style applied to the agent NPC temporarily through the game performance system.

According to a preferred feature of the invention, the game state transition generation engine, game state transition generation module for generating and collecting the game state transition path from the game information progressed by the individual test gamers in the game operating system; A game state transition memory buffer in which the game state transition path generated by the game state transition generation module is stored; A game state transition storage module for storing game state transition odds information for each gamer with reference to the game state transition memory buffer; And a game state transition odds DB for each gamer in which the game state transition odds information for each gamer generated by the game state transition storage module is stored.

In addition, the game state transition storage module is distributed by reducing the number of state components that define the game state or the number of partitions for dividing the state components with respect to the individual game states stored in the game state transition memory buffer, The amount of memory resident data is reduced by grouping the game state locally to reduce the total number of game states.

When the game state transition path is stored, the game state transition storage module configures the game state transition path as a current game state or a game state immediately preceding the current game state and stores the game state transition path in the game state transition odds DB for each gamer. By storing, the game state transition path constitutes a transition path including at least one or more game states, and as the previous game states increase, reliability of the winning rate is increased based on empirical information.

The optimal game style extraction engine, which constitutes the capability improvement system of the present invention, uses the game state transition path based on the game state transition path information generated by the game state transition generation engine as a game state transition path. Accumulating game state accumulation module; A game status win rate DB for storing game player transition rate win information of all gamers accumulated by the game state accumulation module; An optimum win rate extraction module for extracting only the highest win rate record for each game state transition stored in the win rate DB for each game state; An optimal win rate DB for each game state, which is extracted by the optimum win rate extraction module and stores the highest win rate record; A winning rate DB for each game state of the master gamers, in which the winning rate information for each game state recorded by the owner gamers is stored; It is possible to extract win rate information of a game state that matches the game state recorded in the master gamers' game rate win rate DB from the optimal win rate DB for each game state, and the win rate of the master gamer may be higher than the maximum win rate for the matched game state. An application target optimum game style extraction module for deleting the record of the game state from the optimal win rate DB for each game state; And the application target optimal game style information including the style information and the winning rate of the specific gamers which are optimally determined by comparing the optimal game style extraction module with the highest win rate and the owner gamers for individual game state transitions. It is characterized in that the application comprises the optimal game style DB.

The optimal game style application determination engine constituting the capability improvement system of the present invention includes: a memory loading module that reads data of an application target optimal game style DB extracted by the optimal game style extraction engine into a memory; A game state extraction module that receives a game state progressed by the agent NPC from the game operating system; And retrieves the game style of the particular gamer from the memory in a game state that matches the game state extracted by the game state extraction module, and transfers the game state to the game operating system if the search fails. Optimal game style that improves the capability of agent NPC by temporarily utilizing the game style of a specific gamer by utilizing the agent NPC and determining whether it is applied if the search is successful and transferring the game state and the style vector to the game execution system. Characterized in that comprises a utilization module.

In addition, the optimal game style utilization module, when the search for the style vector of the particular gamer is successful, the difference between the odds of the specific gamers and the owners of the gamers, the game execution time of the agent NPC, the previous game of the specific gamers It is determined whether to deliver to the game operating system in consideration of the number of styles provided, and if it is determined to deliver, the game state and the style vector of the specific gamer are delivered to the game operating system a predetermined number of times, and the number of game styles provided. It is characterized by increasing the.

In the method of improving the capability of an agent NPC in which a game style of a specific gamer is implanted according to the present invention, an agent NPC that learns an optimal game style for each game state transition based on a winning percentage from test gamers and temporarily applies it to an agent NPC of a game operating system A method for improving capability, comprising: (1) receiving game information progressed by test gamers from a game operating system; (2) generating, from the game information, game state transition odds information for each gamer including an ID, a game state transition path, a win rate, and a style vector of the individual test gamers; (3) extracting the best odds information for each game state transition from the game state transition odds information for each gamer, and comparing the best odds information with the game winning rate information recorded by the owner gamers in a specific game state transition, and applying the optimal game style information. Extracting; And (4) extracting game style information of a specific gamer that matches the current game state of the agent NPC from the application target optimal game style information, and determining whether to apply the game style information of the specific gamer temporarily to the agent NPC. Characterized in that comprises the step of applying to.

