JPH10307864A - Group escape action prediction method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the result having high prediction accuracy by adding the quantized model of a mental state to the action model of an individual and performing simulation. SOLUTION: By individually modeling human beings in a facility space 201 as individual objects 202 and calculating the plural pieces of actions, group escape action is calculated. Movement can not be performed through a part where the individual objects 202 provided with size are crowded. When a disaster such as a fire is generated, the range 203 is dispersed. In order to evade the disaster, the individual objects 202 move toward an exit 204. To such an individual action management function by an individual object group, an individual mentality management function for changing mental parameters for which the mental states of the individual human beings are quantized based on a peripheral environment is additionally connected. Then, by selecting an action pattern based on a mental parameter value, the complicated group escape action is simulated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for directly estimating human evacuation behavior in a spatial spread such as an underground passage or a building passage.

[0002]

2. Description of the Related Art There has been proposed a method for estimating human evacuation behavior at the time of disaster using a computer. For example, "Evacuation safety analysis of large-scale underground space using evacuation behavior simulation: 9th Japan Earthquake Engineering Symposium (199
4) As disclosed in “pp2353-2357”, a space is divided into uniform meshes, data representing a person is arranged in each mesh, and a disaster factor or an obstacle is defined as a danger potential, and the magnitude of the potential is defined. Has proposed a method for controlling the movement of human data.

[0003] However, in the above-mentioned method, there is no proposal for performing prediction in consideration of human psychology. In particular, the peculiarity of psychological state at the time of panic has been pointed out,
An evacuation behavior model that considers these situations is needed.

[0004] On the other hand, from a psychological point of view, attempts have been made to analyze behavior at the time of danger by a model experiment in which a human is used as a subject. As disclosed in "Panic Experiments-Social Psychology in Crisis-Naoki Kugihara, Nakanishiya Publishing", a method of simulating behavior by the frequency of pressing buttons has also been proposed. The method of actually predicting a human as a subject has not been able to evaluate behavior involving movement. The need to maintain a sense of urgency has also been pointed out.

[0005]

In order to solve the above-mentioned problems, it is necessary to predict the behavior by using a personal model capable of expressing a human psychology, especially a panic state, by a computer. The present invention proposes a model capable of selecting / controlling human behavior by adding a quantification model of a psychological state to an individual behavior model.

[0006]

According to the present invention, a human is individually represented as a personal object and personal object parameters are managed. This personal object has a behavior simulation function and a psychological simulation function for determining the behavior pattern. The above problem is solved by a psychological simulation function. The psychological simulation function determines the human behavior at the time of crisis that changes depending on the psychological state.
The group evacuation behavior is predicted by moving a plurality of personal objects having such a psychological simulation function. At this time, the personal object does not know the goal of the action (escape from the exit), and moves while searching for the goal while grasping the surrounding environment.

In order to realize such a function by a computer, a personal behavior management function based on a personal object group is provided with a personal psychological management function of changing a psychological parameter quantifying an individual's psychological state based on the surrounding environment. Make additional connections. Then, a complex group evacuation behavior is simulated by selecting a behavior pattern based on a psychological parameter value. As parameters describing the psychological state of a human, a belief parameter indicating the degree of certainty with respect to one's own behavior and a panic parameter indicating the certainty of executing an interruption action, which is a selfish action due to anxiety, are used.

