JP3457427B2 - Method and apparatus for generating motion in virtual space - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention, under a predetermined environment model,
The present invention relates to an action generation method and an action generation device for causing a virtual human model to perform an action with individual differences and displaying the action.

[0002]

2. Description of the Related Art Conventionally, designers have come to use various viewpoints for designing public places such as station premises and shopping malls, and working places such as monitoring rooms for monitoring and controlling power plants and factories. I went based on the sketches I drew.

As a checkpoint for designing public places, a destination display board (especially a bulletin board for guiding an emergency exit or an emergency staircase during an emergency evacuation) is easy to see and is located at an easy-to-understand position, or a fare table for buying a ticket. Or whether the map is easy to see. An important checkpoint is whether there are sufficient routes to reach the emergency exits and stairs, and whether there is sufficient safety in reaching the escalator and boarding.

In designing a work place such as a monitoring room, whether a large screen display on the wall can be seen from the position where the user sits on the monitoring table, or a plurality of workers arrive when operating the site panel in the event of an abnormality. An important checkpoint is whether or not sufficient routes are secured.

The designer draws a sketch while taking these check points into consideration. However, even if the confirmation of the visibility of a large screen is only one, in order to deal with workers of different heights, sketch images must be drawn from various points of view, and the work becomes enormous, which can be dealt with very much. is not.

In order to solve these problems, an environment is created using CG (computer graphics),
A method has been adopted in which images taken from various viewpoints are time-lapsed and recorded to create an animation so that walk-through can be performed.

[0007] However, in the walkthrough by the normal CG, it takes several minutes to generate one image, and it takes a lot of labor and computer power to create an animation for several minutes, which is different. It was difficult in terms of cost to deal with confirmation of visibility from height.

In order to solve such a problem, a human model is virtually generated, and from the viewpoint of this human model,
Alternatively, a system has been developed that enables confirmation of visibility from various viewpoints by observing the working state of this human model (reference document, Miwako Doi et al., “For virtual subjects. Motion Generation Method ", N in the 9th session
ICOGRAP Thesis Contest Proceedings, Japan Computer Graphics Association, 1993).

[0009]

In the above-mentioned technique, the virtually generated human model can perform exactly the same behavior when the work is designated in a certain environment. In other words, the above-mentioned technology is a system that allows confirmation of visibility from various viewpoints, assuming that all information known by the system is also known by a human model. Human models can only do the same actions. And, in a work place such as a surveillance room, it is considered that a person who has been trained to some extent works.
There is no problem even with the above-mentioned technique in which the behavior of human models having different viewpoints and physiques can be observed while performing the same action.

However, in places where a large number of unspecified persons act, such as public places, the actual actions of humans do not differ because of individual differences. A problem in designing a work place is a problem that is completely different from what the designer thinks, that is, a problem that the designer does not see even when he thinks.

Conventionally, there has not been a human model capable of performing an action that the designer does not consider, that is, a method or an action generating device for generating an action having individual differences.

[0012]

According to the present invention, based on the individual difference data which is a characteristic of each human model, the arrangement of objects in the virtual space is used to determine the arrangement of objects to be recognized in the visual field by the human model. Information about an external memory model
From the action strategy, which is a rule for the human model to perform the action that the human model should act, based on the determined external memory model, the action generating unit determines the action that the human model can take. , The action generation process of the action of the determined human model according to the action pattern that is a rule for the action of the human model to generate action.
It is characterized in that it is generated by stages .

[0013]

According to the present invention, the method of creating the external memory model of the human model differs depending on the individual difference. In other words, the presence or absence and arrangement of objects in the virtual space recognized by the individual human models are different, so that even if the same action strategy is used in the same environment model, the individual human models take different actions. No special knowledge about psychology, CG, etc. is required because it is only necessary to manipulate variables that are easy to understand, such as attention level and height, in order to generate actions that differ among individuals.

