JP6679645B2 - Simulator, simulation method and action determination device - Google Patents

Description

  The present invention relates to a simulator that predicts a behavior of a crowd, a simulation method, and a behavior determination device that determines the behavior of an individual person who is one of the individuals that make up the crowd.

In order to properly guide the evacuation of the crowd in the event of a disaster or to design a space where evacuation can be performed smoothly, it is necessary to investigate the mechanism that causes panic behavior. Therefore, there is a method of using a simulator that simulates the behavior of a crowd.
One of the explanations of the irrational behavior of the crowd, such as the panic behavior of the crowd at the time of a disaster, is entrainment behavior by cognitive bias, and studies such as approaching it as a leader-follower problem have been conducted. It was The approach as a leader-follower problem is that, in a crowd consisting of a leader who decides his own behavior and a follower who decides his own behavior by the behavior of another person, as a result of the influence of individuals, irrational behavior as a crowd This is a modeling-based approach of taking.

  For example, in the simulation device described in Patent Document 1, the relationship between a plurality of objects is represented by a hypergraph, and the behavior of the plurality of objects is simulated by using the hypergraph, so that the leader follows a specific follower. The relationship was always expressed and the relationship that the follower always followed with a particular leader.

JP, 2003-346071, A

  However, in the conventional technique, it is difficult to simulate a specific behavior under unknown conditions because it is necessary to arbitrarily set conditions for simulating a specific behavior such as panic behavior in a crowd. However, it was difficult to determine the conditions under which the specific behavior occurred.

For example, in the simulation device described in Patent Document 1, the hypergraph can be generated each time the simulation is performed, but the relationship represented by the hypergraph is maintained during the simulation.
So, for example, in order to simulate panic behavior in which evacuees concentrate on one of the two exits, create a hypergraph in which the majority of individuals in the crowd are followers of a very small number of leaders. , It is considered necessary to make arbitrary settings to induce panic behavior.

The present invention has been made in view of the above problems, a simulator capable of simulating a specific behavior such as panic behavior in a crowd under unknown conditions, and a simulator capable of facilitating the investigation of the condition in which the specific behavior occurs in the crowd. It is an object to provide a method and a behavior determining device.
In this specification, "crowd" means a group of people, and "crowd" means a group of social organisms including a person.

  In order to achieve such an object, the present invention is a simulator for predicting the behavior of a crowd in a predetermined space capable of achieving a predetermined goal, and at least for each of a plurality of individuals constituting the crowd, An attribute value indicating whether it is a leader or a follower, a state value including a position in the space, and an individual information storage unit that stores an action parameter that acts on the state value, and a stimulus that indicates a degree of a stimulus that triggers the action. A stimulus value setting unit that sequentially sets a value, and each time the stimulus value is set, a first winning probability based on the stimulus value is set to an individual whose attribute value is a follower among the individuals, A probability setting unit that sets a predetermined second winning probability for an individual whose attribute value is a leader, and the attribute value of the individual for which the first winning probability is set is the leader at that probability. A leader action determination unit that sets a value that brings the state value of the individual closer to the achievement of the purpose in the action parameter of the individual whose attribute value is a leader, and the position from the individual for each individual. A surrounding individual information acquisition unit that extracts surrounding individuals within a predetermined distance, and the attribute value of the individual for which the second winning probability is set is changed to a follower with the probability, and the attribute value is a follower. A follower action determination unit that sets a value similar to the action parameter of the individual surrounding the individual in the action parameter of the individual, and individual information update that updates the state value of each individual with the action parameter of the individual And a simulator provided with the section.

  Further, the individual information storage unit further stores a reaction threshold value of each individual, and the probability setting unit, the higher the stimulus value is higher for each individual, the higher the lower the reaction threshold, the higher the first winning. It is preferable to set the probability.

  Further, it is preferable that the stimulus value setting unit sets the stimulus value for each individual.

  Further, further comprising a risk value acquisition unit that sequentially acquires a risk value representing the magnitude of the risk occurring in the space, the individual information storage unit, a cognitive threshold value representing the magnitude of the risk recognized by each of the individuals Further, the stimulation value setting unit stores the stimulation value set for each individual in the individual information storage unit, and a large difference between the risk value for each individual and the cognitive threshold of the individual is large. It is preferable to set the stimulus value by increasing the value stored in the individual information storage unit more for an individual.

  Further, the stimulation value setting unit stores the stimulation value set for each individual in the individual information storage unit, and the individual information storage unit stores the individual information as the number of the surrounding individuals extracted by the surrounding individual information acquisition unit increases. It is preferable to set the stimulus value by greatly increasing the value stored in the unit.

  Further, in order to achieve such an object, the present invention is a simulation method for predicting the behavior of a crowd in a predetermined space capable of achieving a predetermined purpose, the computer comprising a plurality of individuals constituting the crowd. For each of the above, at least an attribute value indicating whether it is a leader or a follower, a state value including a position in the space, and an action parameter that acts on the state value are stored, and the degree of stimulation that triggers the action is represented. A process of sequentially setting a stimulus value, and each time the stimulus value is set, a first winning probability based on the stimulus value is set to an individual whose attribute value is a follower among the individuals, and the attribute value is set. The process of setting a predetermined second winning probability to the individual whose is the leader, and the attribute value of the individual for which the first winning probability is set to the leader with the probability. In addition, a process of setting a value that brings the state value of the individual closer to the achievement of the purpose in the action parameter of the individual whose attribute value is a leader, and the position of the individual from the individual is a predetermined distance. With the process of extracting surrounding individuals that are within, and changing the attribute value of the individual for which the second winning probability is set to a follower with the probability, the attribute value is related to the action parameter of the individual who is a follower. To execute a follower action determination unit that sets a value similar to the action parameter of the surrounding individual of the individual, and an individual information update unit that updates the state value of each individual with the action parameter of the individual. There is provided a simulation method characterized by:

  Further, in order to achieve such an object, the present invention is a behavior determination device for determining the behavior of one's own individual, which is one of the individuals constituting a crowd in a predetermined space capable of achieving a predetermined objective, An individual information storage unit that stores at least an attribute value indicating whether the subject is a leader or a follower, a state value that includes a position in the space, and an action parameter that acts on the state value for the own individual; A stimulus value setting unit that sets a stimulus value that represents the degree of stimulus that triggers action, and sets a first winning probability based on the stimulus value when the attribute value is a follower, and the attribute value is a leader. A probability setting unit that sets a predetermined second winning probability in that case, and when the first winning probability is set, the attribute value is changed to a leader with the probability, and the attribute value is changed. In the case of a leader, a leader action determination unit that sets the action parameter to a value that brings the state value closer to the achievement of the purpose, and among the individuals, surrounding individuals that are present within a predetermined distance from the position of the own individual. When the second winning probability is set, the attribute value is changed to a follower with the probability, and when the attribute value is a follower, Provided is an action determination device including a follower action determination unit that sets the action parameter similar to the action.

  According to the present invention, the leader and the follower are stochastically changed even if the same setting is performed. This makes it possible to simulate a specific behavior such as a panic behavior in the crowd under unknown conditions, and facilitate the investigation of the condition in which the specific behavior occurs in the crowd.