According to a preferred feature of the present invention, the step (2) comprises: (2-1) generating a game state from the game information, and generating and storing a game state transition path according to the game progress of the individual test gamers; (2-2) Locally distributed game states by reducing the state component defining the game state or reducing the number of divisions for dividing the state components with respect to the individual game states constituting the game state transition path. Grouping to reduce the total number of game states; And (2-3) storing the game state transition odds information for each gamer including game state transition paths, odds, and style vectors for each ID of the individual test gamers with reference to the reduced game state transition paths. It is characterized by.

The game state transition path may include a current game state or a game state immediately preceding the current game state, thereby forming a transition path including at least one or more game states. As a result, the reliability of the odds ratio is increased.

The step (3) of constructing the capability improving method of the present invention comprises: (3-1) accumulating and storing the odds information of all test gamers by using the individual game state transition paths as the key from the game state transition rate information for each gamer. step; (3-2) extracting and storing only the record having the highest odds among the odds information for each test gamer with respect to the stored game state transition path; (3-3) obtaining win rate information for each game state recorded by the owner gamer; (3-4) The owner gamers' odds of step (3-3) and the highest of steps (3-2) with respect to the game state of step (3-2) that match the game state of the owner gamers of step (3-3) Comparing the odds ratio; And (3-5) when the owner gamer win rate is greater, deleting the record corresponding to the game state of step (3-2), and including the style information and the win rate of the specific gamer determined optimally. And storing the style information.

The step (4) of constructing the capability improving method of the present invention includes: (4-1) reading the application target optimal game style information into a memory; (4-2) receiving a game state of the agent NPC from the game operating system; (4-3) retrieving the style vector of the particular gamer from the memory of the step (4-1) in the game state that matches the game state of the step (4-2); (4-4) if the search of the step (4-3) is unsuccessful, transferring the game state to a game operating system, and the agent NPC proceeds to play the game by the owner's style vector; And (4-5) if the search of the step (4-3) is successful, it is determined whether or not to apply and transfers the game state and the retrieved style vector of the specific gamer to a game operating system to apply the game style of the specific gamer. And temporarily improving the capability of the agent NPC.

Further, the step (4-3), when retrieved from the memory of the step (4-1) with respect to the game state, there is (4-3-1) real-time game state transition information having a high odds And displaying the game screen.

In addition, the step (4-5), (4-5-1) the difference between the odds of the specific gamers and the owners of the gamers, the total game execution time of the agent NPC and the number of game styles provided by the specific gamers before Determining whether or not the transfer to the game operating system is considered; (4-5-2) if it is determined to be delivered, transferring the game state and the style vector of the specific gamer to the game operating system a predetermined number of times; And (4-5-3) increasing the number of times the game style is provided after delivering the predetermined number of times.

The agent NPC capability improvement system and method constituting the technology of the present invention as described above, after learning the game state transition path of the entire test gamers, and extracts the game state transition path of the individual gamers having the highest odds for the entire game state By constructing the optimal game state transition information, in the game progress of the agent NPC in which the master gamer's style is transplanted, the optimal game style can be temporarily implanted into the agent NPC in a specific game state. Therefore, certain gamers' styles have the advantage of improving the temporary capability of the agent NPC and correcting the game styles of the owner gamers through porting.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, terms or words used in the specification and claims should not be construed as having a conventional or dictionary meaning, and the inventors should properly explain the concept of terms in order to best explain their own invention. Based on the principle that can be defined, it should be interpreted as meaning and concept corresponding to the technical idea of the present invention. Therefore, the embodiments described in the specification and the drawings shown in the drawings are only the most preferred embodiment of the present invention and do not represent all of the technical idea of the present invention, various modifications that can be replaced at the time of the present application It should be understood that there may be equivalents and variations.