Further, the behavioral patterns such as the walking pitch and the speed can be changed by the individual difference even with the same walking motion, so that the remarkable behavior due to the individual difference can be generated.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The operation of the present invention will be described below with reference to the drawings. FIG. 1 is a schematic configuration diagram of the present invention. The environment model storage unit 1 is, for example, a storage unit for storing an environment as a work target as shown in FIG. 2, that is, an arrangement of objects in a virtual space. FIG. 2 considers a case where the layout of an automatic ticket gate is evaluated as an example of a social human interface (usability of a device used by an unspecified number of users). The three-dimensional shape and the position of the object arranged in the work environment as shown in FIG. 2 are stored in a format in which the position of each object as shown in FIG. 3 is represented by the center coordinates in the virtual space and the coordinates of the circumscribed rectangular parallelepiped. ing. Figure 2
In the example, there are two ticket gates, but since the shape data to be read is the same for each, the same pointer P0 is entered in the column of points to the shape data, but the ticket gate 0 and the ticket gate are the objects. It is different like 1 . When an object has a movable part, the movable part is stored as a separate object for each part.

The individual difference data storage unit 4 includes various variables and a plurality of values for causing a virtual user, which is a user's alter ego, to perform actions with individual differences (hereinafter referred to as individual difference data).
I remember. Specifically, people who carefully judge their surroundings, such as people who look at it once in a wild rush and then act blindly, such as human habits and habits that come out at the timing of looking around. I remember.
For example, it is stored in the format shown in FIG. 4 below.

FIG. 4 shows, for each individual, the name of the file in which the external memory model, which is the layout of the objects in the recognized virtual space, is stored (this may be a pointer to the storage location), and the characteristics. Is remembered. Figure 4
In the example, "normal" (normal) "hasty"
For example, three types of attribute values are written, such as "quiet".

These attributes control, for example, the speed of action. That is, if it is normal, for example, if the user looks around 10 times a second, it indicates that this is normal attention. By looking around, the information on the outside world that has become clear is stored in the individual's outside world memory model.

On the other hand, hasty is a stern character, which means that the user looks around for 20 times a second and frequently checks the surroundings. Since the frequency of looking around is high compared to normal, more information can be obtained and the latest information can be kept.

On the other hand, the quiet character has a relaxed personality, and therefore, for example, looks around only five times a second. That is,
Since quiet is laid-back, it sees only half of normal, so the corresponding external memory model of that person is based on older information about the virtual environment.

Specific values of normal, hasty, and quiet can be stored as a ratio to the parameter to be controlled, as shown in FIG. 5, for example. To change values or add new attributes, edit the file in FIG.

The information acquisition control unit 3 refers to the characteristics of the individual stored in the individual difference data storage unit 4 as shown in FIG. 4, and stores the contents stored in the external world model storage unit 2 shown below. I will update.

In the external world memory model storage unit 2, for example, in the format as shown in FIG. 6, among the objects in the work environment by the virtual user, the object recognized at the present time has a state of “known”,
Objects that are not recognized are remembered because the state is "unknown".

In the example of FIG. 6, the states are only known and unknown, but whether the object is movable (movable / unmo
vable), whether it is an operable object (operatable
/ unoperatable), etc. various states can be stored. In addition to this, the target is within reach (reachable
) Or not (unreachable), but since this changes depending on the current position of the human model, it is not stored in the format shown in FIG. Instead, the action generator 6 changes the reacha according to the distance between the human model and the target.
It is designed to judge ble / unreachable.

The behavioral strategy storage unit 5 stores how to achieve a task (goal) when a task (goal) is given, how to achieve the task (strategy), that is, to achieve the goal. The rules are remembered. For example, a strategy for achieving a certain work (goal) is stored in a format as shown in FIG. According to this strategy, the action generation unit 6 determines the action as follows.

For example, "gaze" has an object (OBJ), and knowing the object is "gaze".
Has become a condition for. By "gaze",
Since the target state does not change, the resulting item is null.

Further, "touch" has an object (OBJ), and it is conditioned not only that the object is known but also that the object is within reach (reachable). In the case of mere “walking”, there is no target (NULL), and there is no condition (NULL), but in the case of “heading”, unlike the “walking”, the target (OBJ) to go to is known (known).
) Things and unreachable are needed. Furthermore, by "going", the state of the target is unre
From achable to reachable (within reach).

"Finding" is different from "gaze", and the condition is that the object (OBJ) is unknown. In other words, "finding" is the process of searching for and knowing something that you do not know.

The style storage unit 7 stores the control items when the actions stored in the action strategy storage unit 5 are executed and the differences due to individual differences. Specifically, the same "walking"
But normal, hasty, and quiet are different. The difference
For example, as shown in FIG. 9, it is stored as a ratio to the normal parameter.

That is, FIG. 9 stores the part of the action that produces the individual difference, the default value in the case of normal, and the method of applying the ratio when reflecting.

