JP3426571B2 - Rule generator - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a rule generation device, and in particular, for example, a first autonomous system reproduced in a virtual space.
Based on multiple first rules that define the character's behavior
Second autonomous character newly reproduced based on virtual space
Generate a plurality of second rules that define the behavioral patterns of
And a rule generator .

[0002]

[0003]

[0004]

[0005]

2. Description of the Related Art As a conventional technique for evolving an autonomous character reproduced in a virtual space, there has been a role-playing game in which a hero character improves his / her special skills and talents as he / she gains experience. Further, as a conventional technique for reproducing a manual character and an autonomous character in a virtual space, for example, a plurality of players have appeared, such as a soccer game, and one of them has to be manually controlled.

[0006]

However, in the above-mentioned role-playing game, the direction of evolution is predetermined and cannot be freely changed by the user. Further, in the soccer game, although the character to be manually controlled can be arbitrarily changed depending on the situation, there is a problem that the virtual space displayed on the monitor is flat and lacks power.

Therefore, a main object of the present invention is to provide a rule generation device capable of giving a freedom degree to an action pattern of an autonomous character newly reproduced in a virtual space.

[0008]

[0009]

[0010]

[0011]

According to the present invention, a virtual
Prescribes the action pattern of the first autonomous character reproduced in space
Newly reproduced in virtual space based on multiple first rules
Of a plurality of second autonomous characters that define the action pattern
The rule generation device that generates the second rule includes a plurality of first rules.
First acceptance that accepts the setting of the desired probability common to all
Means, locked or unlocked for the plurality of first rules
The second receiving means for receiving the individual setting of the locked state, the lock
First generating means for generating the second rule so as to have a predetermined relationship with the first rule set in the locked state , and unlocking
A second generation unit that randomly mutates the first rule set in the state with a desired probability to generate the second rule is provided.
And are characterized .

[0012]

[0013]

[0014]

[0015]

[Operation] The line of the first autonomous character reproduced in the virtual space
The motion pattern is defined by a plurality of first rules. First
The accepting means has a desired probability common to the plurality of first rules.
The setting is accepted, and the second accepting means sets a plurality of first rules.
For individual settings of locked or unlocked status
Stick to. The first generation unit is configured to lock the first state.
Generate a second rule so that it has a predetermined relationship with one rule,
The second generation means is configured to set the first loop set in the unlocked state.
Randomly mutate the rule with a desired probability to generate the second rule
To do. Second autonomous character newly reproduced in virtual space
The behavior pattern of the
Therefore, it is defined.

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023]

According to this invention, according to the present invention, when the first rule is set to the unlocked state, the second rule by mutating the first rule at random are generated, the second rules generated A second autonomous character having an action pattern based on is reproduced in the virtual space. For this reason, Ru can have a degree of freedom in the behavior of the autonomous character to be newly reproduced in the virtual space.

The above-mentioned objects, other objects, features and advantages of the present invention will become more apparent from the following detailed description of embodiments with reference to the drawings.

[0025]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT Referring to FIG. 1, a three-dimensional graphic device 10 of this embodiment includes a CPU 12. CPU 12
Memory 14 and disk drive 1 by bus 18
6 and is connected to the bus 18 and the I / F circuit 2
A display (video monitor) 22, a keyboard 24, a microphone 26, and a mouse 28 are connected via 0.

The disk drive 16 accesses a recording medium 30 such as a CD-ROM in response to a command from the CPU 12, and installs the three-dimensional graphic software recorded on the recording medium 30 in the memory 14. When the operator 24 inputs a predetermined command through the keyboard 24, the microphone 26 or the mouse 28, the CPU 12 processes the three-dimensional graphic software according to the given command. A three-dimensional virtual space corresponding to the process is reproduced on the display 22.

Specifically, the three-dimensional virtual space is a space as shown in FIG. In other words, one private house exists on the land surrounded by the sea, and a hill is formed at a position separated from the private house. There are two rocks on the hill, and one tree grows on one rock. Moreover, the disk floats on these rocks. Fruits (not shown) such as oranges, apples, and bananas are scattered on the land, and fish swim in the sea. In such a virtual space, for example, three types of characters such as "bird", "kappa" and "girl" as shown in FIG. 3 are reproduced. The manual mode is set for any one of the reproduced characters, and this character acts according to the command from the operator. On the other hand, an autonomous mode is set for the remaining characters, and this character takes an action pattern according to preset rules.

