Method, device and electronic device for simulating transcending behavior based on action element

Abstract

The present disclosure provides a method, device and electronic device for simulating overtaking behavior based on action elements, wherein the method comprises: determining the target personality parameters of a target pedestrian based on the association relationship between pre-constructed trait factors and the personality parameters of the overtaking pedestrian, and determining at least one action meta-model matching the target pedestrian based on the mapping relationship between the personality parameters and the pre-constructed action meta-model; during the movement of the target pedestrian, judging whether the current walking environment of the target pedestrian meets the overtaking condition; when it is determined that the current walking environment does not meet the overtaking condition, determining the target action meta-model that meets the energy consumption requirements in the at least one action meta-model, and determining the movement trajectory of the target pedestrian through the target action meta-model and the social force model.

Description

Technical Field

[0001] The present disclosure relates to the technical field of data processing, and in particular to a method, device and electronic device for simulating transcending behavior based on action elements. Background Art

[0002] As an important part of transportation, pedestrian travel behavior has always been a matter of great concern. Compared with other components of transportation, pedestrians have higher heterogeneity in travel decisions, both physically and mentally, and have greater freedom, which leads to the diversity of pedestrian travel behavior.

[0003] However, the related technical solutions still have pedestrian travel accidents. The main causes of the accidents are the speed differences of pedestrians in high-density environments and the lack of advance anticipation by managers. For example, the main cause of these accidents is that the pedestrians behind want to reach the front (i.e., overtake the pedestrians in front). The related technical solutions simulate the overtaking behavior of pedestrians, but the simulation scheme does not simulate the overtaking behavior well, resulting in the inability to accurately simulate the travel behavior and movement trajectory of heterogeneous pedestrians. Summary of the invention

[0004] The embodiments of the present disclosure at least provide a method, device and electronic device for simulating transcending behavior based on action elements.

[0005] In a first aspect, an embodiment of the present disclosure provides a method for simulating overtaking behavior based on action meta-models, comprising: determining target personality parameters of a target pedestrian based on a correlation relationship between pre-constructed trait factors and personality parameters of the overtaking pedestrian, and determining at least one action meta-model matching the target pedestrian based on a mapping relationship between the personality parameters and a pre-constructed action meta-model; during the movement of the target pedestrian, determining whether the current walking environment of the target pedestrian meets the overtaking condition; when it is determined that the current walking environment does not meet the overtaking condition, determining a target action meta-model that meets the energy consumption requirements in the at least one action meta-model, and determining the movement trajectory of the target pedestrian through the target action meta-model and the social force model.

[0006] In the second aspect, an embodiment of the present disclosure provides an overtaking behavior simulation device based on action meta-models, including: a first determination unit, used to determine the target personality parameters of a target pedestrian based on the association relationship between pre-constructed trait factors and the personality parameters of the overtaking pedestrian, and determine at least one action meta-model matching the target pedestrian based on the mapping relationship between the personality parameters and the pre-constructed action meta-models; a judgment unit, used to judge whether the current walking environment of the target pedestrian meets the overtaking condition during the movement of the target pedestrian; a second determination unit, used to determine the target action meta-model that meets the energy consumption requirements in the at least one action meta-model when it is determined that the current walking environment does not meet the overtaking condition, and determine the movement trajectory of the target pedestrian through the target action meta-model and the social force model.

[0007] In a third aspect, an embodiment of the present disclosure further provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the above-mentioned first aspect, or any possible implementation of the first aspect are performed.

[0008] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned first aspect or any possible implementation of the first aspect are executed.

[0009] The present disclosure provides a method, device and electronic device for simulating overtaking behavior based on action elements. In an embodiment of the present disclosure, first, based on the association relationship between the pre-constructed trait factors and the personality parameters of the overtaking pedestrian, the target personality parameters of the target pedestrian are determined, and based on the mapping relationship between the personality parameters and the pre-constructed action meta-model, at least one action meta-model matching the target pedestrian is determined; then, during the movement of the target pedestrian, it is determined whether the current walking environment of the target pedestrian meets the overtaking condition; when it is determined that the current walking environment does not meet the overtaking condition, a target action meta-model that meets the energy consumption requirements is determined in at least one action meta-model, and the target pedestrian's motion trajectory is determined through the target action meta-model and the social force model.

[0010] In the above implementation, the personality characteristics of the target pedestrian can be determined by pre-constructing the association between the trait factors and the personality parameters, and at least one action meta-model matching the personality characteristics of the target pedestrian can be determined by pre-constructing the mapping relationship between the personality parameters and the action meta-model. When the current walking environment of the target pedestrian meets the overtaking condition, the target pedestrian's action trajectory can be determined by the target action meta-model that meets the energy consumption requirements, so that the overtaking behavior can be simulated according to the pedestrian's personality characteristics, thereby achieving a more realistic simulation of the overtaking behavior.