It is a block diagram which shows the schematic structure of the simulator 1 which concerns on 1st embodiment. It is a functional block diagram of the simulator 1 which concerns on 1st embodiment. 3 schematically illustrates information stored in the environment information storage unit 41. 3 schematically illustrates the information stored in the individual information storage unit 42. 3 schematically illustrates the information stored in the individual information storage unit 42. It is an example of an intermediate prediction result. It is another example of an intermediate prediction result. It is an example of the prediction result imaged as a graph by statistically processing. 3 is a flowchart showing the operation of the simulator according to the first embodiment. It is a flow chart of action determination processing of a simulator concerning a first embodiment.

[First embodiment]
As a first embodiment of the simulator of the present invention, a fire broke out in a room R having _two exits E ₁ and E _2, and a plurality of people (crowd) staying in the room R were discharged from either exit. An example of the simulator 1 for predicting evacuation behavior will be shown.
That is, in this embodiment, the space to be simulated (target space) is the room R, the individuals that make up the crowd are people, and the purpose of the crowd or individuals is to evacuate from the event of fire. Has the exits E ₁ and E ₂ , which makes it possible to achieve the purpose.
In addition, the exemplified simulator 1 includes an action determination device which is another aspect of the present invention, and the action determination device sequentially and randomly determines the action of each individual to predict the action of the crowd.

<Structure of Simulator 1>
FIG. 1 is a block diagram showing a schematic configuration of the simulator 1 according to the first embodiment.
The simulator 1 includes an operation input unit 2, a file input / output unit 3, a storage unit 4, a control unit 5, and a display unit 6. Of these, the file input / output unit 3, the storage unit 4, and the control unit 5 can be realized by a so-called computer, and the operation input unit 2 and the display unit 6 can be realized as peripheral devices of the computer.

  The operation input unit 2 is a user interface such as a keyboard and a mouse, and is operated by a user and used to input data. The operation input unit 2 is connected to the control unit 5, converts the user's operation into an operation signal, and outputs the operation signal to the control unit 5.

  The file input / output unit 3 is a DVD (Digital Versatile Disc) drive, a USB (Universa Serial Bus) interface, a network interface, etc., one of which is connected to an external device (not shown), a recording medium, a network, etc., and the other is a control unit. 5, the data is input to the control unit 5 as a file, and the data is output from the control unit 5 as a file.

  The storage unit 4 is a memory device such as a ROM (Read Only Memory) and a RAM (Random Access Memory), and stores various programs and various data. The storage unit 4 is connected to the control unit 5 and inputs / outputs these pieces of information from / to the control unit 5.

  The control unit 5 is configured by an arithmetic device such as a CPU (Central Processing Unit), a DSP (Digital Signal Processor), and an MCU (Micro Control Unit). The control unit 5 is connected to the storage unit 4 and operates as various processing units by reading and executing a program from the storage unit 4 to store and read various data in the storage unit 4. The control unit 5 is also connected to the operation input unit 2, the file input / output unit 3, and the display unit 6, and the user operates the operation input unit 2 to acquire the input data, and the file input / output unit 3 is operated. Data is externally acquired and output as a file, a simulation based on the obtained data is performed to predict the behavior of the crowd, and data of the prediction result is output as a file from the file input / output unit 3 and / or prediction The resulting data is imaged and displayed on the display unit 6.

  The display unit 6 is a display device such as a liquid crystal display or a CRT (Cathode Ray Tube) display, is connected to the control unit 5, and displays a prediction result by the control unit 5. The user visually recognizes the displayed prediction result and considers the behavior of the crowd.

<Function of Simulator 1>
FIG. 2 is a functional block diagram of the simulator 1 according to the first embodiment.

  The operation input unit 2 functions as a part of the condition setting unit 20 or the like. The file input / output unit 3 functions as a part of the condition setting unit 20 and a part of the prediction result output unit 57. The storage unit 4 functions as the repetition condition storage unit 40, the environment information storage unit 41, the individual information storage unit 42, and the like. The control unit 5 includes a risk value acquisition unit 50, a surrounding individual information acquisition unit 51, a stimulus value setting unit 52, a probability setting unit 53, a leader action determination unit 54, a follower action determination unit 55, an individual information update unit 56, and a prediction result output. It functions as a part of the unit 57 and the like.

  The condition setting unit 20 receives inputs of various conditions necessary for simulation, and stores the input conditions in the storage unit 4. The condition setting unit 20 is realized, for example, by the operation input unit 2 and the control unit 5 cooperating with each other, and the control unit 5 causes the storage unit 4 to store the value input by the user operating the operation input unit 2. . The condition setting unit 20 may include a file input / output unit 3 for inputting a file in which some or all of the conditions are written, and the control unit 5 sets the file designated by the user operating the operation input unit 2 as a file. The value obtained through the input / output unit 3 and written in the file is stored in the storage unit 4.

  The simulation is performed while advancing the virtual time within a predetermined range. Also, trials are repeated for virtual times within the same range. Although the conditions used in each trial are the same, the behavior of each individual at each virtual time of each trial is probabilistically determined, so that the simulation results between trials are rarely the same.

  The virtual time is a simulation of the passage of the actual time, but is advanced every time the simulator finishes the calculation for one hour, and is ticked faster than the passage of the actual time. For example, one virtual time is associated with an actual one second, and the virtual time is represented by an integer starting from 0 and incrementing by 1.

  In the simulation of this embodiment, assuming that the upper limit virtual time is T, the scheduled number of trials is M, and the virtual time is changed from 0 to (T-1), the trial of updating the environment and individual information for each virtual time is M. Repeated times.

  The repetition condition storage unit 40 stores the upper limit virtual time T and the scheduled trial count M. These values are set in advance by an input operation by the user using the condition setting unit 20.

  The environment information storage unit 41 stores information derived from the target space. Specifically, the environment information storage unit 41 stores a map of the target space including the exit area and a risk value indicating the magnitude of the risk occurring in the target space. The risk value is a function of virtual time. These values are set in advance by an input operation by the user using the condition setting unit 20.

  FIG. 3 schematically illustrates information stored in the environment information storage unit 41.

FIG. 3A shows a map 100 of the room R in a two-dimensional xy coordinate system. In the map 100, the floor surface of the room R is divided into blocks each having 128 squares in the x direction and 40 squares in the y direction as a movable area of an individual, and a part of the block is the area of the exit E ₁ and another of the blocks. Part of the area is the exit E ₂ . It is assumed that the center of the room R is the origin of the xy coordinate system and the x direction is north. The environment information storage unit 41 lists data in which x-direction grid numbers assigned to the centers of gravity of the blocks, y-direction grid numbers, and combinations of codes that distinguish the exit E ₁ from the exit E ₂ and others. Is memorized. The map 100 is a simple example for explaining the principle, but the target space may be a complicated space composed of a plurality of rooms and passages. In that case, the movable area has a complicated shape excluding areas such as walls and furniture.

  FIG. 3B schematically shows the risk value. The environment information storage unit 41 stores risk value sequences r (0), r (1), r (2), ..., R (T-1) from virtual time 0 to (T-1). In the present embodiment in which the event is a fire, the risk value is a value according to the scale of the fire, and is set to become larger as the virtual time advances.