[0008] The personal object simulates a complex human action by selecting an action based on a belief parameter and a panic parameter quantifying a human psychological state. Here, the ratio of the belief parameter and the panic parameter to the maximum value is obtained, and the behavior pattern is selected by comparing the ratio with a randomly generated value. The belief parameter value increases when it matches the behavior of another personal object, and decreases when it does not match. When the belief parameter value is high, even if the behavior of the personal object does not match the behavior of the personal object in the visual field, the user's behavior is not changed. However, when the value of the belief parameter is low, an attempt is made to match the behavior of the individual object around. Panic parameter values increase when a disaster comes into view,
It is low when outside the field of view. When the panic parameter value is high, the self-priority action by interruption is executed. When the panic parameter value is low, the deceleration action or the interference avoidance action by changing the moving direction is executed when the personal object is crowded at a specific place.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION In the present invention, a method of estimating a group evacuation behavior by taking a human evacuation behavior at the time of a fire as an example will be described. FIG. 2 shows a facility space 20 such as a building passage or an underground passage.
1 shows a simulation example. Here, the collective evacuation behavior is calculated by individually modeling a human as the personal object 202 and calculating a plurality of movements. This personal object has a size and cannot move through a dense area of personal objects. When a disaster such as a fire occurs, the area 203 spreads. The personal object 202 moves so as to avoid this disaster, and moves toward the exit 204.

[0010] The personal object has parameters relating to the movement and the visual field depending on its position and moving speed. And the psychological state is set for each individual object,
The psychological parameter indicating the psychological state is updated autonomously according to the state of the surrounding environment. As the psychological parameters, a belief parameter indicating the degree of certainty regarding the user's own behavior and a panic parameter indicating the certainty of executing the interrupting action, which is a selfish action due to anxiety, are used. If the value of the belief parameter for one's action is improved, even if another personal object is performing another action, he or she will stick to his own action. When the panic parameter value increases, the user completes his or her action regardless of the actions of other personal objects. Belief parameter values increase when avoiding an impending disaster and when the behavior of the surrounding personal objects matches their behavior. Further, the panic parameter value is improved when a group of personal objects around is crowded.

The personal object recognizes the situation of the surrounding personal object and the position of the spatial data. For this purpose, a visual field parameter is used. The personal object can grasp only the personal object in the visual field, the disaster situation, and the spatial data. This means that the situation of the whole space cannot be grasped. This models a situation in which humans can only acquire the surrounding information visually.

An action pattern is selected according to a change in a psychological parameter. This choice is probabilistic. There are two types of behavior patterns: simple movement-free interference behavior and mutual interference behavior such as collision avoidance with other personal objects. The prediction of the group evacuation behavior is performed by selecting the behavior pattern according to each situation for all the individual objects.

FIG. 1 shows a processing hierarchy for simulating a collective evacuation behavior, and FIG. 3 shows a detailed functional configuration based on the functional hierarchy. The processing layer is composed of three layers: an environment management layer, a personal behavior management layer, and a personal psychology management layer, which perform processing relating to space, individuals, and psychology.

<Environment management layer> The environment management layer includes spatial data such as a diagram of a facility such as an underground passage or a map, a spatial data management 101 for managing a diffusion range of a disaster, and a disaster simulation 202 for predicting a disaster such as a fire spread. The spatial data management 101 includes a simulation execution time management unit 301,
Disaster mapping unit 303, spatial data management memory 30
4. Disaster simulation function 202 corresponds to disaster simulation execution unit 302.

The simulation execution time management unit 301 manages the transition of the simulation time. The disaster simulation unit 302 predicts a disaster such as a fire spread simulation. The disaster mapping unit 303 associates the result of the fire spread with the spatial data. The status display unit 305 displays the spatial data, the current position of the personal object, and the disaster data diffusion range. The spatial data management memory 304 stores spatial data.

When the time is advanced by the simulation execution time management section 301, a notification is sent to the disaster simulation execution section 302 and the individual group management section 307, and the simulation of the disaster and the collective evacuation behavior is advanced, respectively.

<Personal Behavior Management Layer> The personal behavior management layer is composed of a personal group management 104 for managing behavioral information such as the position and moving direction / velocity of a personal object representing an individual, and a personal behavior simulator 103 for calculating the behavior of the personal object. . The individual group management 104 includes an individual group initial arrangement unit 306 and an individual group management unit 307. Personal behavior simulator 103
Is composed of an environment data acquisition unit 308, an environment determination unit 309, a behavior parameter management memory 312, a behavior calculation unit 310, a behavior selection unit 311 and a behavior parameter management memory 312.