1 to 3 illustrate the basic concepts of the prior art on which the technology of the present invention is based. The present invention adopts the basic environment that the game style of the gamers through the specific game output in the individual game state corresponding to the game progress as in the prior art.

4 illustrates an environmental concept of an agent NPC capability improvement system that constitutes the techniques of the present invention. The prior art is composed of a game operating system 100 and a game performance system 200. The solid line shown in FIG. 4 is the data transfer in the process of improving the capability of the agent NPC, and the dotted line is the data transfer of the process in which the agent NPC is based on the style of the owner gamer as in the prior art.

The conventional game performance system 200 receives game information corresponding to a game log of test gamers performing a game from the game operating system 100, and builds a style vector DB by digitizing the style vector S of individual test gamers. In addition, the weight vector DB is constructed by inferring and learning the weight vector W having the correlation information between the output vector O actually caused by the individual style vector S. Then, the style vector S corresponding to the game style of the owner gamer is obtained.

Details of the prior art are introduced in Patent Application No. 2004-27918. In this specification, the game style of the owner gamers in the conventional game performance system 200 is briefly summarized in real time. The conventional game performance system 200 receives the game state of the current agent NPC from the game operating system 100, performs an operation with the style vector S of the owner gamer and the weight vector W of the received game state and outputs the result. By generating the vector (O) (see Equation 1) and passing the game output matching the output vector (O) back to the game operating system 100, the agent NPC proceeds the game based on the owner's game style. Do it.

However, in the present invention, the capability improvement system 1 receives the game state of the current agent NPC from the game operating system 100, searches for an optimal game style corresponding to the received game state, and determines whether to apply it. In the case of improving the capability of the agent NPC, the game state and the optimal game style retrieved are transmitted to the conventional game execution system 200 so that the agent NPC is not the game style of the owner gamer for a while but the game style of the specific gamer determined to be optimal. Apply to improve ability. When the agent NPC is based on the game style of the owner gamers, only the game state received by the capability improving system 1 is transmitted to the conventional game performance system 200.

In other words, the technique of the present invention improves the capability of the agent NPC by applying the game style of a specific gamer with a high odds of success, rather than the game style of the master gamer, in the specific game state of the agent NPC in which the game style of the master gamer is transplanted. It is to let.

Therefore, in the prior art, the propensity of the owner gamers is followed as it is, and the game proceeds according to the owner's game style regardless of whether the game is won or lost. However, in the technique of the present invention, the game game system 200 temporarily provides an optimal game strategy style of the specific gamers extracted with the highest win rate to temporarily help the game of the owner gamers (agent NPCs).

In the present invention, the term 'optimal' does not necessarily lead to a perfect game state transition through the 100% win rate or the highest win rate. However, it has a combination of game style information that is relatively superior to the odds of the owner gamers in the game state progressed by the owner gamers, and records the excellent odds of others (test gamers) even in the game states where the owner gamers have not progressed. The term optimal is given in consideration of the degree to which the transition path to the state matches the host gamers. For example, game state transitions of test gamers determined to be optimal in a particular game state may temporarily lead to an agent NPC in an adverse game state. This is because test gamers deliberately create an adverse game state and then switch to an advantageous game state by obtaining higher points.

5 illustrates the concept of a game state transition path that constitutes the techniques of the present invention. In the present invention, a model of the progress of the game is operated based on the highest odds information. Indeed, the best odds information refers to the complete odds of winning 100 times by playing 100 games, but in practice, since this is a virtual goal win rate, the game of certain gamers with excellent odds from the test gamers who are the subject of learning in the present invention. The optimal game style is extracted by learning the optimized transition path by combining the state transition path.

In the present invention, it is assumed that the deciding factor of the victory is that the gamer has the final victory by going to the next game state which is more advantageous in the current game state. Therefore, the test gamers learn about the game state transition path that is passed through the game progress.