For example, in the case of "gaze", the item reflecting the individual difference is the gaze time, the default gaze time is 1 second, and the ratio of the individual differences is "1 / n". Therefore, it has been shown to be applied at 1 / n times the default gaze time. In other words, in the normal case, the gaze time is 1 second, and in the case of hasty, the gaze time is (1 second × 1/2 = 1/2 second), so the gaze time is shortened. It will be a movement of feeling. On the other hand, in the case of quiet, the gaze time becomes (1 second × 1 / (1/2) = 2 seconds), so the gaze time becomes longer and the motion becomes slower.

The same applies hereinafter, and the details of each will be described in the following operation generation. The action generation unit 8 generates an action according to the flowcharts shown in FIGS. 10 to 13, and the result is presented to the presentation unit 9 as shown in FIGS. 22 to 25, for example. 10 to FIG.
Although a flow chart for generating an action for one person is shown, for a plurality of persons, steps 1 to 37 are processed in parallel for the number of persons. FIG. 22 shows a desk, a pot,
Mr. A, Mr. B, Mr. C in the environment where notes and telephones are placed
It shows the situation with 3 people. FIG. 23 shows a state in which the environment of FIG. 22 is viewed from above, and the positional relationship in the horizontal direction of each object can be confirmed.

The operations of the action generator 6 and the action generator 8 of this embodiment will be described below with reference to FIGS. First, in step 1, the individual difference data stored in the individual difference data storage unit 4 in the format as shown in FIG. 4 is read.
According to the individual difference data read, as explained earlier,
Since the individual difference is determined to be any of hasty, normal, and quiet, the action cycle tc is set by this (step 2). By this action cycle, information is periodically acquired, and as a result, the known external environment changes and the action changes. Taking the case of generating an action for Mr. A as an example, since the characteristic is normal, the action cycle tc has a default value 1/1, for example.
0 seconds is set. That is, in the case of Mr. A, 1 per second
The information will be acquired 0 times and the behavior will be reconsidered. In the case of Mr. B, the ratio is 2 as shown in FIG.
/ 20 seconds is set. Similarly, in the case of Mr. C, q
Since it is a uiet, the ratio is 1/2 as shown in FIG. 5, so a default value of 1/2, 1/5 second, is set.

Next, in step 3, the behavioral patterns stored as shown in FIG. 9, for example, are read from the pattern storage section 7. Further, in step 4, the action strategy stored in the action strategy storage unit 5 as shown in FIG. 8 is read.

First, before entering the action generation, an action flag i indicating which layer of the target action is present is initially set to 0 (step 5). When the action flag i is 0, it means that the player is at the target level 0. When the action flag i is not 0, it indicates that the player is in a sub-action to achieve the target action. For example, if i is 2, it indicates that two more sub-actions are necessary to achieve the target action.

Next, it waits for the setting of some desired behavior (step 6). If the target behavior is set, it is confirmed whether the target is necessary (step 7). If necessary, it is checked whether the target has been set (step 8). If it is not set, a message is issued to prompt the target setting (step 9).

If the target behavior is set correctly,
Set to i (step 10). Subsequent steps are G
It depends on i and the loaded action strategy. Step 4
The action strategy read in step is developed from step 13 to step 37. First of all, branch off for each action,
Next, it is checked whether the condition is satisfied, and if it is satisfied, the actual action is generated. If it is not satisfied, an action having the condition as a result is generated as a sub-action.

Specifically, it is as follows. In the case of action (finding), the condition is “state (OBJ) =
unknown ”. Therefore, it is checked whether this condition is satisfied (step 13). In this case, A
Investigate the target state on Mr. A's external map A. For example,
Since it is stored in the format as shown in FIG. 6, the state is known if the target OBJ is (desk) and unknown if it is (pot).

If this condition is satisfied, a (find) operation is generated (step 14). FIG. 14 shows a flow chart of the generation of the (find) operation in step 14. First, from FIG. 9, the individual difference appears in the pitch. Therefore, the pitch of the (finding) operation according to the individual difference is set. In case of normal, the default is 5 times, and in case of hasty,
In the case of twice, that is 10 times per second, and in the case of quiet, the (finding) operation is repeated at a pitch of 1/2 that time and 2.5 times per second. In the flow chart of FIG. 14, the determinations of step 144 and step 151 are repeated at this pitch. In other words, if it cannot be found if the judgment in step 144 or step 151 is not made 10 times, quiet can only repeat 2.5 times per second.