[0011] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following is a brief introduction to the drawings required for use in the embodiments. The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used together with the specification to illustrate the technical solutions of the present disclosure. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can also be obtained based on these drawings without creative work.

[0013] Figure 1 A flow chart of a method for simulating a transcending behavior based on action elements provided by an embodiment of the present disclosure is shown;

[0014] FIG2 (a) shows a schematic diagram of a force analysis of a pedestrian walking normally provided by an embodiment of the present disclosure;

[0015] FIG2( b ) shows a schematic diagram of force analysis of a turn provided by an embodiment of the present disclosure;

[0016] FIG3 (a) shows a schematic diagram of a pedestrian density identification area provided by an embodiment of the present disclosure;

[0017] FIG3( b ) shows a schematic diagram of another pedestrian density recognition area provided by an embodiment of the present disclosure;

[0018] FIG3 (c) shows a schematic diagram of a region division of an action region provided by an embodiment of the present disclosure;

[0019] Figure 4 A schematic diagram of force analysis of a pushing behavior provided by an embodiment of the present disclosure is shown;

[0020] Figure 5 An AMMDM framework flow chart provided by an embodiment of the present disclosure is shown;

[0021] Figure 6 A schematic diagram showing a fitting result of a relationship between an angle and a speed provided by an embodiment of the present disclosure;

[0022] Figure 7 A schematic diagram of a transcendence experiment provided by an embodiment of the present disclosure is shown;

[0023] Figure 8 A simplified schematic diagram of a local density decision area provided by an embodiment of the present disclosure is shown;

[0024] Fig. 9 A schematic diagram showing a pedestrian recognition area provided by an embodiment of the present disclosure;

[0025] Fig.10 A schematic diagram showing a parameter setting provided by an embodiment of the present disclosure is shown;

[0026] Fig.11 A schematic diagram showing the setting of a simulation experiment provided by an embodiment of the present disclosure is shown;

[0027] Fig.12 A schematic diagram showing trajectory simulation results of different pedestrians provided by an embodiment of the present disclosure;

[0028] Fig.13 A schematic diagram showing the number of each alternative behavior occurring in a simulation experiment provided by an embodiment of the present disclosure;

[0029] FIG14( a ) is a schematic diagram showing a comparison between the model provided by an embodiment of the present disclosure and a basic diagram of related studies;

[0030] FIG14( b ) is a schematic diagram showing a comparison of the unidirectional channel simulation capacity of the model provided by the embodiment of the present disclosure and the original SFM;

[0031] Fig.15 A schematic diagram of a transcending behavior simulation device based on action elements provided by an embodiment of the present disclosure is shown;

[0032] Fig.16 A schematic diagram of an electronic device provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the technical scheme in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. The components of the embodiments of the present disclosure generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure provided in the drawings is not intended to limit the scope of the present disclosure for protection, but merely represents the selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present disclosure.

[0034] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0035] The term "and / or" herein only describes an association relationship, indicating that three relationships may exist. For example, A and / or B may represent the following three situations: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C may represent including any one or more elements selected from the set consisting of A, B, and C.

[0036] The research found that the relevant technical solution simulated the overtaking behavior of pedestrians, but the simulation solution did not simulate the overtaking behavior well, resulting in the inability to accurately simulate the travel behavior and movement trajectory of heterogeneous pedestrians.

[0037] Based on the above research, the present disclosure provides a method, device and electronic device for simulating overtaking behavior based on action elements. In the embodiment of the present disclosure, first, based on the association relationship between the pre-constructed trait factors and the personality parameters of the overtaking pedestrian, the target personality parameters of the target pedestrian are determined, and based on the mapping relationship between the personality parameters and the pre-constructed action meta-model, at least one action meta-model matching the target pedestrian is determined; then, during the movement of the target pedestrian, it is determined whether the current walking environment of the target pedestrian meets the overtaking condition; when it is determined that the current walking environment does not meet the overtaking condition, a target action meta-model that meets the energy consumption requirements is determined in at least one action meta-model, and the target pedestrian's motion trajectory is determined through the target action meta-model and the social force model.