Referring to FIG. 5, when the state transition path 2 is mainly a game style path, and the state transition path 1 is a game style path mainly winning, the game is performed according to the game style mainly lost in the game state x that is in an intersecting state. This is a method that temporarily reflects a certain part of the winning game style to the agent NPC. This is due to the behavior of human gamers who model their propensity to play games in real life and improve their gaming skills.

The game state in the present invention is defined as the number of partitions assigned to the game state component and the unit game state component as described above with reference to FIG. In order to construct the optimal game state transition path, it is necessary to have the odds information for the transition from one game state to another game state.

In a game space with a total of n game state nodes, assuming that the game state unit node has the odds weight information of the path transitioning to another node based on a blind path search, the total number of weight information is ( n-1) * n. This is calculated as 20,249 * 20,250 in the present invention, and there is a maximum of about 400,000,000 weight information. This is not practical in practice because of memory loading and processor operations.

Therefore, the present invention considers only the transition paths that the actual test gamers actually perform based on heuristic information, reduces the number by grouping the game state, and compares the optimal win rate with the win rate of the owner gamers. The memory loading and operation problems are solved through optimization techniques such as having only the odds information.

6 illustrates a configuration of an agent NPC capability improvement system according to an embodiment of the present invention. The agent NPC capability improvement system 1 according to the present invention comprises a game state transition generation engine 2, an optimal game style extraction engine 3, and an optimal game style application determination engine 4.

The game state transition generation engine 2 receives game information performed by test gamers from the game operating system 100. The game state is extracted based on the received game information, and the game state transition path is constructed by tracking the variation of the game state. When the game state transition path is configured, the game state transition odds information for each gamer is generated based on gamer information, odds information, and style vector information. Here, the test gamers preferably adopt a group having a high win rate as the test gamer, but it is also possible to configure a game state transition path by adopting any general gamer group operating in the game operating system 100.

The optimal game style extraction engine 3 extracts the highest odds information of the test gamers' odds for each game state transition based on the game state transition odds information for each gamer generated by the game state transition generation engine 2, and is recorded by the owner gamer. If the game odds information is larger than the maximum odds information, the optimal game style information to be applied is extracted by deleting the highest odds information.

The optimal game style application decision engine 4 is an engine that improves the capability of the agent NPC based on the game style of the owner gamers in real time after the capability improvement system 1 is established through the optimal game style extraction engine 3.

The game style information of the specific gamers corresponding to the optimal style information of the game state that matches the current game state progressed by the agent NPC is extracted from the extracted application target game style information, and the game style of the specific gamers is limitedly presented according to the conditions. The style vector S is transmitted to the game performance system 200 together with the game state.

7 is a detailed configuration of a game state transition generation engine 2, an optimal game style extraction engine 3, and an optimal game style application determination engine 4 constituting the capability improving system 1 according to an embodiment of the present invention. To show.

-Game State Transition Fish Engine (2)

The game state transition generation engine 2 includes a game state transition generation module 21, a game state transition memory buffer 22, a game state transition storage module 23, and a game state transition odds DB 24 for each gamer. do.

The game state transition generation module 21 receives game information progressed by the individual test gamers from the game operating system 100, analyzes the game state, generates a game state transition path, and temporarily generates the game state transition memory buffer 22. Save as.

The game state transition storage module 23 stores the game state transition odds information for each gamer in the game state transition odds DB 24 for each gamer by referring to the game state transition memory buffer 22 whenever the game is finished.

The game state transition odds information for each gamer stored by the gamers DB 24 stores the game state transition paths for the individual test gamers' IDs, the odds and the style vectors of the win / loss / empty records recorded so far by the gamers. S).

Table 1 below shows the specifications of the game state transition odds DB by gamers.