  The individual information storage unit 42 stores at least an attribute value, a state value, and an action parameter for each of the plurality of individuals that form the crowd. Further, the individual information storage unit 42 may further store one or more of a reaction threshold value of each individual, a cognitive threshold value of each individual, and a stimulation value for each individual.

An identifier indicating whether the individual is a leader or a follower is set in the attribute value. Leaders and followers are the roles of individuals in the crowd. Leaders decide their actions based on their will. A follower decides his / her behavior according to the behavior of another individual.
In the present invention, the attribute value is dynamically changed as the virtual time passes. However, the initial value of the attribute value is common in each trial. That is, the initial value of the attribute value is set by the input operation by the user using the condition setting unit 20, the initial value is set to the attribute value each time the trial is performed, and the attribute value is set to the leader action determination unit 54 and It is changed by the follower action determination unit 55 at any time.
Incidentally, in this embodiment, the initial value of the attribute value is a follower common to all individuals.

The state value represents the state of each individual at each virtual time of each trial. The state value is the position of the individual in the target space and is represented by a grid number.
The initial value of the state value is also common in each trial and is updated as the virtual time passes. That is, the initial value of the state value is set by the input operation by the user using the condition setting unit 20, the initial value is set to the state value for each trial, and the state value is set by the individual information updating unit 56 described later. Updated from time to time.
Incidentally, in the present embodiment, the initial value of the position of each individual is a value randomly set from the grid numbers in the movable area of the room R.
In addition, in the present embodiment, the state value is one element only of the position, but in another embodiment, the state value may include a plurality of elements such as a value representing the position and the posture.

The action parameter represents the action of each individual at each virtual time of each trial, and is a parameter that acts on the state value. The behavior parameter is basically the purpose of the behavior of each individual, the means of the behavior determined by the individual to achieve the purpose, and the state value per hour by the individual performing the means of the behavior. It is defined as the amount of change (state change amount).
For example, the purpose is evacuation, and an identifier representing evacuation is stored as the purpose. However, in the present embodiment, the purpose is one type of evacuation, and thus the storage of the purpose may be omitted.
Further, for example, there are two kinds of means, that is, the movement to the exit E ₁ and the movement to the exit E ₂ , and the code representing the selected exit is stored as the means. That is, when the former is selected, the code E ₁ is stored in the means, and when the latter is selected, the code E ₂ is stored in the means.
Further, for example, the state change amount is a movement amount in a grid unit, and a set of an increment value in the x direction and an increment value in the y direction of the grid number is stored as the state change amount. That is, in the target space represented by the map of FIG. 3 (A), for example, when moving one grid to the north (0, -1), when moving one grid to the northwest (-1, -1), ... And so on. If the movement amount of the individual whose virtual time t is (5, 10) at the number of trials m is (-1, -1), the position of virtual time (t + 1) at the number of trials m is (4, 9). ) Is updated.
Incidentally, the initial value of the selection exit of each individual is not defined in this embodiment, and the initial value of the movement amount of each individual is (0, 0). The behavior parameter is changed at any time by the leader behavior determination unit 54 and the follower behavior determination unit 55, which will be described later.

The reaction threshold and the cognitive threshold are parameters expressing the individuality of an individual, and are preset by an input operation by the user using the condition setting unit 20.
The reaction threshold for each individual is a threshold for the stimulus value of the individual, and represents the difficulty of causing a reaction that causes an action by the stimulus value, that is, the difficulty of being changed to a leader. Individuals with higher response thresholds are less sensitive to stimuli, and individuals with lower response thresholds are more sensitive to stimuli. Therefore, an individual having a higher reaction threshold is less likely to be changed to a leader, and an individual having a lower reaction threshold is more likely to be changed to a leader.
The cognitive threshold for each individual represents the magnitude of the risk recognized by the individual. The higher the cognitive threshold, the more difficult it is for the stimulus value to increase in accordance with the increase in the risk value, and the lower the cognitive threshold, the easier the stimulus value increases. Therefore, an individual with a higher cognitive threshold is less likely to be changed to a leader, and an individual with a lower cognitive threshold is more likely to be changed to a leader.

The stimulus value is a value indicating the degree of stimulus that triggers an action. For example, the stimulus value is set as a value that increases every moment in response to an increase in the scale of a fire or a broadcast that prompts evacuation. The stimulus value for each individual is the stimulus value that the individual receives at each virtual time.
In this embodiment, the stimulus value is updated as the virtual time passes. However, the initial value of the stimulation value is common in each trial. That is, the initial value of the stimulation value is set by the input operation by the user using the condition setting unit 20, the initial value is set to the stimulation value each time the trial is performed, and the stimulation value is set by the stimulation value setting unit 52 described later. It is updated for each individual and for each virtual time.
By the way, in this embodiment, the initial value s (−1) of the stimulation value is 0 in common for all individuals.

  4 and 5 schematically illustrate information stored in the individual information storage unit 42.

An individual ID for identifying the individual is given to each of the N individuals, and the individual ID and information about the individual are stored in association with each other.
The information 110 of FIG. 4 includes the reaction thresholds θ ₀ , θ ₁ , ..., θ _N−1 and the cognitive threshold μ that are stored for each individual whose individual ID is 0, 1, ..., (N−1). ₀ , μ ₁ , ..., μ _N−1 are illustrated.
FIG. 5 shows an example of the information 111 at the end of the simulation. As the information 111, the individual IDs at the virtual times 0, 1, 2, ..., (T-1) in the 1, 2, ..., Mth trials are 0, 1, ..., (N-1). ), The history of stimulus values, the history of attribute values, the history of state values, and the history of action parameters are stored. For example, in the information 111, the stimulus value at each virtual time given to the individual whose individual ID is 0 in the first trial is s _1,0 (0), s _1,0 (1), s _1,0 (2), ..., s _1,0 (T-1), and in response to this, the attribute value of the same individual changes in the same trial as follower, leader, leader ,. , The selected exits of E ₁ , E ₁ , E ₁ , ..., E ₁ have changed, and the movement amount of the same individual in the same trial is (0, 0), (-1, -1), (-1, -1). , ..., (-1, -1), and the grid where the same individual is located in the same trial is (0, 0), (-1, -1), (-2, -2), ..., ( -64, -20) and the fact that the exit E ₁ is reached is stored.

  The risk value acquisition unit 50 sequentially acquires the risk value indicating the magnitude of the risk occurring in the target space and outputs it to the stimulation value setting unit 52. Specifically, the risk value acquisition unit 50 receives a request from the stimulation value setting unit 52 each time the virtual time t advances, reads the risk value r (t) at the virtual time t from the environment information storage unit 41, and performs stimulation. The value is output to the value setting unit 52.

The surrounding individual information acquisition unit 51 acquires, for each individual, information about surrounding individuals existing within a predetermined distance from the position of the individual, and outputs the information to the stimulus value setting unit 52 and the follower action determination unit 55.
One of the pieces of information acquired by the surrounding individual information acquisition unit 51 is the number of surrounding individuals, and the surrounding individual information acquisition unit 51 responds to the request from the stimulus value setting unit 52 specifying the individual ID, and the surroundings of the specified individual. Output the number of individuals. For example, the distance threshold T _D is set in advance to the radius of a circle that includes the neighboring 8 grids. The surrounding individual information acquisition unit 51 refers to the individual information storage unit 42, extracts an individual whose position within the radius T _D is set from the position of the specified individual as a surrounding individual, and extracts the number of surrounding individuals. Is counted and output.
Further, another piece of information acquired by the surrounding individual information acquiring unit 51 is the behavior of the surrounding individual, and the surrounding individual information acquiring unit 51 is designated in response to the request from the follower behavior determining unit 55 designating the individual ID. Outputs the behavior of individuals around them. Specifically, the surrounding individual information acquisition unit 51 extracts the surrounding individual in the same manner as above and outputs the action parameter set for the extracted surrounding individual.