The individual group initial arrangement unit 306 arranges individual objects in spatial data. The personal group management unit 307 selects a personal object. Environmental data acquisition unit 308
Acquires the spatial data included in the visual field of the personal object, the position of the personal object group in the vicinity, and the disaster position. The environment determining unit 309 determines the surrounding situation from the acquired spatial data and the position of the personal object group. The action calculation unit 310 calculates a movement position based on the selected action pattern and the movement speed of the personal object. At this time, in the case of collision with spatial data or other personal objects, self-intervention or avoidance by interruption is performed. The action selecting unit 311 selects an action pattern according to the state of the psychological parameter. The action parameter management memory 312 stores parameters relating to the position, velocity, and visual field of the personal object, spatial data obtained from the visual field of each personal object,
Stores disaster data and personal objects.

<Personal Psychological Management Layer> A personal psychological simulator 1 that manages parameters indicating the psychological state of a personal object and updates the psychological parameters according to the situation of the surrounding environment.
05. The personal psychological simulator 105 includes a conviction parameter value calculation unit 313, a panic parameter value calculation unit 314, and a psychological parameter management memory 315.

The belief parameter value calculation unit 313 reinforces the persistence of the personal object on the currently selected action pattern based on the environment data. The panic parameter value calculation unit 314 updates the degree of fear when it is determined that the goal to be realized cannot be achieved by the action being performed at present. Psychological parameter management memory 31
Reference numeral 5 stores a belief parameter value and a panic parameter value of each personal object.

FIG. 4 shows an algorithm for performing a group evacuation behavior prediction using the functions described above.

The space data is displayed by the status display unit 305 (S401 to S415 indicate steps). Next, the individual group initial arrangement function 306 randomly arranges the designated number of personal objects in the spatial data (S402). The individual objects are arranged so as not to hit each other. Furthermore, personal objects are not placed outside the spatial data. When it is arranged outside the spatial data, it is arranged inside the spatial data by recalculating the arrangement position.

When the personal object is placed, the behavior parameters and the psychological parameters are initialized (S403). For the personal object, its identification number (ID) and the behavior parameter are stored in a personal group management table (PDM: Persons Dat).
a Management Table). The identification number of the personal object is uniquely assigned to the personal object. FIG. 5A shows the movement parameters of the personal object, and FIG. 5B shows the structure of the PDM. The number of individuals 510, which is information of the PDM 508, indicates the total number of personal objects that have not escaped from the exit. The personal object 501 has position coordinate data 502 and velocity vector data 5 as action parameters.
03, as the visual field data, data of the length of the visual line 506 and the visual angle 507 indicating the spread of the visual field are stored. The visual field range is different between 504 in a normal state and 505 in a panic, and narrows during a panic. The position coordinate data 502 and the velocity vector data are stored as two-dimensional coordinates (X, Y) (Vx, V). The speed of the personal object is determined randomly at a maximum in the range of 1.5 to 3 m / s. The direction of the velocity vector is assumed to be parallel to the figure that is the spatial data. That is, when moving on a road, the vehicle moves along the direction of the road. The viewing direction is made coincident with the moving direction, the length of the line of sight is 10 m, and the viewing angle is 135 degrees.

The psychological parameter has a belief parameter value of 10
0 and the panic parameter value is 0. A belief parameter value of 100 indicates that one's behavior is not affected by the movement of another personal object. Further, a panic parameter value of 0 does not indicate a selfish action such as an interrupting action even when a group of personal objects is gathered and movement is hindered. When the personal objects are arranged in the space, all the personal objects are displayed by the status display unit 305 (S404).

Next, a fire spread is calculated by the disaster simulation unit 302 and stored in the spatial data management memory 304 by the disaster mapping unit 303 (S405).
Indicated by 05. The action calculation unit 310 calculates the destination coordinates according to the velocity vector of the personal object stored in the action parameter management memory 312 (S407). The behavior calculation function 310 calculates a moving position at the next time based on the position of the personal object and the velocity vector (S413).