[0040]

[Equation 1] And it takes 4 seconds. On the other hand, in the case of normal, it can be repeated 5 times per second,

[0041]

[Equation 2] Then, it takes 2 seconds. And for hasty,

[0042]

[Equation 3] And you can find it in just 1 second.

In step 142, the flag B is set to 0 in order to remember the position of the body. Next, the neck position is initialized. At the initial position, an interference check is first performed to see if the field weight and the OBJ interfere with each other. FIG. 15 is a diagram in which the relationship between the field weight and the object is projected on a two-dimensional plane for easy understanding. Since O1 interferes with the field weight having the view angle α,
Within the field of view. However, O2 is outside the field of view because it does not interfere with the field weight.

If it is within the field of view, it is successful (step 145), so the process proceeds to step 16 and the state is changed.
If it is not within the field of view, the neck is rotated by F degrees (step 146). Check if the rotation angle is within the normal range of movement so that it is not twisted strangely (step 14).
7). If it is within the normal range, check the interference again to see if the target object is in the field of view.
Repeat until the normal range is exceeded.

If it exceeds the normal range, the neck is returned to its original position (step 148) and rotated in a different direction (step 149). The same check is performed for the opposite direction (step 150, step 151).

If the target is not found even when looking in the opposite direction, the body is rotated 180 degrees (step 15).
4) Repeat in the same way. When the person has already looked around, the flag B1 of the body position is set (step 155), which means that it has failed (step 153). In the case of failure, the process proceeds to step 15 where an error message indicating that the item was not found is output.

Due to the (find) operation, the target state is un
It is changed from known to known (step 16). That is, for example, in the example of FIG. 6, when the target OBJ is a pot, the state (pot) is changed by the (find) operation.
Since it was changed from unknown to known, it looks like Figure 7.

If the condition is not satisfied, in this case, "state (OBJ) = known", but kn
Being own means outputting the error message because the (finding) operation itself is not necessary in the first place (step 15). It also outputs an error message if it cannot be found.

In the case of action (walking), since the condition is null, a simple (walking) action is generated (step 17). A flow chart of step 17 is shown in FIG.

In FIG. 9, the individual difference appears in the pitch, and the default is 3 times per second, so the pitch is first set (step 171). hasty is 6 times a second, quiet is a second 1.
It will be 5 pitches. In the example of FIG. 9, the stride (step length) does not show individual differences, so the stride simply sets the default (for example, 30 cm) (step 172). Next, the direction is set by obtaining the position of the target OBJ from the environment model storage unit 1 (step 173). Then, the actual walking motion is generated (step 174). Here, for example, it is assumed that the walking motion is generated by the rotational motion of each joint of the limb. However, the interval at which this rotational movement is repeated is controlled by the pitch set in step 171.

In the case of action (touch), there are two conditions. First, the first condition is “state (OBJ) = known”.
Is. Therefore, it is first checked whether this condition is satisfied (step 18). If the condition is not satisfied, an action that satisfies the condition, that is, an action having this condition as a result is generated as a sub-action.

In this case, "state (OBJ) = knwon"
As the result is a (finding) action, it is generated as a sub-action. That is, the action flag is incremented (step 21) and Gi is set to (find). Since the target does not change, keep it as it is. With regard to the sub-action, steps 13 and thereafter will be similarly performed.

If the first condition is satisfied, the second condition "state (OBJ) = reachable" is examined.
Whether it is reachable or unreachable is not remembered in the external map. Since this changes automatically when the position of the operating human model changes, it is not stored in the format shown in FIG. Instead, the distance between the human model and the target is directly calculated, and it is determined whether it is reachable. Specifically, the position of the target can be known because it is stored in the environment model storage unit 1 in the format as shown in FIG. If the target position is (xi, yi, zi) and the current value of the human model is (Xi, Yi, Zi), the distance d between the two is

[0054]

[Equation 4] And is required.

This distance is the range within which the human model can reach β
On the other hand, if d <= β, then it is reachable. In step 19, this judgment is performed. If it is reachable, all the conditions necessary for the (touch) motion are satisfied, so that the (touch) motion is generated in step 20. FIG. 17 shows a flow chart of the generation of the (touch) operation in step 20.