The stimulus value setting unit 52 sets a stimulus value for each individual for each virtual time and outputs the set stimulus value to the probability setting unit 53. Specifically, the stimulus value setting unit 52 sets the stimulus value s _i (t) of each individual i (individual ID is i) at each virtual time t according to the following equations (1) to (4). .

However, s _i (t−1) is the stimulus value set for the individual i one time ago, and the stimulus value setting unit 52 acquires this from the individual information storage unit 42. Further, the stimulus value setting unit 52 causes the individual information storage unit 42 to store the stimulus value set at the current time for setting after one time. δ is the stimulus increase amount per one time, and a value larger than 0 is preset for the stimulus increase amount δ. α is a scale factor for suppressing stimulation, and a value of 0 or more is preset for the scale factor α. Further, r (t) is the above-described risk value, and the stimulation value setting unit 52 acquires this from the risk value acquisition unit 50. g is a non-negative constant. μ _i is the cognitive threshold for the individual i among the cognitive thresholds described above, and the stimulation value setting unit 52 acquires this from the individual information storage unit 42. _ni is the number of individuals around the individual i, and the stimulation value setting unit 52 acquires this from the surrounding individual information acquisition unit 51. N _max is the maximum number of surrounding individuals, and a value of 1 or more is set in advance.
Further, max {·, 0} in the formula (1) and “if · 0 otherwise” in the formulas (2) and (4) respectively indicate that the value is restricted to a non-negative value.

R _p of the equation (3) approaches 1 as the difference between the risk value r (t) and the cognitive threshold μ _i increases, and the stimulation value s _i (t) of the equation (1) increases as R _p approaches 1. To do. That is, the smaller the cognitive threshold μ _{i, the} greater the difference from the risk value r (t) and the more likely the stimulation value s _i (t) increases. This models the state of the real world in which the smaller the cognitive threshold, the more sensitive the risk is and the more likely a leader becomes.
In this way, the stimulus value setting unit 52 sets the stimulus value for each individual, and in particular, the greater the difference between the risk value r (t) and the cognitive threshold μ _i of the individual i, the more the individual information storage unit 42 stores. By greatly increasing the stored value s _i (t−1) and setting the stimulation value s _i (t), it is possible to make a prediction closer to the real world.

Further, F in Expression (4) approaches 0 as the number n _{i of} surrounding individuals increases, and the stimulation value s _i (t) in Expression (1) increases greatly as F approaches 0. That is, the stimulus value s _i (t) is likely to increase as the number of surrounding individuals increases. This is because the area where evacuation is less advanced promotes an increase in stimuli that trigger actions (the atmosphere in which evacuation is to be promoted), and the area where evacuation is advanced models the state of the real world where stimuli saturate. There is.
In this way, the stimulus value setting unit 52 sets the stimulus value for each individual, and in particular, the individual information storage unit 42 stores the individual having a larger number n _{i of} surrounding individuals extracted by the surrounding individual information acquisition unit 51. By setting the stimulus value s _i (t) by greatly increasing the value s _i (t−1), it is possible to make a prediction closer to the real world.

  The control of the number of trials and the virtual time is performed by the stimulation value setting unit 52.

  Each time the stimulus value setting unit 52 sets the stimulus value, the probability setting unit 53 sets the first winning probability based on the stimulus value to the individual whose attribute value is a follower among the individuals and sets the first winning probability. The winning probability is output to the leader action determining unit 54, a predetermined second winning probability is set for the individual whose attribute value is the leader, and the second winning probability is output to the follower action determining unit 55.

Specifically, the probability setting unit 53 reads out the individual ID (assumed to be i) of the individual whose attribute value is a follower and its reaction threshold θ _i from the individual information storage unit 42, and inputs them from the stimulus value setting unit 52. The stimulus value s _i (t) for the individual i is specified from the stimulus values, and the first winning probability P _i (X _i = 0 → X _i = 1) of the individual i is set according to the following equation (5).

However, X _i is a random variable corresponding to the attribute value of the individual i, X _i = 1 that the individual i is a leader, X _i = 0 that the individual i is a follower, and (X _i = 0 → X _i = 1) represents changing the individual i from a follower to a leader. That is, P _i (X _i = 0 → X _i = 1) is the probability of changing the individual i from a follower to a leader.

The first winning probability P _i (X _i = 0 → X _i = 1) of the equation (5) is higher as the stimulation value s _i (t) is higher, and is higher as the reaction threshold θ _i is lower. This models the state of the real world where the stronger the stimulus is, the easier it is to become a leader, and the more sensitive an individual is to become a leader.
As described above, the probability setting unit 53 has a higher first winning probability P _i (X _i = 0 → X _i = 1), which is higher as the stimulation value s _i (t) is higher and lower as the reaction threshold θ _i is lower for each individual. To set.

Further, the probability setting unit 53 identifies the individual ID (i) of the individual whose attribute value is the leader from the individual information storage unit 42, and the second winning probability P _i (X _i = 1 → X _i ) of the individual _i. = 0) is set according to the following equation (6).

Since X _i is the same as the description of the equation (5), P _i (X _i = 1 → X _i = 0) is the probability of changing from the leader to the follower. A value of 0 or more is set in advance in ε.

  Equations (5) and (6) are formulated following the reaction threshold model, and the probability setting unit 53 is characterized by an original approach in which the reaction threshold model is applied to change between a leader and a follower.

  The leader action determination unit 54 changes the attribute value of the individual for which the first winning probability is set by the probability setting unit 53 to the leader with the probability, and sets the state of the individual as the action parameter of the individual whose attribute value is the leader. Set a value that brings the value closer to achieving the objective.

Specifically, the leader action determination unit 54 performs a lottery with the first winning probability P _i (X = 0 → X = 1) for the individual i for which the first winning probability is set, and wins the lottery. In this case, the individual information storage unit 42 stores that the attribute value X _{i of the} individual (winning individual) _i is the leader.
In the above-mentioned lottery, for example, the leader action determination unit 54 determines one random number in the range [0, 1], and if the random number is less than P _i (X _i = 0 → X _i = 1), it is won, If P _i (X _i = 0 → X _i = 1) or more, it is determined that the player has failed.
Further, the leader action determination unit 54 reads the position of the individual whose attribute value is the leader from the individual information storage unit 42, calculates the distance from the position to each exit stored in the environment information storage unit 41, The nearest exit is determined as the selection exit of the winning individual.
The leader action determination unit 54 also calculates the distance from each of the eight neighborhood grids of the position of the individual whose attribute value is the leader to the selected exit, and determines the amount of movement from the position of the individual to the nearest neighborhood grid. . Then, the leader action determination unit 54 stores the determined evacuation purpose, the determined selection exit and the determined movement amount in the individual information storage unit 42 as the action parameter of the individual whose attribute value is the leader.