X (t + Δt) = X (t) + Vx · t (1) Y (t + Δt) = Y (t) + Vy · t (2) At this time, the personal object that has exited from the exit is It is deleted from the PDM (S408). Further PD
The action parameters of the personal object registered at the end of M are copied to the deleted area. And the number of individuals 5
09 is updated.

In the case where the object comes into contact with another personal object or jumps out of the space data, the collision interference is calculated and the object is arranged at the closest position where the object does not contact. If it intersects with the spatial data, it is located just before it intersects with the wall. When a collision occurs with another personal object, the collision is avoided by lowering the speed when the panic parameter value is low, or the collision is avoided by changing the moving direction. In particular, when a group of personal objects is gathered at the exit, the direction is changed without selecting that exit, and another exit is selected. If the panic parameter value is high, the collision course is selected as it is and the personal object is brought into contact (S409).

Next, each individual object grasps its surroundings (S410). The environment data acquisition unit 308 searches the spatial data management memory 304 and the behavior parameter management memory 312 for the identification number and spatial data of the personal object included in the field of view. The acquired identification number and the range coordinates of the spatial data are the VAI shown in FIG.
(Visible Area Information Table) 509 (S410). The PDM and VAI are stored in the behavior parameter management memory.

Next, the value of the psychological parameter is calculated (S
411). Psychological parameters change depending on the surrounding environment. A belief parameter value and a panic parameter value are calculated as psychological parameters.

The belief parameter value is updated by the belief parameter value calculation unit 313. As shown in FIG. 6A, when the moving direction of the user is the same as the moving direction of another personal object, the belief parameter value is set to 1
to add. In the opposite case, the belief parameter value is decremented by one. The environment determination unit 309 detects the number M of personal objects that do not match the number N of personal objects that match the moving direction of the individual objects around the own object. For the moving direction, the absolute value of the angle difference between the velocity vector of the PDM for the personal object and the surrounding personal object is calculated. If this is smaller than a predetermined threshold value, it is determined that the moving directions match. If it is larger than the threshold value, it is determined that the moving directions do not match. In this manner, the rule if N <M subtracts 1 from the belief parameter value... (3) else The belief parameter value is updated by adding 1 to the belief parameter value.

The panic parameter value is updated by the panic parameter value calculation unit 314. When the movement is interrupted by the presence of another personal object, or when the spread of the disaster is approaching as shown in FIG. 6B, the panic parameter value is added. If it is not blocked, or if a disaster is not approaching and there is room, the panic parameter value is subtracted.

The environment determining unit 309 determines whether the disaster diffusion range has entered the field of view and whether the number of surrounding individual objects is larger than the threshold value (whether the object is congested). Then, the panic parameter value calculation unit performs the following calculation. P is the number of surrounding personal objects,
Assuming that Pth is the threshold, the rule if (disaster diffusion range OR P> Pth) is added to the panic parameter value by 1 ... (4) else The panic parameter value is subtracted by 1 (where OR is "or""Means") to update the panic parameter value.

Next, an action pattern is selected based on the psychological parameters (S412). The selection of the behavior pattern is performed by the behavior selection unit 311.

FIG. 7 shows the behavior pattern of the personal object. Basic movement patterns include simple movement 702 and mutual interference 703. The simple movement 702 is calculated based on the velocity vector. Furthermore, when the turning angle enters the field of view, the direction change is determined stochastically. Mutual interference 703 is
It is defined between a plurality of personal objects and spatial data and the personal object of interest. As the mutual interference, “overtaking 704”, “collision avoidance / matching 705”,
There are “interruption 706” and “deceleration 707”.

"Overtaking" means that when the moving speed of a personal object is faster than that of another personal object, the personal object is overtaken from the side. At this time, since the personal object has a size, the overtaking position is changed so as not to collide with the personal object. “Collision avoidance / matching” means an action of avoiding a collision when personal objects move face to face. As in the case of "overtaking", the personal object also has a size in this case, so that the moving direction is changed so as not to collide with the personal object. In this case, the movement stops immediately before the collision. The “interruption” is a selfish action that moves irrespective of the state of other personal objects in order to give priority to one's own movement. “Deceleration” is an operation of reducing the speed so as not to collide in a crowded area, contrary to “interruption”. "Change of moving direction" is an operation for avoiding mutual interference.