Game state transition odds DB by gamers Gamer ID (key) Game state transition path (St _0- > St ₇ , St _7- > St ₄ , St ₄ -St ₁₁₂ ....) Win rate (62% / 28% / 10%) Style vector (S ₁ , S ₂ .... S _n )

In particular, the game state transition storage module 23 reduces the state component defining the game state for each game state stored in the game state transition memory buffer 22 or reduces the number of divisions for dividing the state component. It is desirable to group the distributed game state locally. Grouped game states serve to reduce the amount of memory-resident data by reducing the total number of game states. For example, in the present invention, five state components constituted the divided number of 15, 15, 3, 3, and 10, respectively, where the fourth state component is not related to a game and is not considered, and the fifth state component is configured. If grouping is performed by setting the number of divisions of the element to 3, 15 * 15 * 3 * 3 = 2,025 pieces can be downsized to 1/10 of the actual game state number 20,250.

 In addition, when the game state transition storage module 23 stores the game state transition path, the game state transition path composed of the current game state or the previous game state immediately before transitioning to the current game state and the game state game state for each gamer It is desirable to construct a transition path including at least one or more game states by storing in the transition odds DB 24. The main reason is that the more the previous game state, the higher the reliability of the optimal game style information based on the homogeneity of the style, because the transition path of the optimal odds information and the transition path of the owner gamers match based on the empirical information. Therefore, the number of previous game states in the game state transition path may be a criterion for determining an optimum.

-Optimal game style extraction engine (3)

The optimal game style extraction engine 3 includes a game state accumulation module 31, a game win rate DB 32 for each game state, an optimum win rate extracting module 33, a game win rate DB 34 for each game state, and a game player win state for each game state. And a DB 35, an application target optimal game style extraction module 36, and an application target optimal game style DB 37.

The game state accumulation module 31 extracts individual game state transition paths from the game state transition odds DB 24 of individual gamers generated by the game state transition generation engine 2, and sets the key to win the total test gamers. The information is accumulated and stored in the winning rate DB 32 for each game state. Thus, a plurality of test gamer odds information is listed for the unit game state transition path.

Information stored in the odds database for each game state is shown in Table 2 below.

Odds DB by Game Status Game State Transition Path (key) St _0- > St ₇ Gamer ID (A Gamer) + Odds + Style Vector B Gamer + Odds + Style Vector. . . St _0- > St ₈   . . .

As described above, the winning rate DB 32 for each game state accumulates the winning rate information of all test gamers by using each game state transition path as a key.

The optimal win rate extraction module 33 extracts only the highest win rate record for each game state transition stored in the win rate DB 32 for each game state and stores the highest win rate record for each game state in the optimum win rate DB 34. Therefore, only the game style of the game player having the highest odds in each game state is extracted and the data of the remaining gamers are ignored and the data is downsized.

Depending on the strategy, it is also possible to modify to extract information of gamers having a plurality of high odds. In this case, a plurality of optimum odds information is provided. As mentioned earlier, 'optimal' is not necessarily the term meaning the best, and it is possible to transform it to obtain optimized data by applying appropriate empirical information through repeated trials.

Table 3 below shows the specifications of the optimum odds DB (34) for each game state.

Optimum odds DB by game state Game State Transition Path (key) St _0- > St ₇    Gamer + Odds + Style Vector St _0- > St ₈ X Gamer + Odds + Style Vector    . . .    . . .

The application target optimal game style extracting module 36 extracts the win rate information of the game state corresponding to the game state recorded in the master gamer game state win rate DB 34 from the optimum win rate DB 34 for each game state, and matches If the game player's winning rate is higher than the highest winning rate, the record of the game state is deleted from the optimum winning rate DB 34 for each game state.

The reason is that the master gamers have a higher chance of winning the game state transition path of the host gamers accumulated in the master gamers game state winning rate DB 34. Because you don't need any more help in that game. Thus, the optimal data is downsized once again to valid data.