The follower action determination unit 55 changes the attribute value of the individual for which the second winning probability is set by the probability setting unit 53 to a follower with the probability, and sets the behavior parameter of the individual whose attribute value is the follower to the surroundings of the individual. Set a value similar to the behavior parameter of the individual.
Specifically, the follower action determination unit 55 performs the lottery with the second winning probability P _i (X = 1 → X = 0) for the individual i for which the second winning probability is set, and wins the lottery. attribute value X _i of the winning individual i is to be stored to be a follower to the individual information storage section 42 when.
In the lottery, for example, the follower action determination unit 55 determines one random number in the range [0, 1], and if the random number is less than P _i (X _i = 1 → X _i = 0), the winning If P _i (X _i = 1 → X _i = 0) or more, it is determined that the player has failed.
Further, the follower behavior determination unit 55 causes the surrounding individual information acquisition unit 51 to acquire the behavior parameter of the individual surrounding the individual whose attribute value is the follower, and selects the winning outlet with the largest number of selected exits among the acquired behavior parameters. Decide on the exit.
Further, the follower action determination unit 55 calculates the distance from each of the eight neighboring grids of the position of the individual whose attribute value is a follower to the selected exit, and determines the amount of movement from the position of the individual to the closest neighboring grid. . Then, the follower behavior determination unit 55 causes the individual information storage unit 42 to store that the purpose is evacuation and the determined selection exit and movement amount as behavior parameters of the individual whose attribute value is follower.

The individual information updating unit 56 updates the state value of each individual with the action parameter of the individual.
In the present embodiment, the individual information updating unit 56 updates by adding the movement amount of the action parameter to the position of the state value. Specifically, the individual information updating unit 56 reads the position and the movement amount of each individual from the individual information storage unit 42, adds the read movement amount to the read position for each individual, and obtains the position after the addition. It is stored in the individual information storage unit 42.

  The prediction result output unit 57 outputs the prediction result which is the result of the simulation. The prediction result includes at least a part or all of the information stored in the individual information storage unit 42, and preferably further includes a part or all of the information stored in the environment information storage unit 41. Further, the prediction result output unit 57 statistically processes the information stored in the individual information storage unit 42, or the information and the information stored in the environment information storage unit 41, and uses the obtained statistical amount as the prediction result. You may output.

  The output timing of the prediction result may include the middle of the simulation in addition to the time when the simulation is finished. That is, part or all of the above information obtained up to the middle of the simulation may be output as the intermediate prediction result.

  In addition, the prediction result output unit 57 may output the prediction result as a data file, an image file, or an image, and may use one or more of these formats. You may output.

  Each of FIG. 6 and FIG. 7 is an example of an intermediate prediction result at a virtual time close to the upper limit virtual time at a certain number of trials. These prediction results are obtained from the same settings. The prediction result output unit 57 images the map of the target space including the exits, superimposes a figure representing the individual on the coordinates corresponding to the position of each individual on the image to generate an intermediate prediction result, and the display unit Output to 6.

FIG. 8 is an example of a prediction result in which a statistic generated from information at the time when the simulation is finished is imaged as a graph. The horizontal axis of the graph is the index value E _q that represents the uniformity of the selected outlets, and the vertical axis is the frequency of trials. The index value E _q is a value obtained by multiplying the proportion S _{r of} individuals heading to the south exit E ₂ by the proportion N _r of individuals heading to the north exit E ₁ (E _q = S _r · N _r ). The prediction result output unit 57 counts the data whose virtual time is (T-1) among the data of the selection exits stored in the individual information storage unit 42 for each selection exit for each number of trials and calculates the total number of individuals. Is divided by to calculate S _r and N _r at each trial. Then, to calculate the E _q by multiplying the S _r and N _r for each number of trials, to calculate the frequency of each section was divided by attempts estimated count by counting the E _q of 13 sections each. Then, the image in which the frequency for each section is drawn as a graph is output to the display unit 6.

The intermediate prediction result 110 in FIG. 6 is an intermediate prediction result obtained in the course of the trial in which E _q close to 0.25 was calculated, and is the trial in the bin on the rightmost side in the graph of FIG. is there. That is, it was predicted that the trial in which the intermediate result of FIG. 6 was obtained did not cause a panic and the evacuation was performed by using the two exits almost equally.

On the other hand, the intermediate prediction result 120 of FIG. 7 is an intermediate prediction result obtained in the course of the trial in which E _q close to 0.00 was calculated, and is the trial prediction included in the leftmost bin in the graph of FIG. It is a thing. That is, in the trial in which the intermediate result of FIG. 7 was obtained, it was predicted that a panic in which evacuees were concentrated at the south exit E ₂ occurred.

  For example, if the prediction result by the simulator 1 is used, for example, an analysis is performed to compare the position of the leader and its time transition in the trial illustrated in FIG. 6 and the trial illustrated in FIG. It becomes possible to do. Based on the results of the investigation, it is possible to accurately and efficiently formulate a plan such as at which position the guide means should be installed or the guide member should be placed, and at what timing the guide means and the guide member should perform the guide. It becomes possible to do.

<Example of simulator 1 operation>
An operation example of the simulator 1 will be described with reference to the flowcharts of FIGS. 9 and 10.

  First, the condition setting unit 20 receives an input from the user and sets the conditions for simulation (step S1). That is, the simulation conditions obtained by the user manually inputting using the condition setting unit 20 and / or the user using the condition setting unit 20 to specify a file and the corresponding condition setting unit 20 are applicable. The condition setting unit 20 sets the simulation condition obtained by reading from the file in the iterative condition storage unit 40, the environment information storage unit 41, and the individual information storage unit 42.

  When the simulation condition is set, the stimulation value setting unit 52 initializes the number of trials to 1 (step S2), and sets the loop process of steps S2 to S12.

  Furthermore, in the loop processing of the number of trials, the stimulation value setting unit 52 initializes the virtual time to 0 (step S3), and sets the loop processing of steps S3 to S10.

  In the virtual time loop process, first, the action determination process is performed (step S4).

  The action determination process of step S4 will be described with reference to the flowchart of FIG.

  In the action determination process, first, the risk value is acquired (step S401). That is, the stimulus value setting unit 52 requests the risk value acquisition unit 50 to acquire the risk value by specifying the virtual time being set, and the risk value acquisition unit 50 stores the risk value corresponding to the specified time in the environmental information storage. It is read from the unit 41 and output to the stimulation value setting unit 52.

Next, the stimulation value setting unit 52 refers to the individual information storage unit 42 to identify an individual whose position has not yet reached the exit E ₁ or E ₂ , and sequentially sets the identified individual as a processing target. (Step S402), the loop processing of steps S402 to S417 is set.

  In the individual loop processing, first, surrounding individuals are counted (step S403). That is, the stimulus value setting unit 52 specifies the individual ID of the processing target individual and requests the surrounding individual information acquisition unit 51 to acquire information about the surrounding individual, and the surrounding individual information acquisition unit 51 refers to the individual information storage unit 42. Then, the surrounding individuals of the designated individual are counted, and the counted number of surrounding individuals is output to the stimulus value setting unit 52.

