KR101265051B1 - Method for providing non player character artificial intelligence according to game user level - Google Patents

Abstract

PURPOSE: An NPC AI providing method according to the game user's level is provided to control the level of NPC measured against the game user according to the level of the game played. CONSTITUTION: An NPC AI providing method according to the game user's level comprises a step(S110) of deciding the level according to the game's characteristics using genetic algorithm; a step(S120) of deciding the game level of the game user and storing the data according to game pattern; a step(S130) of deciding the level of NPC AI according to the game level of the user; a step(S130) of generating parameters with the NPC AI according to the stored data and the determined level of the NPC AI; and a step of providing NPC AI according to the level of the user by applying the generated parameters to the game. [Reference numerals] (AA) Start; (BB) End; (S110) Determine a level matched to the characteristic of a game; (S120) Determine the game level of a game user and store data according to game patterns; (S130) Determine an NPI Al level according to the game level of the game user; (S140) Generate parameters with respect to the NPC AI according to the stored data and the determined NPC AI level; (S150) Apply the generated parameters to the game

Description

How to provide NPC AI according to the level of game users {METHOD FOR PROVIDING NON PLAYER CHARACTER ARTIFICIAL INTELLIGENCE ACCORDING TO GAME USER LEVEL}

The present invention relates to an NPC AI providing method according to the game user level, and more particularly, to an NPC AI providing method according to the game user level provided by applying various levels of artificial intelligence NPC according to the game user's game ability level. .

In general, the role of NPCs in games varies. The most important of them is to play the role of opponent of the game user. Game developers have used AI techniques to implement the intelligent behavior of characters. The user doesn't want the opponent to be clumsy in the game. Even if the opponent of the game wins the game because of clumsy behavior, the fun is not great. Rather, it is easy to lose interest in the game and easily get tired of it.

For example, in a competitive action game, if the character controlled by the gamer is far away and the opponent character is not approaching and doing the wrong action from afar, the game user may lose interest in the game or be bored.

Thus, many game users have desired and demanded that the opponent's character or mob behave intelligently.

The AI techniques used in the game range from traditional techniques such as finite state machine (FSM) and fuzzy state machine (FUSM) to artificial life and neural networks.

Even with this method, the game developer eventually had to implement the NPC AI by coding in accordance with the planner's intention.

Related prior arts include Korean Patent Publication No. 2010-0104898 (published date: September 29, 2010, titled: Preset adjustment method and runtime game balancing adjustment method) for game balancing.

The present invention is to provide a method for providing an NPC AI according to the user level to adjust the level of the NPC to compete with the game user according to the level of the game user playing the game.

The technical problems to be achieved by the present invention are not limited to the technical problems mentioned above.

A method for providing an NPC AI according to the user level of the present invention for achieving the above object, in the method of generating a non-player character artificial intelligence (NPC AI) according to the game user level by using a genetic algorithm (Genetic Algorithm), Determining a level of fit level; Determining a game level of the game user and storing data according to a game pattern; Determining a level of NPC AI according to the game level of the game user; Generating a parameter for the NPC AI according to the stored data and the determined level of the NPC AI; And applying the generated parameter to a game to provide an NPC AI according to a user level.

Specifically, the game pattern may include the behavior of the character controlled by the game user and the state of the game.

In addition, the game level may be determined by at least one of the score of the NPC, the score of the game user, the damage to the game user NPC, the damage to the NPC, the hit rate of the NPC, and the hit rate of the game user.

The generating method of generating a parameter for the NPC AI may include selecting a parameter capable of changing a level in the NPC AI; Initializing the selected parameter; Calculating a goodness of fit of a level suitable for game characteristics and data according to the game pattern of the game user stored above; Calculating a parameter by a genetic operator by arbitrarily selecting a parent in proportion to the goodness of fit; And generating a calculated parameter when the goodness of fit is equal to or greater than a preset value.

In addition, the parameter may mean a value or information that is changed to intelligently operate the NPC.

The method may further include repeatedly performing a parameter operation until the goodness of fit is less than or equal to a preset value.

In addition, the method of arbitrarily selecting a parent may be selected using any one or more of a goodness-of-fit proportional selection method, a roulette selection method, an expected value selection method, a ranking selection method, and a tournament selection method.