From FIG. 9, (touch), individual differences appear in speed, the default speed is 30 cm per second, and the individual difference is applied n times. Therefore, first, in step 201, the speed is set. normal is 30 cm / sec, hasty is 60c
m and quiet are 15 cm.

Next, the coordinates of the target OBJ to be touched are obtained from the environment model storage unit 1 (step 202). In step 204, inverse kinematics (Badler, NI,
Barsky, BA, and Zeltzer, D., "Making the Move,"
MorganKaufmann Pub., Inc., USA, 1991)
In order to calculate the moving position of the hand, an interpolation position corresponding to the speed is calculated in step 203. That is, as shown in FIG. 18, when the hand position is P0 and the target position is O1,
normal takes P2 between P0 and O1, and P2 and O1
While the inverse kinematics are removed for two points of, while the operation is slower for quiet, a finer interpolation position is required. Therefore, in addition to P2, P
Take Inverse Kinematics by taking P1 and P3 as well.

As a result, the movement of the hand is calculated (step 205). In step 19, if d> α,
In order to be unreachable and satisfy the condition,
An action having (going) with a state (OBJ) = reachable is generated as a sub action. Therefore, first, the action flag is incremented (step 23) and Gi is set to (go). The target does not change, so keep it as it is. With regard to the sub-action, steps 13 and thereafter will be similarly performed.

In the case of action (gaze), there is one condition, "state (OBJ) = known". It is checked whether this article is satisfied (step 25). If it is satisfied, the coordinate value of the target is known from FIG. 2 of the environment model storage unit 1, and the action is generated so that the face of the human model faces in that direction.

FIG. 19 shows a flow chart of the generation of the (gaze) operation in step 26. From FIG. 9, the individual difference appears in the gaze time. Since the default gazing time is 1 second, 1 second is set to normal, 1/2 of 0.5 seconds is set to hasty, and 2 seconds is set to 2 times that of quiet (step 261). Next, the coordinates of the target OBJ to be watched are set from the environment model storage unit 1 in FIG. 2 (step 262). The neck is rotated in the set direction and the gazing is performed for the time set in step 261 (step 263).

If the condition is not satisfied, an action (finding) having this condition as a result is generated as a sub-action. That is, the action flag is incremented (step 27) and Gi (find) is set. The target does not change, so keep it as it is. Regarding sub actions,
Step 13 and the subsequent steps are similarly performed.

In the case of action (heading), there are two conditions. However, if the action flag is not 0 (step 2
9) Since these two conditions have already been satisfied because they are generated as sub-actions of other actions, it is not necessary to check them and the action (directing) is generated immediately (step 33).

FIG. 20 is a more detailed flow chart of step 33. It is almost the same as the (walking) operation of FIG.
Only step 334 is different. That is, FIG.
As can be seen from (ahead), individual differences appear in the reach distance (how close to the target) in addition to the pitch. In other words, while normal approaches only 50 cm away from the target, hasty approaches half that distance, 25 cm. In other words, hasty walks too fast and feels too close to the target, which makes the motion somewhat dull. On the other hand, quiet is the norm
You can only get up to 100 cm, which is twice the distance of al. This will not reach, so it is necessary to approach the remaining 50 cm again. Therefore, in the case of quiet, it approaches the target a little further away, and then again,
I feel like approaching carefully.

When the action flag is 0, the condition is checked (steps 30 and 31). “Status (OBJ) = kn
When "own" is not satisfied, the action (finding) is generated as a sub action as in the other cases (step 3).
4, 35).

If both of the two conditions are satisfied, a (heading) motion is generated (step 3).
3). If the "state (OBJ) = unreachable" is not satisfied, an action (going) is not necessary in the first place, and an error message is issued (step 32).

When the action is generated, step 36 checks whether the generated action is a sub-action. If it is a target action, the process ends, but if it is a sub-action, an action one level higher than that needs to be generated, so the action flag is decremented (step 37) and the process returns to (A).

Also, during the action generation, if the action cycle comes, the process returns to step 12 of information acquisition to acquire information.
This is interrupt B. As a result, the content of the external world memory map changes. As a result, even in the same environment, different actions can be generated due to individual differences in the information acquisition cycle.