  Next, the stimulation value is updated (step S404). That is, the stimulus value setting unit 52 reads the stimulus value and the cognitive threshold of the individual to be processed one hour before from the individual information storage unit 42, the read stimulus value, the cognitive threshold, and the risk value acquired in step S401. Then, the number of surrounding individuals acquired in step S403 is applied to equations (1) to (4) to calculate a stimulus value, and the stimulus value is added to the individual information storage unit 42.

  Subsequently, the probability setting unit 53 refers to the individual information storage unit 42 and confirms the attribute value of the individual to be processed (step S405).

  When the attribute value is a follower (YES in step S405), the probability setting unit 53 reads the reaction threshold value and the latest stimulus value of the processing target individual from the individual information storage unit 42, and applies these to Expression (5). The first winning probability is calculated (step S406). Then, the leader action determination unit 54 performs a lottery with the calculated first winning probability (step S407), and when the winning is achieved (YES in step S408), the attribute value of the processing target individual in the individual information storage unit 42 is set. Change to a leader (step S409), and the process proceeds to step 413. If not won (NO in step S408), it is determined that the attribute value of the processing target individual is not changed, step S409 is omitted, and the process proceeds to step S415.

  On the other hand, when the attribute value is the leader (NO in step S405), the probability setting unit 53 sets the second winning probability based on the equation (6), and the follower action determining unit 55 sets the second winning probability. Lottery is performed (step S410). If the player has won (YES in step S411), the follower behavior determination unit 55 changes the attribute value of the processing target individual in the individual information storage unit 42 to a follower (step S412), and the process proceeds to step S415. If not won (NO in step S411), it is determined that the attribute value of the processing target individual is not changed, step S412 is omitted, and the process proceeds to step S413.

In step S 413, the leader action determination unit 54 stores the selection outlets of the individual whose attribute value has been changed to the leader and the individual outlets of the individual whose attribute value has not been changed from the leader in the individual information storage unit 42. An exit closer to the stored location of the individual is determined.
Then, the leader behavior determination unit 54 calculates the amount of movement of the position of the processing target individual to the nearest neighboring grid of the eight neighboring grids to the selected exit as the amount of movement of the individual (step S414).

On the other hand, in step S415, the follower behavior determination unit 55 makes the selection outlets of the individuals whose attribute values are determined to be changed to followers and the individuals whose attribute values are determined not to be changed from followers similar to the surrounding individuals. To decide. The follower behavior determination unit 55 causes the surrounding individual information acquisition unit 51 to acquire selection outlets of individuals around the processing target individual, and determines the largest number of selection outlets of the selection outlets of the processing target individual. Note that when the virtual time is 0 or close to 0, many selection outlets of surrounding individuals may be undecided. In that case, the selection outlets of the processing target individuals are also undecided.
Then, the leader action determination unit 54 calculates the amount of movement of the position of the processing target individual to the neighboring grid closest to the selected exit among the eight neighboring grids as the amount of movement of the individual (step S416).

  In the following step S417, it is determined whether or not the loop processing of the individual has ended. The stimulus value setting unit 52 confirms whether or not all the individuals who have not reached the exit have been processed, and if not processed (NO in step S417), returns the process to step S402 and Continue the loop processing of.

  On the other hand, when all the individuals who have not reached the exit have been processed (YES in step S417), the leader behavior determination unit 54 and the follower behavior determination unit 55 cause the behavior of each individual stored in the individual information storage unit 42. The parameters are updated (step S418). That is, the leader action determination unit 54 stores the selected exit determined in step S413 and the movement amount determined in step S414 in the individual information storage unit 42, and the follower action determination unit 55 determines the selected exit determined in step S415 and step S416. The moved amount is stored in the individual information storage unit 42.

  When the action parameter is updated, the action determination process ends, the process proceeds to step S5 in FIG. 9, and the individual information updating unit 56 uses the movement amount of each individual updated in step S4 to position the individual. To update.

  In the subsequent step S6, the prediction result is output. The prediction result output unit 57 refers to the environment information storage unit 41 to draw a map of the room R, refers to the individual information storage unit 42 to draw the latest position of each individual on the map, and draws the result. Is displayed on the display unit 6 as an intermediate prediction result.

Then, the end of the virtual time loop process is determined. The stimulus value setting unit 52 increases the virtual time by 1 (step S7), and refers to the individual information storage unit 42 to check whether all the individuals have reached the exit (step S8). Furthermore, when there is an individual that has not reached the exit (NO in step S8), the stimulation value setting unit 52 confirms whether the increased virtual time is equal to or more than the upper limit virtual time (step S9).
If there is an individual that has not reached the exit and the virtual time is less than the upper limit virtual time (NO in step S9), the stimulation value setting unit 52 returns the processing to step S4 and continues the loop processing of the virtual time. .

  On the other hand, when all the individuals have reached the exit (YES in step S8) or when the virtual time is equal to or more than the upper limit virtual time (YES in step S9), it is determined whether or not the loop processing for the number of trials has ended. The stimulation value setting unit 52 increases the number of trials by 1 (step S10), and confirms whether or not the increased number of trials exceeds the scheduled number of trials (step S11). When the increased number of trials is equal to or less than the planned number of trials (NO in step S11), the stimulation value setting unit 52 returns the process to step S3 and continues the loop process of the number of trials.

If the increased number of trials exceeds the scheduled number of trials (YES in step S11), the simulation ends and the prediction result is output.
The prediction result output unit 57 refers to the individual information storage unit 42, totals the frequencies of the index values E _q , draws a frequency graph, and displays the drawn graph image on the display unit 6 as a prediction result. The prediction result output unit 57 also writes all the data in the individual information storage unit 42 and all the data in the environment information storage unit 41 to a file, and outputs the file from the file input / output unit 3 as another prediction result.

[Second embodiment]
In the first embodiment, the individual information storage unit 42, the surrounding individual information acquisition unit 51, the stimulus value setting unit 52, the probability setting unit 53, the leader action determination unit 54, the follower action determination unit 55, and the individual information update unit 56. An example of performing processing for a plurality of individuals has been shown. Instead of these, the simulator 1b in the second embodiment has an individual information storage unit 42b-i, which performs processing for one individual, a surrounding individual information acquisition unit 51b-i, a stimulus value setting unit 52b-i, The behavior determining apparatus 70b-i including the probability setting unit 53b-i, the leader behavior determining unit 54b-i, the follower behavior determining unit 55b-i, and the individual information updating unit 56b-i is provided by the number N of individuals. However, i = 0, 1, 2, ..., N-1.
That is, in each of the action determination devices 70b-i (where i = 0, 1, 2, ..., N-1) included in the simulator 1b, the storage unit 4 functions as a memory region in parallel or in a time-division manner, It is realized by the control unit 5 executing processes that operate in parallel or in a time-division manner.