First, a moving speed / direction is determined in selecting an action pattern, and an interaction with another personal object is determined.

A method for determining the moving direction and the speed will be described. When no personal object is included in the field of view, the movement is simple. The calculation method of the moving direction and the speed when the personal object is included in the visual field will be described below.

As shown in FIG. 8A, when the belief parameter value is higher than the threshold value, the moving direction and speed are not changed. On the other hand, when the belief parameter value is low, the moving direction and speed are changed.

When the belief parameter value is low, assuming that the moving speed is V, V = maxVi / max (i) (i is the number of individual objects in view) (5). If the moving directions of the surrounding individual objects do not match, V may be 0.

Regarding the moving direction, if the belief parameter value is B, if B / Bmax ≦ Random Changing the moving direction (6) else the moving direction is not changed (Bmax: the maximum value of the belief parameter value). Represents a randomly generated number up to 1.

According to the equation (6), when the belief parameter value is small, the probability of selecting the movement direction changing action increases, and when the belief parameter value is large, the movement direction of the surrounding individual object is kept as it is. Is more likely to be maintained.

For the speed of the individual object in the visual field, a personal object identification number in the visual field is searched from the information table VAI in the visual field, and a speed vector of the corresponding personal object is searched from the PDM using the identification number as a key. Obtained by:

Next, interference calculation is performed based on the panic parameter value. As shown in FIG. 8B, when the panic parameter value decreases, interference with other personal objects is avoided, and when the panic parameter value increases, interference occurs. Probability is used to select an action pattern according to the panic parameter value. When the panic parameter value is D, if D / Dmax ≦ Random interference execution (7) else Avoidance of interference (Bmax: maximum value of belief parameter value) When the panic parameter value is small, the moving direction is avoided or the moving speed is reduced. When the panic parameter value is large, the probability of selecting a selfish action like interruption is high.

As shown in FIG. 8B, when the panic parameter value is low, whether to decelerate or change the moving direction is selected stochastically.

If Random ≦ 0.5 Decrease of moving speed (8) else Change of moving direction When the panic parameter value is high, the size of the visual field which is an action parameter is changed. Specifically, line of sight 5
06, and the viewing angle 507 is reduced. Specifically, the line-of-sight length is changed in the range of 2 m to 10 m and the viewing angle is changed in the range of 60 to 135 degrees in proportion to the magnitude of the panic parameter value.

The processing from step S407 to step S412 is performed for all personal objects (S413). When the individual group management unit 307 refers to the behavior parameter management memory 312 and determines that all personal objects have escaped from the exit (S414),
The simulation ends (S415). If there are still unexposed individual objects, the processes from step 405 to step 416 are repeatedly executed to continue the simulation (S40).
4).

FIG. 9 shows an embodiment of a hardware configuration for performing a group evacuation behavior simulation according to the present invention.
Here, the calculation units (902, 903) for performing a group evacuation behavior simulation and a disaster simulation to which a large simulation is added are separately described. LA
N901 connects a computer and peripheral devices. Printer 90
Reference numeral 4 outputs graphic data of a prediction result.

[0048]

According to the present invention, an evacuation behavior prediction simulation based on a special psychology at the time of crisis can be performed because the selection of evacuation behavior patterns by a group of humans is performed based on the psychological parameters of each individual. Then, since the simulation is performed by modeling the actual space, individual behavior, and mental state, it is possible to perform the simulation in the actual state, and to obtain a result with high prediction accuracy.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing a processing hierarchy of a simulation of predicting a group evacuation behavior according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an execution example of a group evacuation behavior prediction simulation according to one embodiment of the present invention.

FIG. 3 is a block diagram showing a functional configuration of a group evacuation behavior prediction simulation according to one embodiment of the present invention.

FIG. 4 is a processing flowchart showing a group evacuation behavior prediction simulation algorithm according to one embodiment of the present invention.