The display 22 displays an image of the virtual space viewed from the character in the manual mode. That is, when "bird", "kappa", and "girl" are present as shown in FIG. 3, and the manual mode is set for "girl", the image shown in FIG. Is displayed in. Also, "kappa"
When the manual mode is set to, the image shown in FIG. 5 viewed from “Kappa” is displayed on the display 22, and when the manual mode is set to “bird”, the image shown in FIG. 6 viewed from “bird” is displayed. 22 is displayed.

Referring to FIG. 7, each character has a visual sensor and a tactile sensor that are effective in the autonomous mode. The visual sensor is a sensor that detects what kind of object one is looking at, and the tactile sensor is a sensor that detects what kind of object one is touching.

With respect to the visual sensor, in addition to the horizontal and vertical viewing angles, the respective short-range and long-range reach ranges are set. Therefore, when the object is detected by the visual sensor, each character can determine not only whether the object has entered the viewing angle, but also whether the object is located at a short distance or a long distance. Here, for characters such as "girl" and "kappa" whose eyes face forward, the viewing angle should be set narrow. On the other hand, for characters such as "birds" whose eyes are facing left and right, the viewing angle should be set wider. As described above, since the visual sensor characteristic differs depending on the character, the visual sensor characteristic table shown in FIG. 8 is prepared for each character, and the viewing angle and the reach range are adjusted by this table.

According to FIG. 8, four items of "near distance sensor", "long distance sensor", "angle θ" and "angle φ" are prepared. Of these, the "short range sensor" and the "long range sensor" determine the short range and the long range, and the "angle θ" and the "angle φ" determine the horizontal and vertical viewing angles. The maximum reachable range for both the "short range sensor" and the "long range sensor" is 20 times the height of the character. On the other hand, with regard to the minimum reachable range, the “short range sensor” is 0, but the “long range sensor” is the maximum value of the reachable range actually set in the short range sensor. For viewing angle,
Both “angle θ” and “angle φ” can be set in the range of 0 degrees to 180 degrees. Each autonomous character recognizes the existence of an object by such a visual sensor.

On the other hand, when the tactile sensor is used, each autonomous character can detect not only the object with which it is in contact but also the place where it is. For example, when the character is on the land, it detects that the soles of the feet are in contact with the ground, determines that the character is standing on the land, and walks on the land at a speed that considers air resistance. When it detects that the body has come into contact with an obstacle such as a rock or a private house while walking, it judges that it cannot break through the obstacle in front of it and stops moving forward. In addition, when it jumps from land to the sea and detects that the whole body comes into contact with seawater, it determines that it is in the sea, changes the method of movement from walking to swimming, and considers the speed of movement in the sea considering water resistance. To decide. In this way, the autonomous character recognizes where it is now by the tactile sensor and behaves according to the recognition result.

An evolution criterion table shown in FIG. 9 is also prepared for each character. According to FIG. 9, there are three parameters of "fullness", "vigor" and "mood" as survival conditions, and the threshold value of each parameter (threshold value for determining whether or not survival is possible) is arbitrarily adjusted by the operator. To be done.

"Fullness" and "vigor" possessed by each character
The parameter values of “and mood” change with the action of the character. For example, if you find and eat your favorite food, your character will be full, cheerful, and happy. On the other hand, if the feed cannot be found for a long time, the character becomes hungry, tired and tired, and further displeased. The character dies (disappears) when any one of the parameter values of "satiety", "vigor" and "mood" decreases below the threshold of the survival condition.
To do. Here, the dying character is not limited to the autonomous character. That is, the character acting in the manual mode also dies when the parameter value falls below the survival condition.

Each character evolves upon death. That is, immediately after the character's death, a character that inherits any one of the remaining surviving characters (the probability that an excellent character is selected) is generated and reproduced in the virtual space. At this time, the gene of the next-generation character may have a mutation in the gene of the parent character. This mutation rate is determined by the mutation rate set in the evolution reference table of FIG. There are two mutation rate parameters, "point mutation" and "dominance mutation", and the threshold values of these parameters are also arbitrarily adjusted by the operator. The point mutation rate and the dominant mutation rate will be described later.