In addition, the genetic operator may adopt any one or more of crossover, mutation, inversion, substitution, duplication, addition, and deletion.

As described above, the present invention has an effect that the development time is shortened because the game developer does not implement algorithms for all cases.

The present invention has an effect that can cause the interest of the game user by providing an artificial NPC according to the level of the game user.

The present invention can generate and obtain NPC behaviors in various cases even if the NPCs are not matched to the level of the game users, thereby providing an NPC having various behavior patterns to many game users.

In addition, it can be applied to the game development stage, so that various levels of NPC can be easily and quickly created for game development.

1 is a flowchart illustrating an NPC AI providing method according to a game user level according to an embodiment of the present invention.
2 is a flowchart illustrating an NPC AI generation method of an NPC AI providing method according to a game user level according to an embodiment of the present invention.
FIG. 3 is a diagram illustrating an example of parameters that may be selected in the NPC AI generation method of FIG. 2.
4 is a diagram showing a cross operation.
5 is an image for explaining an embodiment of the present invention.
6 is a diagram illustrating an example of selectable parameters according to the embodiment of FIG. 5.
7 is a graph showing an embodiment of the fitness of the individual.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Like elements in the figures are denoted by the same reference numerals wherever possible. In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail.

1 is a flowchart illustrating a method of providing an NPC AI according to a game user level according to an embodiment of the present invention. The method of providing an NPC AI according to a game user level is as follows.

Here, NPC AI (Non Player Character Artificial Intelligence) means an artificial character without a player, hereinafter referred to as 'NPC AI'.

First, a level of a level suitable for a game characteristic is determined (S110). At this time, since the criteria for determining the level are different for each game, the level is determined according to the characteristics of each game, and the level suitable for the game characteristics may be determined by the developer.

That is, some games may belong to a high score of 100 points, but some games may belong to a low score of 100 points. Therefore, it is preferable to set a standard for determining a level based on a criterion corresponding to each game.

Table 1 below lists the levels of levels according to embodiments for any game.

<-200 -199 to -100 -99 to 100 101-200 > 201 Very low level Low level Equivalent level High level Very high level

For example, as shown in Table 1 above, if a game scores less than -200 in a game, it is classified as a very low level, and if it acquires points between -199 and -100, it is classified as a low level. 100 may be classified at the level equivalent to NPC, 101 to 200 may be classified at a higher level than NPC, and 201 or more at a very high level.

In the embodiment of the present invention, although divided into five categories, it may be divided into three to ten categories according to the type of game or the developer.

Next, the game level of the game user is determined and data according to a game pattern is stored (S120). In this case, the game user's game level is determined by one or more of the score of the NPC, the score of the game user, the damage the NPC has inflicted on the game user, the damage the game user has inflicted on the NPC, the hit rate of the NPC, and the hit rate of the game user, The game level may indicate the user's game level.

That is, the game level of the game user may be determined according to the level classification of the level determined in S110 as a result obtained while the game user plays the game for a predetermined time.

In addition, the game pattern may include game behavior and game state of the character controlled by the game user.

Here, the game action of the user represents all operations or actions of the game user character that the user has performed on the game, and the game state includes all information except for the character that the user operates in the game.

In the step of storing data according to the game pattern of the game user, the game user may compete with the basic NPC, and store the game pattern for a preset time.

Next, the level of the NPC AI according to the game level of the game user is determined (S130). In this case, the level of the NPC AI is determined based on the game level of the game user, whether the level of the NPC AI is lower than that of the game user, the same level as the game user, or higher than the game user.

The level of NPC AI determined at this stage allows you to determine the NPC AI that matches the user level of the game.

In addition, the level of the NPC AI determined in this step may be determined before the step S120 whether to have the same level as the game user, have a weak level, or have a strong level before knowing the game user's level in advance.

Subsequently, a parameter for the NPC AI is generated according to the stored data and the determined level of the NPC AI (S140). In this case, the parameter is a value assigned to operate the intelligent NPC, and means a value or information that is changed to intelligently operate the NPC.

Here, the generation method for generating the parameter for the NPC AI will be described again below.