Therefore, in the optimal winning rate DB 34 for each game state extracted and constructed by the application target game style extraction module 36, the game state transition path of the gamers having a higher odds than the master gamers and the game state where the owner gamers have never progressed at all. The optimal route information from is stored so that only valid data is actually stored.

The application target optimal game style extraction module 36 from the downsized game state optimum win rate DB 34 applies the target game style information to be applied including the game state transition path, the win rate, and the style vector. 37). The data stored in the application optimum game style 37 is data resident in memory, and may be downsized once again through various techniques depending on the size and distribution of the data.

-Engine for determining optimal game style

The optimal game style application determination engine 4 includes a memory loading module 41, a game state extraction module 42, and an optimal game style utilization module 43.

The memory loading module 41 resides in the memory the data of the application target optimal game style DB 37 extracted by the optimal game style extraction engine 3, and the game state extraction module 42 is an agent from the game operating system 100. Receive real-time game state progressed by NPC.

The optimal game style utilization module 43 retrieves a game state corresponding to the current game state from data resident in memory and determines whether to apply a game style of a specific gamer determined to be optimal in the retrieved game state to the agent NPC. If applied, the agent's optimal game style is used to improve the agent NPC's ability. If not, the agent's game style follows the owner's game style.

The optimal game style utilization module 43 determines whether the optimal game style is applied according to the following criteria when the style vector of the specific gamer is successfully searched. When the difference between the odds of the specific gamers and the owners of the gamers is greater than or equal to the threshold value, the application may be made in consideration of the game execution time of the agent NPC. It is determined whether to transfer to the game operating system (100).

In addition, the case where the optimal game style is not applied is as follows. This is the case where the above-mentioned application criteria are not met or when there is no game state matching the data residing in the memory. In the latter case, the owner gamers have no experience of game transition, and the test gamers have no experience of game transition.

When it is determined that the optimal game style is applied, the optimal game style utilization module 43 transmits the game state and the retrieved style vector S to the game performance system 200 a predetermined number of times. The reason for delivering the predetermined number of times is that the performance of the agent NPC can be improved only by repeating to some extent according to the game style of a specific gamer. If the delivery is made a predetermined number of times, the optimal game style is provided.

If it is determined that the optimal game style is not applied, the optimal game style utilization module 43 transmits only the game state to the game performance system 200. Since this is the same as the input of the conventional game performance system 200, it controls the agent NPC through the style vector (S) of the owner gamer.

8 is a flowchart illustrating a method for improving agent NPC capability according to an embodiment of the present invention.

In the game information receiving step (S10), test gamers receive real-time game information performed by the game operating system 100.

The game state transition odds information for each gamer including the game state transition path, the odds ratio, and the style vector of the individual test gamers is generated from the game information received in the gamer-specific game state transition odds information generation step (S20).

In the step of extracting the optimum game style information to be applied (S30), the highest win rate information is extracted for each game state transition from the game state transition odds information for each gamer, and the optimal game style information to be applied is compared with the game win rate information recorded by the owner gamers. Extract it.

In the optimal win rate information style vector transplanting step (S40), the optimal win rate information is extracted from the target game style information to be applied by extracting the optimal style information of the game state that matches the current game state of the agent NPC, and determining whether to apply it. The style vector of the specific gamer is reflected in the agent NPC.

9 is a detailed flowchart of a game state transition generation process according to an embodiment of the present invention.

In the game state transition path generation / storage step (S21), a game state is generated from the received game information, and a game state transition path is generated and stored by tracking a game progress of a test gamer.

In the total game state number reduction step (S22), the game states distributed by reducing the number of state components defining game states or reducing the number of divisions for dividing the state components with respect to individual game states constituting the game state transition path. Grouping locally reduces the total number of game states.

In the game state transition odds information storing step for each gamer (S23), the game state transition odds information for each gamer including the ID, the game state transition path, the odds record, and the style vector is stored for each test gamer with reference to the reduced game state transition path. do.

10 is a flowchart illustrating an optimal game style extraction process according to an embodiment of the present invention.