FIG. 5 is an explanatory diagram showing a personal object management method according to an embodiment of the present invention.

FIG. 6 is an explanatory diagram showing a relationship between a psychological parameter of an individual object and an action pattern.

FIG. 7 is an explanatory diagram showing a behavior pattern of a personal object.

FIG. 8 is an explanatory diagram showing action selection based on psychological parameters.

FIG. 9 is a block diagram showing a configuration of simulation hardware according to one embodiment of the present invention.

[Explanation of symbols]

301: simulation time management unit, 302: disaster simulation execution unit, 303: disaster mapping unit, 3
04: Spatial data management memory, 305: Status display unit, 3
06 individual group initial arrangement unit, 307 individual group management unit, 30
8 ... environment data acquisition unit, 309 ... environment determination unit, 310 ...
Behavior calculation unit, 311: Behavior selection unit, 312: Behavior parameter management memory, 313: Belief parameter value calculation unit, 31
4. Panic parameter value calculation unit 315: Psychological parameter management memory

Claims (4)

[Claims] 1. A simulator for estimating collective evacuation behavior by simultaneously activating a plurality of personal objects modeling individual humans in spatial data. An environment management function for predicting a disaster diffusion range and managing spatial data. And the personal behavior management function that calculates the psychological state in addition to the personal behavior management function that calculates the behavior of the personal object, and is calculated from the influence of the behavior of the surrounding personal objects in the personal psychology management function A group evacuation behavior prediction method characterized in that a complex group behavior is simulated by selecting a behavior pattern based on a psychological state of an individual object. 2. A personal psychological management function of a simulator for predicting a group evacuation behavior according to claim 1, wherein a psychological parameter indicating a psychological state is determined by a belief parameter indicating a certainty factor of the adopted behavior and a belief parameter indicating an uneasiness. It is represented by a panic parameter that indicates the certainty of executing the interrupting action that is a selfish action, and the ratio of each parameter value to the maximum value is determined, and an action pattern is selected by comparing the value with a randomly generated value. Group evacuation behavior prediction method. 3. An environment management function comprising a simulation time management unit for managing elapsed time, a disaster simulation unit for predicting a disaster, and a disaster mapping unit for associating the status of disaster data with a space, and a personal object in space data. A personal group initial placement unit, a personal group management unit that manages identification numbers, positions, movements, visual fields, and personal object / space data within the visual field; and an environment data acquisition unit that acquires surrounding spatial data. An environment determination unit that determines a situation from the obtained environment data, an action calculation unit that calculates the movement of a personal object and interference of arrangement with another personal object, and an action selection unit that selects an action pattern from psychological parameters. A belief that calculates a belief parameter indicating the degree of certainty of the behavior in the constructed evacuation behavior prediction simulator By adding a personal psychological management function consisting of a parameter value calculation unit and a panic parameter value calculation unit that calculates a panic parameter that determines selfish behavior, it is arranged in the spatial data by the individual group arrangement unit and managed by the individual management function Created personal object
Disaster diffusion data obtained from the disaster simulation unit placed in the space by the disaster mapping unit and the arrangement of surrounding individual objects are acquired by the environment data acquisition unit,
The belief parameter and the panic parameter are updated by the belief parameter value calculation unit and the panic parameter value parameter value calculation unit based on the congestion degree of the personal object group obtained by the environment determination unit and the distance between the personal data and the spatial data or the disaster data, respectively, Collective evacuation characterized by repeatedly selecting an action pattern and calculating a moving speed by an action selecting unit for all personal objects, and predicting a group evacuation action by displaying an obtained result by a situation display unit. Behavior prediction method. 4. A simulator for predicting a group evacuation behavior according to any one of claims 1 to 3, wherein an exit position of space data serving as an action destination target of an individual object is not known in advance and moves in the space. A group evacuation behavior prediction method characterized in that the exit is found by visual field calculation of an environmental data acquisition unit, and whether to aim for the exit is determined based on psychological parameters.