Finally, the generated parameter is applied to the game to provide an NPC AI according to the user level (S150). At this time, the NPC AI applied to the game is set to the set level.

As shown in FIG. 2, the generation method S140 of generating a parameter for the NPC AI is as follows.

First, a parameter for changing a level in the NPC AI is selected (S141). In this case, as illustrated in FIG. 3, when the parameter is implemented as a finite state machine (FSM), a parameter that can change a parameter that can be changed, such as a parameter that indicates a distance or a parameter that indicates an attack time, may be selected. Can be.

In Figure 3, the distance and the attack time are shown, but numerical parameters such as the number of attacks and the strength of the force can be selected.

Subsequently, the above-selected parameter is initialized (S142). In this case, initialization of the parameter means initialization of each individual in the genetic algorithm. The object in the present invention includes the parameters selected in the previous step. Create objects as many as the number of objects selected in advance, and then initialize the values of the parameters included in each object to predefined or random values.

Subsequently, the fitness of the level corresponding to the data and game characteristics of the game user's game pattern stored above is calculated (S143). At this time, the method of calculating the fitness is as follows.

First, inputting data that stores a certain amount of time of the game user, the game is executed to create a function that calculates the value of the criterion that determines the level selected in the previous step.

This function does not need to include elements such as graphics, sounds, networks, etc. that accompany the game's movements; it only needs to calculate the reference values.

If such a function is created, the result value is calculated by reflecting the stored data and the value of each object in the function that calculates the reference value. This result is compared with the level of the AI level to be created by comparing the level of the level as shown in [Table 1], and the fitness is selected in inverse proportion to the difference with the level of the AI to be made.

Subsequently, a parent is arbitrarily selected in proportion to the goodness of fit, and a parameter is calculated by a genetic operator (S144). Here, the parent means two entities selected to select children of the next generation.

The method of arbitrarily selecting a parent is selected using one or more of a proportionality selection method, a roulette selection method, an expectation value selection method, a ranking selection method, and a tournament selection method. Means to select a parent.

Genetic operators also employ one or more of crossover, mutation, inversion, displacement, duplication, addition, and deletion to compute parameters. .

Brief description of the hybridization and mutation of the genetic operator is as follows.

Crossing is the creation of a new individual by partially swapping chromosomes between two parameters. At this time, the trait of the parent should be properly inherited to the offspring, which is called character preservingness or crossover neighborhood. As shown in FIG. 4, one-point breeding is a method of exchanging contents after selecting an arbitrary position. For example, two-point crossover represents a method of selecting two arbitrary locations and exchanging content between them.

In this way, by changing their positions, new parameters (objects) are created.

A mutation is a change in the value of an allele by applying a constant probability of mutation for the gene at each locus in the individual. It is a kind of local random search that creates a new object close to the object. It is also used as a means of maintaining diversity by recovering lost genetic traits in a population.

If the goodness of fit is equal to or greater than a preset value, a calculated parameter is generated (S145). At this time, if the goodness-of-fit is less than the preset value, the parameter operation is repeatedly performed until the goodness-of-fit is more than the preset value.

[Example]

The following describes the application of the present invention as described above through a two-player competitive action game such as a game street fighter generally known.

As shown in Fig. 5, the game Street Fighter is executed and the following assumptions are made.

(At this time, the game Street Fighter shown in Figure 5 is a copyrighted game, but was adopted to help explain the present invention.)

First, the male character on the left is the character controlled by the game user, and the female character on the right is the NPC.

Second, each character's health gauge ranges from 0 to 100.

Third, NPC is implemented with FSM.

Fourth, the male character, which is the character of the game user, is less skillful than the female character, which is an NPC, and takes much damage from the female character.

Fifth, we want AI that lowers the characters of NPCs to the same level as game users.

In order to provide NPC AI according to the game user level based on the above assumptions, first, the level level is determined according to the characteristics of the target game. At this time, divided into three stages based on the remaining strength of the NPC after the confrontation for a certain time as shown in Table 2 below.

<-50 -49 to 50 > 51 Low level Normal level High level

Thereafter, the game level is determined by the game user's game play for a certain time and the game pattern is stored. Here, the game level is the game user's normal level if the difference is the remaining health of the NPC-the remaining health of the game user after 40 minutes, the game user corresponds to the normal level, and the user wants the same level as the game user at home. According to the normal level of the three steps.