Finally, the manner in which the action generated by the action generation unit 6 is presented by the presentation unit 9 will be described with reference to FIG. From the state of FIGS. 22 and 23, two people, Mr. B, who is hasty 12 and Mr. C, who is quiet, 13 are telephone 14
It is an example of a presentation when an action (toward) is made. (In this presentation screen, it is assumed that Mr. A 11 is already in front of the telephone. In addition, desk 15, pot 16, notebook 17
There is. FIG. 24 shows a state in which two persons, Mr. B 12 and Mr. C 13, are facing the telephone from the state shown in FIGS. 22 and 23.

FIG. 25 shows the situation when Mr. B 12 arrives at the telephone. 22 and 23, first C
Although Mr. B 12 was farther from the telephone 14 than Mr. 13, hasty B and quiet C
With Mr. B, there is a difference in the pitch when going, and Mr. B goes faster.

Therefore, in FIG. 25, as a result of two persons, Mr. B and Mr. C, heading for the telephone, Mr. B, who is hasty, has a qu
It is shown that he arrived at the telephone before Mr. iet Na. In this way, the action with individual differences generated by the action generation unit 6 can be easily confirmed by the presentation unit 9. (Modification) In the first embodiment, the individual difference is given as a ratio to the default value as shown in FIG. 4, but it is not necessarily limited to this. It is also possible to give it as a value of the action pattern part of FIG. 9 for each attribute of the individual difference.

Further, it is possible to give a controllable range of change with respect to the default. In other words, the default pitch is given as 5 ± k times per second (0 <k <1), and the individual difference is (5 + kn) in the case of hasty for this k.
In the case of times and quiet, it is also possible to give an individual difference so that the upper limit (or lower limit) of the default change is further greatly changed, as in (5-kn) times.

Further, the individual difference does not necessarily have to be stored in the individual difference data storage section, and the user directly inputs the parameter of the individual difference from the input section 10 under the configuration as shown in FIG. A method (for example, a value corresponding to the ratio in FIG. 5) is also possible. With such a configuration, it is possible to realize individual differences that change depending on the situation.

[0073]

As described in the above embodiments, by giving individual differences in a simple format, it is possible to easily realize operations having various different modes. This allows you to easily set various situations rather than the conventional uniform animation,
It is possible to verify while visually confirming behavior patterns of an unspecified number of users.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a work environment evaluation apparatus according to a first embodiment of the present invention.

FIG. 2 is a diagram showing an example of presentation by an environment presentation unit stored in an environment model storage unit of the embodiment.

FIG. 3 is a diagram showing an example of a storage format of a three-dimensional object stored in an environment model storage unit of the same embodiment.

FIG. 4 is a diagram showing an example of a storage format of individual differences stored in an individual difference data storage section of the embodiment.

FIG. 5 is a diagram showing an example of a storage format of individual differences stored in an individual difference data storage unit of the same embodiment.

FIG. 6 is a diagram showing an example of a storage format of an object existing in an environment stored in an external memory model storage unit of the same embodiment.

FIG. 7 is a diagram showing an example of a storage format of an object existing in an environment stored in an external memory model storage unit of the same embodiment.

FIG. 8 is a diagram showing an example of a storage format of a goal achievement strategy stored in a behavior strategy storage unit of the embodiment.

FIG. 9 is a diagram showing an example of a storage format of an action style stored in a style storage unit of the embodiment.

FIG. 10 is a diagram showing an example of a flow of processing of action generation of the action generation unit of the embodiment.

FIG. 11 is a diagram showing an example of the flow of processing for behavior generation of the behavior generation unit of the embodiment.

FIG. 12 is a diagram showing an example of the flow of a process of action generation of the action generation unit of the embodiment.

FIG. 13 is a diagram showing an example of the flow of a process of action generation of the action generation unit of the embodiment.

FIG. 14 (Find) in the action generator of the same embodiment
It is a figure showing an example of the flow of processing of generation of operation.

FIG. 15 (Find) in the action generator of the embodiment
It is explanatory drawing of the determination in operation.

FIG. 16 is a diagram showing an example of the flow of processing for generating a (walking) motion in the behavior generation unit of the embodiment.

FIG. 17 is a diagram showing an example of the flow of processing of generating (touching) motions in the behavior generating unit of the embodiment.

FIG. 18 is an explanatory diagram of interpolation points in a (touching) operation in the action generation unit according to the embodiment.

FIG. 19 is a diagram showing an example of the flow of a process of generating (gazing) an action in the action generator in the example.

FIG. 20 is a diagram showing an example of the flow of processing of (moving) action generation in the action generation unit according to the embodiment.