[0038] In the above implementation, the personality characteristics of the target pedestrian can be determined by pre-constructing the association between the trait factors and the personality parameters, and at least one action meta-model matching the personality characteristics of the target pedestrian can be determined by pre-constructing the mapping relationship between the personality parameters and the action meta-model. When the current walking environment of the target pedestrian meets the overtaking condition, the target pedestrian's action trajectory can be determined by the target action meta-model that meets the energy consumption requirements, so that the overtaking behavior can be simulated according to the pedestrian's personality characteristics, thereby achieving a more realistic simulation of the overtaking behavior.

[0039] To facilitate understanding of this embodiment, a method for simulating overtaking behavior based on action elements disclosed in the embodiment of the present disclosure is first introduced in detail. The execution subject of the method for simulating overtaking behavior based on action elements provided in the embodiment of the present disclosure is generally an electronic device with certain computing capabilities. In some possible implementations, the method for simulating overtaking behavior based on action elements can be implemented by a processor calling computer-readable instructions stored in a memory.

[0040] See also Figure 1 FIG. 1 is a flowchart of a method for simulating overtaking behavior based on action elements provided by an embodiment of the present disclosure, wherein the method includes steps S101 to S103, wherein:

[0041] S101: Determine target personality parameters of a target pedestrian based on a correlation relationship between pre-constructed trait factors and personality parameters of a surpassing pedestrian, and determine at least one action metamodel matching the target pedestrian based on a mapping relationship between the personality parameters and pre-constructed action metamodels.

[0042] In the disclosed embodiment, an action meta-model of at least one action meta-element is pre-constructed, and an association relationship between a trait factor and a personality parameter of a transcendental person is pre-constructed, and a mapping relationship between the action meta-model and the personality parameter is constructed.

[0043] Here, the target personality parameters of the target pedestrian can be determined based on the association relationship. After obtaining the target personality parameters, at least one action metamodel matching the target pedestrian can be determined in the pre-constructed action metamodel based on the mapping relationship.

[0044] S102: During the movement of the target pedestrian, determining whether the current walking environment of the target pedestrian satisfies a surpassing condition.

[0045] In the disclosed embodiment, it is possible to determine whether the current walking environment of the target pedestrian satisfies the surpassing condition by means of global density.

[0046] Here, the density of pedestrians in the decision area of ​​the target pedestrian can be determined, and then the density is determined as the global density; wherein, if it is determined that the global density is greater than or equal to the specified density threshold 1, it is determined that the current walking environment of the target pedestrian does not meet the surpassing condition. If it is determined that the global density is less than the specified density threshold 1, it is determined that the current walking environment of the target pedestrian meets the surpassing condition.

[0047] Ideally, the global density The calculation formula is: ,in, Refers to the number of pedestrians in the decision area of ​​the i-th pedestrian at the t-th time. In real life, the pedestrian's line of sight will be blocked by other pedestrians in front, and the global density cannot be accurately identified. Therefore, the average crowd density in the experimental area can also be determined. Then, this density is determined as the global density. Specifically, in order to determine the impact of different global densities on the pedestrian decision-making mechanism, the technical solution disclosed in the present invention introduces the level of service (LOS) to characterize the density, and takes the average crowd density in the experimental area as the global density. If it is determined that the global density is greater than or equal to the specified density threshold 2, it is determined that the current walking environment of the target pedestrian does not meet the surpassing condition. If it is determined that the global density is less than the specified density threshold 2, it is determined that the current walking environment of the target pedestrian meets the surpassing condition. Among them, the specified density threshold 1 and the specified density threshold 2 may be the same or different.

[0048] S103: When it is determined that the current walking environment does not satisfy the surpassing condition, a target action metamodel that satisfies the energy consumption requirement is determined in the at least one action metamodel, and the action trajectory of the target pedestrian is determined through the target action metamodel and the social force model.

[0049] In the embodiment of the present disclosure, a target action metamodel with the lowest energy consumption can be determined in at least one action metamodel, and then the target action metamodel can be input into the social force model for solution to obtain the position and speed of the target pedestrian at the next moment, thereby determining the movement trajectory of the target pedestrian based on the position and speed.

[0050] The above process will be described in detail below in conjunction with specific implementation methods.

[0051] From the above description, it can be seen that the technical solution disclosed in the present invention pre-constructs the correlation between the trait factors and the personality parameters that surpass pedestrians. The specific construction process is described as follows:

[0052] First, trait factors and personality parameters are selected.