When the NPC AI is generated using the above information, a parameter that may affect the level of the NPC AI is selected. In this case, the parameter may be represented by a number such as 2, 5, 3, etc. which may affect the level of NPC AI through FIG. 6.

The expressed parameter can be initialized to the initial values (2, 5, 3).

In addition, the game pattern of the game user stored for a predetermined time is input to the game to calculate the set level. Since the closer to the NPC AI level to be generated, the higher the fitness, the farther the fitness is set lower.

For example, if the set level is high, the fitness is low. If the level is normal, the fitness is high. At this time, it can be set as a level or it can be calculated with a fitness gauge.

Also, parents are randomly selected in proportion to the goodness of fit. At this time, the parent may be selected randomly in proportion to the ranking of the goodness of fit, or the parent may be selected in the form of a tournament.

If the fitness is calculated as shown in the following [Table 3] according to the individual, as shown in Figure 7, the parent is randomly selected according to the ratio of the fitness. In the present embodiment, but shown in accordance with the ratio, the ratio may vary depending on the goodness of fit.

Object number Fitness One 10 2 20 3 5 4 45 5 20

The selected parents are computed with crossover and mutation operations according to the genetic algorithm. Here, in the case of the crossover operation, since there are three values in the parameter, a one-point crossover operation was used. In the case of the mutation operation, each value is an integer, and thus, each value is performed according to the probability, and is randomly added to or subtracted from each value. At this time, the range of the random value is set to -5 to 5.

Here, when the best fit is found, create the parameters and apply them to the game.

 Therefore, the present invention has the effect that the development time is reduced by the game developer does not implement the algorithm for all cases, and can provide the interest of the game user by providing an artificial NPC according to the level of the game user It works.

In addition, the present invention can generate and obtain NPC behaviors in various cases even if the NPCs are not matched to the level of game users, thereby providing a large number of game users with NPCs having various behavior patterns.

The NPC AI providing method according to the game user level as described above is not limited to the configuration and operation of the embodiments described above. The embodiments may be configured so that all or some of the embodiments may be selectively combined so that various modifications may be made.

Claims (8)

In the method of generating NPC AI (Non Player Character Artificial Intelligence) according to game user level using a genetic algorithm (Genetic Algorithm),
Determining a level corresponding to a game characteristic;
Determining a game level of the game user and storing data according to a game pattern;
Determining a level of NPC AI according to the game level of the game user;
Generating a parameter for the NPC AI according to the stored data and the determined level of the NPC AI; And
And applying the generated parameter to the game to provide an NPC AI according to a user level.
Generation method for generating a parameter for the NPC AI,
Selecting a parameter capable of changing a level in the NPC AI;
Initializing the selected parameter;
Calculating a goodness of fit of a level suitable for game characteristics and data according to the game pattern of the game user stored above;
Calculating a parameter by a genetic operator by arbitrarily selecting a parent in proportion to the goodness of fit; And
And generating a calculated parameter when the goodness of fit is equal to or greater than a preset value.
The method according to claim 1,
The game pattern is NPC AI providing method according to the game user level, including the behavior of the character and the game state controlled by the game user.
The method according to claim 1,
The game level is determined according to the game user level determined by at least one of the score of the NPC, the score of the game user, the damage the NPC has done to the game user, the damage the game user has done to the NPC, the hit rate of the NPC, and the hit rate of the game user. How to provide NPC AI.
delete The method according to claim 1,
The parameter means a value or information that is variable to intelligently behave the NPC NPC AI providing method according to the user level.
The method according to claim 1,
If the goodness of fit is less than a preset value, repeatedly performing a parameter operation until the goodness of fit is equal to or more than a preset value.
The method according to claim 1,
The method of arbitrarily selecting a parent may be selected based on one or more of a suitability proportion selection method, a roulette selection method, an expectation selection method, a ranking selection method, and a tournament selection method.
The method according to claim 1,
The genetic operator may be based on a game user level that adopts at least one of crossover, mutation, inversion, displacement, duplication, addition, and deletion. How to provide NPC AI.

Images