FIG. 21 is a diagram showing a modification of the first embodiment.

FIG. 22 is a diagram showing an example of presentation of actions having individual differences in the presentation unit of the first example.

FIG. 23 is a diagram showing an example of presentation of actions having individual differences in the presentation unit of the first example.

FIG. 24 is a diagram showing an example of presentation of actions having individual differences in the presentation unit of the first example.

FIG. 25 is a diagram showing an example of presentation of actions having individual differences in the presentation unit of the first example.

[Explanation of symbols]

Environment model memory 1 External memory model storage unit 2 Information acquisition control unit 3 Individual difference data storage unit 4 Action strategy memory 5 Action generator 6 Style memory section 7 Action generator 8 Presentation section 9 Input section 10

Continuation of the front page (56) References JP-A-6-274094 (JP, A) JP-A-4-195476 (JP, A) Muneto Unuma, 2 people, "Modeling crowd behavior-People in virtual city space Interaction between flock and environment ", IEEJ Transactions C (Electronics, Information and Systems Division), The Institute of Electrical Engineers of Japan, January 20, 1995, Vol. 115-C, No. 2, p. 212 −221 Kenji Mochizuki, 1 person, “Proposal of a 3D object model world system that integrates language and image media”, IEICE technical research report (CS91-76 to 82 communication method), corporation The Institute of Electronics, Information and Communication Engineers, September 26, 1991, Vol. 91, No. 240, p. 1-6 Naoko Umeki Outside 3 people, "Basic research for virtual space navigation", The Institute of Electronics, Information and Communication Engineers Technical Research Report (MVE95-8-24 multimedia / virtual environment basis), The Institute of Electronics, Information and Communication Engineers, May 26, 1995, Vol. 95, No. 83, p. 61-66 (58) Fields investigated (Int.Cl. ⁷ , DB name) G06T 15/70 CSDB (Japan Patent Office)

Claims (10)

(57) [Claims] 1. An information acquisition control of an external memory model, which is an arrangement of objects to be recognized by the human model in the visual field, from the arrangement of objects in a virtual space based on individual difference data which is a characteristic of each human model. From the action strategy, which is a rule for the human model to perform the action that the human model should act, based on the determined external memory model, the action generation unit determines the action that the human model can act <br / A motion generation method characterized in that the motion generation means generates the motion of the determined behavior of the human model according to a behavior pattern that is a rule for generating a motion that the human model should act. 2. The individual difference data is composed of a plurality of attribute values.
The operation generation method according to claim 1, wherein: 3. The individual difference data is obtained from a ratio of a plurality of attributes.
The operation generation method according to claim 1, wherein 4. The external memory model is the human model.
It is characterized by storing the state of the object according to each
The operation generation method according to claim 1. 5. The behavior pattern is a pattern for each human model.
A ratio to a parameter, which is a ratio.
Described behavior generation method. 6. An environment model storage means for storing an arrangement of objects in a virtual space, an individual difference storage means for storing individual difference data which is a characteristic of each human model, and the individual difference storage means. Based on the stored individual difference data, the human model has a visual field based on the arrangement of objects in the virtual space stored in the environment model storage means.
Information acquisition control means for determining an external world memory model, which is an arrangement of objects recognized in the inside, external world memory model storage means for storing the external world memory model determined by the information acquisition control means, and the human model Is stored in the external memory model storage means from the behavior strategy storage means that stores a behavior strategy that is a rule for performing an action to be performed, and the behavior strategy stored in the behavior strategy storage means. An action generation unit that determines an action that the human model can move based on an external memory model; and a mode storage unit that stores an action pattern that is a rule for generating an action to be taken by the human model. , based on the behavior stored in this manner storing means, have the operation generating means for generating an action operation to be of the human model the behavior generation unit decides Motion generating system according to claim Rukoto. 7. The individual difference data is composed of a plurality of attribute values.
7. The motion generation device according to claim 6, wherein: 8. The individual difference data is obtained from a ratio of a plurality of attributes.
7. The motion generation device according to claim 6, wherein: 9. The external world memory model storage means is the person.
The feature is that the state of the object corresponding to each inter-model is stored.
The motion generating apparatus according to claim 6, which is a characteristic. 10. The style storage means is the human model.
Characterized by storing the ratio to each parameter
The motion generation apparatus according to claim 6.