  That is, each of the behavior determination devices 70b-i is a behavior determination device that determines the behavior of its own individual, which is one of the individuals forming a crowd in a predetermined space that can achieve a predetermined purpose. An individual information storage unit 42b-i that stores at least an attribute value indicating whether a person is a leader or a follower, a state value including a position in space, and an action parameter that acts on the state value of an individual; A stimulus value setting unit 52b-i that sets a stimulus value that represents the degree of stimulus that triggers, and a first winning probability based on the stimulus value when the attribute value is a follower, and when the attribute value is a leader A probability setting unit 53b-i that sets a predetermined second winning probability, and when the first winning probability is set, the attribute value is changed to a leader with the probability and the attribute value is set to Leader action determining unit 54b-i that sets the state value to the action parameter that is close to the achievement of the purpose, and the actions of the surrounding individuals present within a predetermined distance from the position of the individual among the individuals are acquired. The surrounding individual information acquisition unit 51b-i that changes the attribute value to a follower with the probability when the second winning probability is set, and an action similar to the behavior of the surrounding individual when the attribute value is a follower And a follower action determination unit 55b-i that sets a parameter.

[Modifications of First and Second Embodiments]
In each of the above-described embodiments, an example in which the environment information storage unit 41 stores a two-dimensional space map is shown, but a three-dimensional map may be used. In that case, the state value and the behavior parameter stored in each of the individual information storage units 42 and 42-i are also three-dimensional information.

  In each of the above-described embodiments and the modifications thereof, a part of the simulation condition can be a value according to the situation of the real-time target space. For example, if the situation in the near future is predicted by a simulation in which the position of the actual person is set as the position of each individual, it becomes possible to accurately guide the crowd.

  Although the risk value acquisition unit 50 reads the risk value from the environment information storage unit 41 in each of the above-described embodiments and its modifications, in another embodiment, the risk value acquisition unit 50 includes a flame sensor and smoke. It may be configured by a function of the disaster sensor that measures a physical quantity of a real space related to a disaster such as a sensor and a function of the control unit 5 that processes signals from the disaster sensor according to a program. In this case, the risk value acquisition unit 50 quantifies a signal indicating the presence or absence of a disaster and the degree of the disaster into a risk value according to a predetermined rule.

  In each of the above-described embodiments and its modified examples, the individual information updating units 56 and 56b-i each update the position of the individual without considering the size of the individual and eliminating the overlap. The position of the individual may be updated in consideration of the physical constraint of. In that case, for example, the individual information update units 56 and 56b-i respectively set the movement amounts determined by the leader action determination units 54 and 54b-i and the follower action determination units 55 and 55b-i, respectively, in the virtual space. Adjust the amount of movement to a position where individuals do not overlap each other and update. Alternatively, for example, the individual information updating units 56 and 56b-i move the individuals closer to the selection areas determined by the leader behavior determination units 54 and 54b-i and the follower behavior determination units 55 and 55b-i, respectively. The amount of movement, which is the amount of movement to a position where the individuals do not overlap in the virtual space, is calculated and updated.

[Third embodiment]
The autonomous mobile robot 8 according to the third embodiment includes an action determination device 70c that is an example of the action determination device of the present invention, and simulates one individual forming a crowd.

  The autonomous mobile robot 8 has an operation input unit 2c, a file input / output unit 3, a control unit 5, and an operation input unit 2c similar to the display unit 6 of the first embodiment, a file input / output unit 3c, a control unit 5c, and a display unit. 6c, a disaster sensor such as a flame sensor or a smoke sensor, a surrounding individual sensor such as a camera, a timekeeping unit that measures real time, and a movement control unit including a motor, wheels, positioning means, obstacle sensors, and the like. Prepare Each of the disaster sensor, the surrounding individual sensor, the clock unit, and the movement control unit is connected to the control unit 5c. It is desirable that the operation input unit 2c and the display unit 6c be removable.

  The autonomous mobile robot 8 includes the condition setting unit 20, the repetition condition storage unit 40, the environment information storage unit 41, the individual information storage unit 42, the risk value acquisition unit 50, the surrounding individual information acquisition unit 51, and the stimulation value of the first embodiment. Conditions for performing processing for one individual corresponding to each of the setting unit 52, the probability setting unit 53, the leader action determining unit 54, the follower action determining unit 55, the individual information updating unit 56, and the prediction result output unit 57. Setting unit 20c, repetition condition storage unit 40c, environment information storage unit 41c, individual information storage unit 42c, risk value acquisition unit 50c, surrounding individual information acquisition unit 51c, stimulus value setting unit 52c, probability setting unit 53c, leader action determination unit. 54c, a follower action determination unit 55c, an individual information update unit 56c, and a prediction result output unit 57c.

However, the risk value acquisition unit 50c acquires the risk value by quantifying the signal output by the disaster sensor according to a predetermined rule.
In addition, the surrounding individual information acquisition unit 51c analyzes the image output by the surrounding individual sensor, detects a robot and a person of the same type as the autonomous traveling robot 8 existing around the autonomous traveling robot 8 as surrounding individuals, and detects the detected surrounding individuals. The number and behavior of surrounding individuals are acquired by recognizing the number and behavior of.
Further, the real time is used instead of the virtual time in the first embodiment, and the stimulus value setting unit 52c controls the time of the simulation with reference to the real time output by the time counting unit. The number of trials is advanced for each instruction from the user through the operation input unit 2c, and the stimulation value setting unit 52c refers to the instruction signal from the operation input unit 2c to control the number of trials.
In addition, the individual information updating unit 56c rotates the motor in the direction indicated by the movement amount of the action parameter up to the distance indicated by the movement amount, drives the wheels connected to the motor, and moves the autonomous mobile robot 8. Let In addition, the individual information updating unit 56c analyzes a signal from an obstacle sensor that detects an obstacle around the autonomous mobile robot 8 to detect a distance and a direction to the obstacle, and detects the obstacle and the autonomous mobile robot 8. The rotation of the motor is suppressed so that the distance is not less than the predetermined distance. Further, the individual information updating unit 56c analyzes the signal output by the positioning means including the communication interface for communicating with the beacon system and the GPS system, detects the position of the autonomous mobile robot 8, and obtains the state value at the detected position. Update.

  Of these, the individual information storage unit 42c, the surrounding individual information acquisition unit 51c, the stimulus value setting unit 52c, the probability setting unit 53c, the leader action determination unit 54c, the follower action determination unit 55c, and the individual information update unit 56c are the action determination devices. Included in 70c.

  That is, the behavior determining device 70c is a behavior determining device that determines the behavior of the individual, which is one of the individuals forming a crowd in a predetermined space that can achieve a predetermined purpose, and at least for the individual. , An individual information storage unit and 42c that stores an attribute value that indicates whether it is a leader or a follower, a state value that includes a position in space, and an action parameter that acts on the state value, and a degree of stimulation that triggers an action And a stimulus value setting unit 52c that sets a stimulus value that indicates the first winning probability based on the stimulus value when the attribute value is a follower, and a predetermined second winning probability when the attribute value is a leader And a probability setting unit 53c that changes the attribute value to the leader with the probability when the first winning probability is set, and the action parameter when the attribute value is the leader. A leader action determination unit 54c that sets a state value that approximates the achievement of the purpose, and a surrounding individual information acquisition unit 51c that acquires the actions of surrounding individuals within a predetermined distance from the position of the own individual among individuals. And when the second winning probability is set, the attribute value is changed to a follower with the probability, and when the attribute value is a follower, a follower action determination unit 55c that sets an action parameter similar to the action of surrounding individuals And

[Modifications common to the above embodiments]
In each of the above-described embodiments and its modifications, a room is illustrated as the target space, but the target space may be a concert hall, a station, or the like, and is not limited to indoors, but is also used for outdoor shopping malls, parks, towns, and the like. It may be a space, or a space that crosses indoors and outdoors such as at an event venue or school.