In the winning rate information accumulating / storing step (S31), the winning rate information of all test gamers is accumulated and stored using the individual game state transition paths as the key from the winning game state for each gamer in step S23.

In the extracting and storing the highest winning rate information record step (S32), only the record having the highest winning rate information among the winning rate information of all test gamers is extracted and stored with respect to the stored unit game state transition path.

In the step S33 of acquiring the game state of the owner gamers, the owner gamers acquires the odds information of the game state transition path recorded in the past.

In the odds comparison step S34, the master gamer win rate in step S33 and the highest win rate in step S32 are compared with the game state of step S32 that matches the game state of the master gamer in step S33.

In the application target optimal game style information storage step (S35), if the owner gamer win rate is greater than the maximum win rate, by deleting the record corresponding to the game state of the step (S32) and optimizing, based on the data to apply the optimum game style information Create and save.

11 is a flowchart illustrating an optimal game style transplantation process according to an embodiment of the present invention.

In the memory loading step S41, the application target optimal game style information generated in step S35 resides in the memory.

In the agent NPC game state receiving step (S42), the game state of the agent NPC is received from the game operating system 100.

In the game state memory search step S43, a game state that matches the game state in step S42 is searched from the memory in step 4-1. In the step S43, if the matching game state is retrieved from the memory resident data through the real-time optimal odds information presentation step S431, it will inform that the game can be performed on behalf of the current game at a higher game win rate on the game screen in real time. It may be.

If the search fails, in the game state delivery step (S44), only the game state is transmitted to the game performance system 200 based on the game style of the owner gamer without improving the capability of the agent NPC.

If the search is successful, the application criteria are compared in the determination and delivery step 45, and if appropriate, the game style of the specific gamer determined as the optimal style vector in the searched game state is reflected in the agent NPC.

12 is a detailed flowchart of a process of applying an optimal game style to an agent NPC according to an embodiment of the present invention.

In the comparison of the application criteria and the determination of whether or not to deliver the application (S451), it is determined whether to improve the agent NPC based on the predetermined application criteria. The application criterion is to determine the game style of a particular gamer that is optimally determined by comparing the difference between the odds of the specific gamer and the master gamer, comparing the execution time of the agent NPC, and comparing the number of game styles already provided by the particular gamer. Determine whether to deliver to the performance system 200.

If it is determined that the optimal game style is not applied, the game state received from the game operating system 100 is transmitted to the game execution system 200 as it is through the game state delivery step S44 described above.

In the case where the optimal game style is not applied, it is determined that the application criteria are not suitable for the application criteria in the above-described comparison and delivery determination step (S451), or there is no game state that matches the data resident in the memory. In the latter case, the owner gamers have no transition experience in the game state, and the test gamers have no transition experience. If the optimal game style is not applied as described above, the agent NPC transmits the game output to the agent NPC according to the game style of the owner gamer.

In addition, if it is determined that the optimal game style is applied, the game state and style vector delivery step (S452), the specific gamers' style vector (S) determined optimally with the current game state of the agent NPC received from the game operating system 100 To the game performance system 200 a predetermined number of times.

In particular, when it is determined that the optimum game style is applied once, the game state is continuously received from the game operating system 100 in the agent NPC game state receiving step S42 a predetermined number of times, and immediately steps S43 to S451. The game state and style vector transfer step (S452) is performed without performing the step). The reason is to ensure that the game style of the specific gamer determined to be optimal is effectively reflected in the agent NPC. If the game style of a certain gamer is delivered only once or if it is reflected too small, the game style tendency of the owner gamers remains in the agent NPC rather than the game style of the specific gamers.

After step S42 and step S452 are successively performed from the game operating system 100 by a predetermined number of times, the count is increased in the step of increasing the optimal game style providing number (S453).

As described above, although the present invention has been described by way of limited embodiments and drawings, the present invention is not limited thereto and is intended by those skilled in the art to which the present invention pertains. Of course, various modifications and variations are possible within the scope of equivalents of the claims to be described.