The gene program shown in FIG. 10 is set for each character. This gene program sets a rule menu table 48 for arbitrarily creating a rule for determining a character's action pattern, and whether or not to mutate each rule when creating a character (mutation lock / unlock). A rule table 52 for storing rules created by operating the mutation lock setting table 50 and the rule menu table 48 (including mutation lock / unlock), and addition of rules to the rule table 52 And an icon 56 for instructing to delete a rule from the rule table 52.

The rule menu table 48 has three items of "input", "state", and "output". There are three parameters relating to "input": the type 32 of the sensor to be activated, the detection object 34 by the activated sensor, and the detection distance 36 of the visual sensor. Further, there are four parameters relating to the state: the place 38 where one is present, the feeling of fullness 40, the vitality 42, and the mood 44. Furthermore, as a parameter related to output, there is a character action 46.

Here, there are "1: visual sensor" and "2: tactile sensor" as the sensor type 32, and "A: private house door", "B: another character", "C" as the detection object 34. : Rock, "D: Disc", "E: Sea surface",
There are "F: apple", "G: mandarin orange", "H: banana" and "I: fish", and the detection distance 36 includes "1: close" and "2: far". The setting of the detection distance 36 is effective only when the visual sensor is activated.
As the place 38, “A: private house”, “B: undersea”,
There are "C: on the rock" and "D: on the disc", and the feeling of fullness 40 is "1: full" and "2: hungry".
The vitality 42 includes "1: full of energy" and "2: tired", and the mood 46 includes "1: happy" and "2: moody". Furthermore, as for the operation,
"A: Walk around", "B: Approach", "C: Backward", "D: Rest", "E: Enter and leave", "F: Fly", "G: Hirou", "H: Put", " There are “I: eat”, “J: hit” and “K: escape”. By selecting a desired item from the rule menu table 48 with the mouse 28, a rule for determining the action pattern of the autonomous character is created.

It is not always necessary to select parameters for the sensor type 32, the detection distance 36, the place 38, the feeling of fullness 40, the vitality 42, and the energy 44. In this case, the parameter value that is not selected becomes "*", which functions like the wild card of Trump. For example,
If "A: Private house" is set for "Place" and none of the parameters for "Feeling fullness", "Vitality", and "Vigor" is selected, the character will feel fullness when moving to a private house, Take a predetermined action (e.g., eat) in response to detecting a predetermined object (e.g., banana) regardless of vitality and energy.

By operating the mutation lock table 50, it is possible to arbitrarily set whether or not to cause "point mutation" and "dominance mutation" for each created rule. "Point mutation" means a random mutation of the content indicated by multiple items that make up a rule, and "dominance mutation" is the priority (rank) of the set rules.
Means a random mutation of. Mutation lock table 5
At 0, lock / unlock can be set for each of the parameters of “point mutation” and “dominance mutation”. This setting operation is also performed by the mouse 28.

When only "point mutation" is locked,
The content of locked rules is not altered, but the order of rules may be altered. On the contrary, when only the "dominance mutation" is locked, the locked rule order is not changed, but the rule content may be changed. When both "point mutation" and "dominant mutation" are locked, neither the content nor the rank of this rule is changed. On the other hand, when neither "point mutation" nor "dominance mutation" is locked, both the content and the rank of this rule may be mutated.

The parameter value when only "point mutation" is locked is "1", which means "dominant mutation".
The parameter value when only one is locked is "2".
The parameter value when both the point mutation and the dominant mutation are locked is "3", and the parameter value when neither the point mutation nor the dominant mutation is locked. The value is "*". That is, it is not always necessary to select parameters for the mutation lock table 50. At this time, mutations are treated in the same manner as wildcards, and the corresponding rules are subjected to both point mutations and dominance mutations. It

After the rule creation and mutation setting are completed in this way, when the operator clicks the "add" icon with the mouse 28, the created rule and mutation parameters are displayed in the rule table 52.
Added to. On the other hand, when any of the rules set in the rule table 52 is selected and the “delete” icon is clicked, the selected rule is deleted from the rule table.

For example, the sensor type 32 and the detected object 34
And “1. visual sensor” from each of the detection distances 36,
Select "H: Banana" and "1: Close", place 3
8. Select “A: private house”, “2: hungry”, “2: tired” and “2: grumpy” from each of the feelings of fullness 40, vitality 42, and energy 44, and from action 46, “I: eat”.
And lock only “point mutation”, and then click the “Add” icon 54 with the mouse 28, “1H1: A222: I: 3” is displayed in the rule table 5
Set to 2.