[0053] The personality of pedestrians can be described by the strength of several specific trait factors based on personality trait theory. In order to determine these trait factors, various personality models have been established in advance, such as the PEN model, the OCEAN model, and the MBTI model. Among them, the OCEAN model is proposed on the basis of the PEN model, while the dimensions contained in the MBTI model are too broad and not suitable for detailed research on pedestrian overtaking behavior. In addition, there are some self-designed questionnaires to measure the relationship between personality trait factors and behaviors, such as the Pedestrian Behavior Scale (PBS) and the Pedestrian Behavior Questionnaire (PBQ), but these questionnaires are mainly aimed at the travel behavior of road users and are difficult to transplant to a travel environment with only pedestrians. Therefore, the disclosed technical solution selects the PEN model for personality trait testing, and selects the following three personality trait factors (traits):

[0054] Psychoticism is used to measure the pedestrian's aggressiveness and ego strength. The higher the score, the more aggressive the behavior. Extraversion mainly measures the pedestrian's explicit or implicit tendencies. High scores correspond to some active and impulsive behaviors, and low scores correspond to quiet and stable behaviors. Neuroticism refers to the degree of emotional stability, which can correspond to shyness and nervousness.

[0055] In addition, when measuring pedestrian personality in a specific environment, some personality adjectives are usually preset to describe pedestrian behavior. Therefore, based on the research on personality and pedestrian walking behavior, five commonly used personality adjectives are summarized to describe the behaviors that pedestrians may exhibit during the overtaking process:

[0056] Aggressive (Ag): Their goal is to reach the destination in the shortest distance, so they walk fast, change direction less, and do not mind colliding with other pedestrians; Impulsive (I): This type of pedestrian will reach the destination as quickly as possible without colliding with others, so they will constantly adjust their walking direction during the journey, look for gaps and overtake others; Confident (As): This can be called a steady type. They will maintain their own walking speed while keeping a certain distance from others and adjust their behavior according to their perception of the environment; Shy (S): They will walk along the edge of the channel to overtake others and minimize contact or collision with others; Nervous (T): This type of pedestrian will feel nervous when overtaking others, so they usually choose to give up overtaking behavior, reduce speed, and walk behind other pedestrians.

[0057] Secondly, the correlation between trait factors and personality parameters is constructed.

[0058] In the disclosed embodiments, transformation sets characterized by different parameters can be introduced into factor sets (a set of trait factors) and personality sets (a set of personality parameters), and the relationships between the transformation sets and the factor sets and personality sets are respectively constructed, thereby deriving a mapping relationship between trait factors and preset personalities.

[0059] Here, the transformation set B is introduced, and the characteristic factor set and personality set The mapping relationship is as follows: , , where the relationship parameters can be described as: , .

[0060] Among them, in the trait factor set, Psychoticism means psychoticism, Extraversion means extraversion, and Neuroticism means neuroticism. In the personality set, Aggressive means aggressive, Assertive means confident, Shy means shy, Active means active, Tense means nervous, and Impulse means impulsive.

[0061] because and are all non-square matrices, and The number of rows is greater than the number of columns, On the contrary, therefore, No solution, There are infinite solutions. Therefore, in order to determine the relationship between P and F, the equation is optimized according to the least squares method, that is, satisfy: .

[0062] In addition, according to the pseudo-inverse matrix calculation method, It can be calculated as ,in, and The matrices are The transpose and pseudo-inverse of . Therefore, we can conclude that After another pseudo-inverse transformation, the mapping relationship between P and F can be obtained as follows:

[0063] .

[0064] In order to fit the five personality adjectives proposed in the technical solution of the present disclosure, the personality set is modified. The personality that goes beyond pedestrians can be obtained from the scores of the three trait factors measured by the PEN model. The final correlation is as follows:

[0065] .

[0066] Secondly, determine the mapping relationship between personality parameters and action metamodels. Psychoticism represents psychoticism, Extraversion represents extraversion, Neuroticism represents neuroticism, Aggressive represents aggressiveness, Assertive represents self-confidence, Shy represents shyness, Tense represents nervousness, and Impulse represents impulsiveness.

[0067] In real life, pedestrians usually use action elements such as turning (R), detouring (D) and pushing (P) to achieve overtaking behavior, or give up overtaking and adopt the action element of following (F). Therefore, four action elements that may appear in the overtaking process are proposed here. Since the concept of action element is first applied to the simulation of overtaking behavior, based on the above personality classification analysis, a set of optional action elements is boldly and roughly created for pedestrians i with different personalities: , , , and .

[0068] Next, a mathematical model of at least one action element of surpassing the pedestrian is constructed to obtain an action element model; wherein the at least one action element includes at least one of the following: detouring, following, pushing, and turning.

[0069] Here, four action element models will be established to simulate pedestrian motion. Obviously, except for the turning action element, the body postures of the other three action elements are almost unchanged, which can be achieved by updating the driving force or interaction force in the original social force model SFM. Here, the modeling process of the four action elements will be specifically introduced.

[0070] (1) Orbital action modeling.