The system and method for improving the capability of the agent NPC in which the game style of a specific gamer is implanted according to the present invention is optimized by combining and optimizing all the game state transition paths of the test gamers with the game state transition paths of the gamers with excellent odds. By learning the transition path and presenting the best odds information for each game state of the agent NPC, there is an effect of temporarily strengthening the ability of the agent NPC based on the owner gamer to the optimized style of a specific gamer.

In addition, the owner gamers can be informed in real time about the optimal winning rate information about their current game state, and if watching the game played by the agent NPC temporarily implanted with the optimal winning rate information compared to the progress of his game Because it can be analyzed, it has the effect of correcting your game style.

Claims (3)

In the agent NPC capability improvement system that learns the optimal game style for each game state transition based on the winning rate from test gamers and temporarily applies it to the agent NPC of the game operating system, From the game information progressed by the test gamers, the game state transition odds information for each game player including the ID of each test gamer, the game state transition path, the win / loss / fraction odds, and the style vector is generated, and the game state transition odds DB for each gamer is generated. A game state transition generating engine for storing; Extracting the highest odds information for each game state transition from the game state transition odds information for each gamer generated by the game state transition generating engine, and comparing the highest odds information with the game winning rate information recorded by the owner gamer in a specific game state transition. An optimal game style extraction engine for extracting the optimal game style information to be applied and storing the optimal game style DB to be applied; And Whether to extract game style information of a specific game player that matches the current game state of an agent NPC from the application target game style information extracted by the optimal game style extraction engine, and apply game style information of the specific gamer to the agent NPC. Determination engine for applying optimal game style to the agent NPC temporarily through the game execution system System for improving the capability of the agent NPC implanted the game style of a particular gamer, characterized in that comprises a. The method of claim 1, The optimal game style extraction engine, A game state accumulation module for accumulating the odds information of all test gamers by using a game state transition path as a key from the game state transition odds information for each gamer generated by the game state transition generation engine; A game status win rate DB for storing game player transition rate win information of all gamers accumulated by the game state accumulation module; An optimum win rate extraction module for extracting only the highest win rate record for each game state transition stored in the win rate DB for each game state; An optimal win rate DB for each game state, which is extracted by the optimum win rate extraction module and stores the highest win rate record; A winning rate DB for each game state of the master gamers, in which the winning rate information for each game state recorded by the owner gamers is stored; It is possible to extract win rate information of a game state that matches the game state recorded in the master gamers' game rate win rate DB from the optimal win rate DB for each game state, and the win rate of the master gamer may be higher than the maximum win rate for the matched game state. An application target optimal game style extraction module for deleting the record of the game state from the optimal win rate DB for each game state; And The application-optimized game style extraction module stores the application-optimized game style information including the game information and the style information of the specific gamers that are optimally determined by comparing the winning rate and the winning rate of the owner gamers with respect to individual game state transitions. Target optimal game style DB A system for improving the capability of the agent NPC implanted game style of a particular gamer characterized in that it comprises a. In the agent NPC capability improvement method that learns the optimal game style for each game state transition based on the winning rate from test gamers and temporarily applies it to the agent NPC of the game operating system, (1) receiving game information progressed by the test gamers from the game operating system; (2) generating, from the game information, game state transition odds information for each gamer including an ID, a game state transition path, a win rate, and a style vector of the individual test gamers; (3) extracting the best odds information for each game state transition from the game state transition odds information for each gamer, and comparing the best odds information with the game winning rate information recorded by the owner gamers in a specific game state transition, and applying the optimal game style information. Extracting; And (4) Extracting game style information of a specific gamer that matches the current game state of the agent NPC from the application target optimal game style information, and determining whether to apply the game style information of the specific gamer temporarily to the agent NPC Steps to Apply Method for improving the capability of the agent NPC implanted game style of a particular gamer, characterized in that comprises a.

Images