Each autonomous character determines a behavior pattern based on a plurality of rules set in its own rule table 52. If you decide your behavior according to probability theory,
First, matching is performed between the current character input and state and each set rule input and state, and the matching score of each rule is obtained. At this time, the matching score is weighted according to the superiority, but when the rule includes "*", a low score is given. Then, find the sum of each matching score,
The ratio of each matching score to this total is calculated.
Then, each rule is weighted according to the calculated ratio, and one of the rules is randomly selected in consideration of this weighting. The autonomous character takes an action according to the extracted rule.

On the other hand, in the case of determining the action pattern according to the determinism, the matching score is calculated in order from the rule having the highest precedence, and when the matching score exceeds the predetermined threshold, the rule corresponding to this matching score is set. adopt.
The autonomous character takes an action according to the adopted rule.

The CPU 12 is specifically shown in FIGS.
The flow chart (3D graphic software) shown in is processed.

First, referring to the main routine shown in FIG. 11, in step S1, the three-dimensional virtual space shown in FIG. 3 is formed in the work area (not shown) of the memory 14. In step S3, the number of individuals (surviving individuals) n of the characters reproduced in the three-dimensional virtual space is set to an initial value No, and in the following step S5, the number of individuals (disappearing individuals) d of the disappeared (dead) characters is set to " 0 ". The initial value No is, for example, “3”. Moreover, since each character dies when it does not satisfy the predetermined survival condition, the number of individuals d is incremented according to the death of each character. Subsequently, n autonomous characters are reproduced in the three-dimensional virtual space in step S7, a manual mode is set for a predetermined character in step S9, and a virtual camera is placed above the head of the character in manual mode in step S11.

In step S7, for example, "bird", "kappa", and "girl" autonomous characters are reproduced one by one, and identification numbers "0", "1", and "2" are assigned to each character. In step S9, the manual mode is set to "girl", for example. For this reason,
The “birds” and “kappa” act autonomously in the three-dimensional virtual space, while the “girl” acts in the three-dimensional virtual space in response to a command from the operator. Further, the virtual camera is initially placed above the "girl", and the display 22 reproduces the three-dimensional virtual space viewed from the "girl".
“Tori”, “kappa” and “girl” are shown in FIG.
Display 2 when present in a private house as shown in
The image shown in FIG. 4 is displayed on 2.

In step S13, it is determined whether or not an operation command is given by the operator, and in step S17, it is determined whether or not another character is selected by the operator. Both the input of the motion command and the selection of another character are performed through the keyboard 22 or the microphone 26.

When the operation command is input, YES is determined in step S13, and the operation process is performed in step S15. For example, when the "↑" key provided on the keyboard 24 is pressed, the "girl" is moved forward and the same keyboard 2
When the "↓" key provided in 4 is pressed, the "girl" is moved backward. If a voice message “pick up” is input through the microphone 26 when the fruit is falling on the ground, the “girl” can pick up the fruit. When the operation process is completed, the process returns to step S13. In this way, the character to which the manual mode is set behaves in accordance with the operator's command.

When another character is selected, step S
In step S19, the autonomous / manual mode setting is changed. That is, the character whose manual mode is currently set is changed to the autonomous mode, and the selected character is changed from the autonomous mode to the manual mode. When the mode changing process is completed, the process returns to step S11. Therefore, when the "girl" is set to the manual mode and "the operator yells into the microphone 26, for example," kappa ",
The autonomous mode is set for "girl" and the manual mode is set for "kappa". Therefore, the "girl" operates autonomously, and the "kappa" operates according to the instruction of the operator. Further, the virtual camera is placed above the "kappa" by the process of step S11. Therefore, “bird”, “kappa”, and “girl” are shown in FIG.
When it is positioned as shown in FIG.
The images of “bird” and “girl” shown in are displayed.

In step S21, it is determined whether or not an operator gives an end instruction. If NO here, the process returns to the step S13, but if YES, the step S2.
In step 3, a predetermined end process is performed, and then the process ends.