  In each of the above-described embodiments and its modifications, fire was illustrated as an event, but other disasters such as an earthquake, the appearance of a suspicious person, the occurrence of a sick person at an event site, the appearance of a celebrity, a call to the crowd, and question presentation. It can also be applied to events other than disasters such as.

  In each of the above-described embodiments and its modified examples, evacuation was illustrated as the purpose of the action, but rescue actions and field horse actions for the occurrence of sick people, field horse actions for the appearance of celebrities, responses to calls and question presentations, and other events are handled. The purpose can be set appropriately.

In each of the above-described embodiments and the modifications thereof, movement was illustrated as a means of action, but a shift to a leaning posture (evacuation under a sitting desk) as a means of evacuation against an earthquake, a response to a call or question presentation It is also possible to appropriately set means that does not involve movement according to the event and the purpose, such as raising hands and uttering as means of. It is also possible to set the movement to the exit and the transition to the bending posture as an option of the means, and to randomly select one of them.
In addition, in the said embodiment and its modification, although the means of action was two, if it is one or more, it may be other than two.

  In each of the above-described embodiments and its modifications, a crowd of people is illustrated, but the invention can be applied to a crowd of other social organisms such as fish, birds, and ants. In addition, when the individual is other than a person, the target space and the event can be appropriately set according to the type of the individual.

According to the simulator 1, the simulator 1b and the action determination device 70b-i included in the simulator 1b, the autonomous mobile robot 8 and the action determination device 70c included in the autonomous mobile robot 8 described above, even if the same setting is performed. The leader and follower are changed stochastically, and different prediction results can be generated even if the same settings are made. Therefore, each of these can predict a specific behavior such as panic behavior in a crowd under unknown conditions, and by using the prediction results obtained by each, it is possible to easily find out the condition in which the specific behavior occurs in the crowd.

1 ... Simulator 2 ... Operation input unit 20 ... Condition setting unit 3 ... File input / output unit 4 ... Storage unit 40 ... Repetition condition storage unit 41 ... Environmental information storage unit 42 ... Individual information storage unit 5 ... Control unit 50 ... Risk value acquisition unit 51 ... Surrounding individual information acquisition unit 52 ... Stimulation value setting unit 53 ... Probability setting unit 54 ... Reader Action determination unit 55 ... Follower action determination unit 56 ... Individual information update unit 57 ... Prediction result output unit 6 ... Display unit

Claims (7)

A simulator for predicting the behavior of a crowd in a predetermined space capable of achieving a predetermined purpose,
An individual information storage unit that stores at least an attribute value indicating whether a leader or a follower, a state value including a position in the space, and a behavior parameter that acts on the state value, for each of the plurality of individuals forming the crowd. When,
A stimulus value setting unit that sequentially sets stimulus values that represent the degree of stimulus that triggers the action;
Each time the stimulus value is set, a first winning probability based on the stimulus value is set for the individual whose attribute value is a follower among the individuals, and the individual whose attribute value is the leader has a predetermined probability. A probability setting unit that sets the winning probability,
The attribute value of the individual for which the first winning probability is set is changed to a leader with the probability, and the state value of the individual is set to the action parameter of the individual whose attribute value is the leader to achieve the purpose. A leader action decision unit that sets a value to approach,
A surrounding individual information acquisition unit that extracts, for each individual, a surrounding individual whose position from the individual is within a predetermined distance,
The attribute value of the individual for which the second winning probability is set is changed to a follower with the probability, and the behavior parameter of the individual whose attribute value is a follower is similar to the behavior parameter of the surrounding individuals of the individual. A follower action determiner that sets the value
An individual information updating unit for updating the state value of each individual with the action parameter of the individual,
A simulator characterized by having. The individual information storage unit further stores a reaction threshold value of each individual,
The probability setting unit sets the first winning probability higher for each individual as the stimulus value is higher and as the reaction threshold is lower.
The simulator according to claim 1.   The simulator according to claim 1, wherein the stimulus value setting unit sets the stimulus value for each individual. Further comprising a risk value acquisition unit that sequentially acquires a risk value representing the magnitude of the risk occurring in the space,
The individual information storage unit further stores a cognitive threshold representing the magnitude of the risk recognized by each of the individual,
The stimulus value setting unit stores the stimulus value set for each individual in the individual information storage unit, and the individual information has a larger difference between the risk value and the cognitive threshold of the individual for each individual. The stimulus value is set by greatly increasing it from the value stored in the storage unit,
The simulator according to claim 3.   The stimulus value setting unit stores the stimulus value set for each individual in the individual information storage unit, and the larger the number of the surrounding individuals extracted by the surrounding individual information acquisition unit, the more the individual information storage unit. The simulator according to claim 3, wherein the stimulus value is set by greatly increasing the stored value. A simulation method for predicting the behavior of a crowd in a predetermined space capable of achieving a predetermined purpose, comprising:
On the computer,
For each of the plurality of individuals that make up the crowd, store at least an attribute value indicating whether it is a leader or a follower, a state value including a position in the space, and a behavior parameter that acts on the state value,
A process of sequentially setting a stimulus value indicating the degree of stimulus that triggers the action,
Each time the stimulus value is set, a first winning probability based on the stimulus value is set for the individual whose attribute value is a follower among the individuals, and the individual whose attribute value is the leader has a predetermined probability. The process of setting the second winning probability,
The attribute value of the individual for which the first winning probability is set is changed to a leader with the probability, and the state value of the individual is set to the action parameter of the individual whose attribute value is the leader to achieve the purpose. The process of setting the value to approach,
A process of extracting surrounding individuals whose position from the individual is within a predetermined distance for each individual;
The attribute value of the individual for which the second winning probability is set is changed to a follower with the probability, and the behavior parameter of the individual whose attribute value is a follower is similar to the behavior parameter of the surrounding individuals of the individual. A follower action determiner that sets the value
An individual information updating unit for updating the state value of each individual with the action parameter of the individual,
A simulation method characterized by causing to execute. A behavior determination device for determining the behavior of one's own individual, which is one of the individuals constituting a crowd in a predetermined space capable of achieving a predetermined purpose,
For the own individual, at least an attribute value indicating whether it is a leader or a follower, a state value including a position in the space, and an individual information storage unit that stores an action parameter that acts on the state value,
A stimulus value setting unit that sets a stimulus value indicating the degree of stimulus that triggers the action;
A probability setting unit that sets a first winning probability based on the stimulus value when the attribute value is a follower, and sets a predetermined second winning probability when the attribute value is a leader,
When the first winning probability is set, the attribute value is changed to a leader with the probability, and when the attribute value is a leader, the action parameter is set to a value that brings the state value close to the achievement of the purpose. A leader action decision section to set,
Among the individuals, a surrounding individual information acquisition unit that acquires the behavior of surrounding individuals existing at a predetermined distance or less from the position of the own individual,
When the second winning probability is set, the attribute value is changed to a follower with the probability, and when the attribute value is a follower, a follower action that sets the action parameter similar to the action of the surrounding individual is set. The decision section,
An action determination device comprising:

Images