FIG. 12 is a timer interrupt routine which is executed every predetermined period during the processing of the main routine. First, in step S31, the count value i is set to "0", and in step S33, it is determined whether or not the i-th character satisfies the survival condition. As described above, the evolution reference table shown in FIG. 9 is assigned to each character, and the feeling of fullness, vitality, and mood of each character change with action. In step S33, the satiety, vitality, and mood of the focused character are compared with the threshold value,
If any one item is below the threshold value, it is considered that the survival condition is not satisfied, and the process proceeds to step S35. In step S35, the character of interest is erased from the virtual space, in step S37 the number d of extinguished individuals is incremented, and when the processing of step S37 is completed, step S3
Proceed to 9. On the other hand, feeling full about the character of interest,
If all items of vitality and mood are above the threshold,
Assuming that the survival condition is satisfied in step S33,
The process directly proceeds to step S39.

In step S39, the count value i is "n-
If it is NO, the step is judged.
After incrementing the count value i in S41,
Return to step S33. Survival conditions for all characters
When the determination process of is completed, YES is determined in step S39.
If the random number r is determined in step S45,
At step S47, r characters are generated. here,
Random number r is 0 ≦ r ≦ N_{MA X}-(N-d) condition is satisfied
Is a randomly determined integer, and the number of surviving individuals n is N
_MAXNever exceeds. Finish the character generation process
Then, in step S47, the number of surviving individuals n is updated to “n + r”.
After renewing, return to the main routine.

The generation process (evolution process) of r characters in step S45 follows the subroutine shown in FIGS. First, in step S51, the value of the random number r is determined, and if r = 0, the process directly returns to the subroutine shown in FIG. On the other hand, if r> 0, step S
The count value i is set to "0" at 53, and i is set at step S55.
The evaluation value Fi of the th character is calculated. The calculated evaluation value Fi increases according to the number of foods that the i-th character can find per unit time. Step S57
Then, it is determined whether or not the count value i has reached "n-1" (whether or not the evaluation value Fi has been calculated for all the living characters). If NO, the count value i is incremented in step S59. Returns to step S55. On the other hand, if YES in step S57, step S57
In step 61, all evaluation values Fi (0 ≦ i ≦ n−1) are integrated to calculate the total evaluation value ΣF.

In step S63, the count value i is set to "0" again, and in step S65, the relative evaluation value fi of the i-th character is calculated according to equation 1.

[0058]

## EQU00001 ## Then, when the relative evaluation value fi is obtained for all the characters by performing the same processing as steps S57 and S59 in steps S67 and S69, step S71
Create a pie chart as shown in FIG. Characters 0 to n-1 (living characters) are assigned to the n fan-shaped areas formed in the pie chart, and each fan-shaped area has a relative evaluation value fi (0≤i≤n).
-1) It has an area according to. According to FIG. 17, "kappa", "bird" and "girl" exist in the virtual space one by one, the evaluation of "kappa" is highest, and the evaluations of "bird" and "girl" are almost the same. is there.

When the pie chart is created, step S73
The count value i is set to "0" with, and any one character is selected from the pie chart in step S75. Specifically, a character assigned to one of the fan-shaped areas is selected by shooting an arrow on the rotating pie chart.
As a result, the type of the generated (evolved) character (whether it is “girl”, “bird”, or “kappa”) is specified.

When the type of character is specified, the gene program which is the base of the mutation process is specified in step S77. The specified gene program is the program assigned to the character selected in step S75. When the gene program is specified, point mutation processing and superiority mutation processing are performed in steps S79 and S81. That is, the content and order of the rules set in the rule table 52 of the identified gene program are randomly changed to create a new gene program. The created gene program is a program for determining the action pattern of the character identified in step S75. In step S83, the character generated in this way is added to the three-dimensional virtual space, and when the addition processing is completed, the count value i is compared with "r-1" in step S85, and if i <r-1 then step After incrementing the count value i in S87, the process returns to step S75. Therefore, the count value i is "r-
The processes of steps S75 to S85 are repeated until the number reaches 1 ". When r characters are added to the virtual space, YES is determined in step S85, and the process returns to the subroutine shown in FIG.

The point mutation processing in step S79 follows the subroutine shown in FIG. First step S
The count value i is set to "0" at 91, and step S93
Then, the setting state of the point mutation relating to the i-th rule is detected from the rule table 52 of the gene program. In step S95, it is determined whether the detected setting state is locked or unlocked. If the setting state is locked, step S111 is performed. If the setting state is unlocked, step S97 is performed.
To each.

In step S97, the count value j for counting the items (8 in total) that make up each rule is set to "0".
, And in the subsequent step S99, a random number r (0 ≦ r ≦ 1
The real number of) is determined at random. When the random number r is determined, the point mutation rate μ is detected from the evolution reference table shown in FIG. 9 in step S101, and the random number r is compared with the point mutation rate μ in step S103. Here, if μ ≦ r, the process directly proceeds to step S107, but if μ> r, the content of the jth item of the rule of interest is randomly changed in step S105, and then the process proceeds to step S107. In step S107, the count value j is "m-1 (= 7)".
If NO, step S1
After the count value j is incremented at 09, the process returns to step S109. When the processes of steps S99 to S109 are executed for all items of the rule of interest, the process proceeds from step S107 to step S111.

In step S111, the count value i is "k".
It is determined whether or not it has reached -1 ". Here," k "is the total number of rules set in the rule table 52, and if NO in step S111, the count value i is incremented and the process returns to step S93. Therefore, i = k
The processing of steps S93 to S113 is repeated until it becomes -1, and if YES is determined in step S111, it is considered that the point mutation processing for all the rules is completed, and the processing returns to the subroutine shown in FIG.

The dominant mutation processing in step S81 follows the subroutine shown in FIG. First, in step S121, the count value i is set to "0", and then step S
At 123, the setting state of the dominant mutation is detected for the i-th rule. In step S125, it is determined whether the detected setting state is locked or unlocked, and if it is in the locked state, the process proceeds directly to step S141.
If it is in the unlocked state, the process proceeds to step S141 through the processes of steps S127 to S139.

In step S127, the random number r (0≤r≤1
9) is randomly determined, and in step S129, as shown in FIG.
The dominant mutation rate ν is detected from the evolution standard table shown in. When the random number r and the dominant mutation rate ν are obtained,
In step S131, the respective parameters are compared. If ν> r here, random number m (0 ≦ m in step S133)
≤ k-1) is randomly determined, and step S135
In step S137, the setting state of the dominant mutation is detected for the m-th rule, and whether the detected setting state is locked or unlocked is determined in step S137. If the determination result indicates the locked state, steps S133 and S135
When the discrimination result shows the unlocked state, the i-th rule and the m-th rule are exchanged in step S139. When the replacement process is completed, the process proceeds to step S141. When ν ≦ r is determined in step S131, the process proceeds to step S141 without performing the processes of steps S133 to S139.

In step S141, the count value i is "k".
If it is NO, the count value i is incremented in step S143 and then the process returns to step S123. On the other hand, YES in step S141.
If so, the process returns to the subroutine shown in FIG.

As can be seen from the above description, one character acting in the manual mode (manual control method) and a plurality of characters acting in the autonomous mode (autonomous control method) are reproduced in the three-dimensional virtual space. The virtual camera is arranged above the head of the character in the manual mode, and the display shows the three-dimensional virtual space photographed through the virtual camera. Here, when any one of the characters in the autonomous mode is selected, the selected character in the autonomous mode switches to the manual mode,
A character in manual mode switches to autonomous mode. The placement of the virtual camera moves above the character in the manual mode according to the mode switching. As described above, since the arrangement of the virtual camera is also changed according to the mode switching of the character, the virtual space displayed on the monitor can be dynamically changed.

When a character reproduced in the virtual space disappears without being able to satisfy the survival condition, the rule for determining the action mode of any surviving character is determined according to the lock / unlock setting. Mutated and the mutated rules are applied to the evolved character. In other words, if the rule is set to the locked state, this rule will be applied to the evolved character as it is, but if the rule is set to the unlocked state, it will be generated by randomly mutating this rule. The rules are applied to the evolved character. The evolved character acts in the virtual space based on the generated rules.

As described above, if the rule for determining the action pattern of the disappeared character is set to the unlocked state, this rule is randomly mutated to generate a new rule, and the new rule is generated based on the generated rule. A character having a different behavior pattern is reproduced in the virtual space. Therefore, the behavior of the character newly reproduced in the virtual space can be given a degree of freedom.

Here, the probability (mutation rate) of mutating the unlocked rule can be arbitrarily adjusted, and the rule applied to the evolved character is mutated based on the adjusted mutation rate. It

Each rule can be set to either the locked state or the unlocked state in terms of content or rank. If the content is unlocked, the content of the rule is randomly mutated, and if the order is unlocked, the order of the rule is randomly mutated.

In this embodiment, the three-dimensional virtual space is reproduced on a single display.
The three-dimensional graphic devices may be connected to each other via a communication network to reproduce a common three-dimensional virtual space on the display of each three-dimensional graphic device.

[Brief description of drawings]

FIG. 1 is a block diagram showing an internal configuration of the present invention.

FIG. 2 is an illustrative view showing one example of a three-dimensional virtual space.

FIG. 3 is an illustrative view showing another example of a three-dimensional virtual space.

FIG. 4 is an illustrative view showing a three-dimensional virtual space viewed from a girl.

FIG. 5 is an illustrative view showing a three-dimensional virtual space viewed from a kappa.

FIG. 6 is an illustrative view showing a three-dimensional virtual space viewed from a bird.

FIG. 7 is an illustrative view showing a visual sensor and a tactile sensor set for a girl.

FIG. 8 is an illustrative view showing one example of a visual sensor characteristic table.

FIG. 9 is an illustrative view showing one example of an evolution criterion table.

FIG. 10 is an illustrative view showing one example of a gene program.

11 is a flowchart showing a main routine of three-dimensional graphic software applied to the embodiment in FIG.

12 is a flowchart showing a timer interrupt routine of the three-dimensional graphic software applied to the embodiment in FIG.

FIG. 13 is a flowchart showing a part of a character processing routine of three-dimensional graphic software applied to the embodiment in FIG. 1;

FIG. 14 is a flowchart showing another part of the character processing routine of the three-dimensional graphic software applied to the embodiment in FIG. 1;

FIG. 15 is a flowchart showing a point mutation processing routine of the three-dimensional graphic software applied to the embodiment in FIG.

16 is a flowchart showing a superiority mutation processing routine of three-dimensional graphic software applied to the embodiment in FIG.

FIG. 17 is an illustrative view showing one example of a pie chart applied to the embodiment in FIG. 1;

[Explanation of symbols]

10 ... Three-dimensional graphic device 12 ... CPU 14 ... Memory 16 ... Disk drive 22 ... Display 24 ... Keyboard 26 ... Mike 28 ... Mouse 30 ... Recording medium

Front Page Continuation (72) Inventor Takao Nakaguchi 2-2, Kodai, Seika-cho, Soraku-gun, Kyoto Prefecture AT R Human Information Communication Laboratory, Inc. (56) References: JP-A-10-307804 (JP) , A) JP-A-10-307806 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G06T 15 / 70,17 / 40 G06N 3/00-3/12

Claims (4)

(57) [Claims] 1. A first autonomous character reproduced in a virtual space.
Based on a plurality of first rules that define the behavior pattern of
Behavior of the second autonomous character newly reproduced in virtual space
Rule generation for generating a plurality of second rules that define the style
The device receives a desired probability setting common to the plurality of first rules.
First accepting means for imposing, locked state or unlock for the plurality of first rules
Second receiving means for receiving the individual setting of the click state, first generating means for generating a second rule to have a first rule and a predetermined relationship set in the locked state, and before
The first rule set to the unlocked state is set to the desired rule .
Characterized in that it comprises a second generating means for generating a second rule randomly mutated with probability, rule generation device. 2. The content of the first rule in the second acceptance means
About the setting of the locked state or the unlocked state
It includes a content setting accepting means for accepting a constant, the second generating means includes a contents mutation means for mutated the contents of the first rules that are set in the unlocked state by the content setting accepting unit, according to claim claim Rule of 1
Generator . 3. A rank is assigned to the plurality of first rule, the second receiving unit receives setting of the lock state or the unlock state for ranking before Symbol first rule
Include that order setting accepting means, said second generating means includes a ranking mutation means for mutated the sequence for the first rule set in the unlocked state by the order setting accepting unit, according to claim 1 or 2, wherein The roux
Generator . Wherein said first determination means for autonomous character to determine whether a predetermined condition is satisfied, extinguished means to extinguish from the virtual space the first autonomous character when not satisfy the predetermined condition, and prior to serial further comprising reproduction means for reproducing in the virtual space before Symbol second autonomous character can first autonomous character disappears, claims 1
3. The rule generation device according to any one of 3 above.