[0071] The essence of the detour action element is that pedestrians spontaneously adjust the driving force according to their assessment of the surrounding environment to avoid collisions while maintaining walking speed. Therefore, it is necessary to update the driving force of the original SFM. In addition, in order to avoid unreasonable reaction forces caused by obstacle pedestrians to overtake pedestrians, the technical solution of the present disclosure introduces anisotropic parameters to adjust the interaction force. Therefore, the detour action element is modeled as:

[0072] .

[0073] in, and are the detour coefficient and the anisotropy coefficient, is the normalized distance vector The direction of pedestrian i The angle between is a parameter that grows with the strength of the interaction from behind, and They represent correction factors respectively to avoid unreasonable large forces between pedestrians or close obstacles.

[0074] (2) Follow-up action meta-modeling.

[0075] When the density does not support overtaking, the pedestrian will adjust the speed and follow the pedestrian in front to complete the subsequent walking behavior. To distinguish it from the detour action element modeling, here, instead of changing the driving force and interaction force to establish the follow-up action element model, a follow-up force is added to realize the simulation of the follow-up action element:

[0076] .

[0077] Among them, H is the correction parameter, Indicates whether pedestrian j is within the sight range of pedestrian i. is the sight radius of pedestrian i, It means that pedestrians are more inclined to follow pedestrians closer to their current direction. It means that pedestrians tend to follow faster pedestrians. To specify a constant.

[0078] (3) Modeling of small push actions.

[0079] The premise of pushing should be in high-density conditions. Here, it is assumed that the pedestrian who initially overtakes can feel the interaction force of other pedestrians. When the action of overtaking pedestrians is hindered at higher speeds, they will gradually lose the sense of rejection of the pedestrians in front. This loss of feeling is proportional to their actual speed, and the pushing action will occur at the moment when the feeling is completely lost. Therefore, the model of the pushing action element is:

[0080] .

[0081] Where C is a parameter that determines the curvature of the loss of perception of others, It is the threshold speed at which you lose patience.

[0082] (4) Metamodeling of turning action: naive turning model.

[0083] In order to better simulate the pedestrian turning action element, a simplified narrow area scenario and two overtaking scenarios were established for turning behaviors. This application selected an ellipse as the model of the pedestrian's body shape.

[0084] According to the turning formula of the ellipse and considering the psychological distance of pedestrians during the movement, we can get the distance of pedestrian i in , and The minimum acceptable turning angles in the three cases are as follows:

[0085] .

[0086] Where L is the channel width, and are the shoulder width and chest thickness of pedestrian i, m and n are two different pedestrians being overtaken, y m ,y n and w are the position ordinal numbers of pedestrians m, n and obstacle w, respectively, w is the psychological distance between pedestrian i and other pedestrians or obstacles, and are the shoulder widths of pedestrians m and n respectively.

[0087] At the same time, the distance D from the pedestrian to the obstacle or pedestrian when the pedestrian starts to turn is introduced. T and the turning start time T b To determine the timing of turning. According to Newton's laws of motion, the coordinate systems of the three scenes are established. By deducing the coordinates of the known points, the timing of turning in the three scenes can be obtained as follows:

[0088] ;

[0089] ;

[0090] .

[0091] in, is the actual speed of the pedestrian at time t, It's turnaround time.

[0092] In addition, in order to simulate the turning action element, a torsion force F is added to the original SFM. T To change the body posture, its relationship with the turning inertia I and torque T is as follows: , where l=a is the force arm, a r =dω / (dt r ) is the angular acceleration, ω=dθ / (dt r ) is the angular velocity, θ∈[-π / 2,π / 2] is the turning angle, t r is the turning time. The moment of inertia of the ellipse I can be calculated as I=1 / 4m(a 2 +b 2 ), so the torque of pedestrian i can be rewritten as: At the same time, in order to avoid repulsion interfering with the turning action element, the turning intention is also added to the original SFM ,Right now:

[0093]

[0094] It should be noted that the above model is derived under the assumption that the pedestrian's walking speed will not change during the turning process. However, this assumption is unreasonable, and turning will inevitably cause the pedestrian's walking speed to decrease. In addition, in the overtaking process, the most important factor is the walking speed, which is also the basis for ensuring the authenticity of the pedestrian overtaking behavior simulation. Therefore, the present disclosure attempts to reveal the relationship between the turning angle and the walking speed, and to add the speed change process to the turning action meta-modeling.

[0095] (5) Meta-modeling of turning action: a turning model based on the angle-speed relationship.

[0096] According to the principles of biomechanics, all human behaviors consume energy. When a pedestrian turns while walking, the body converts part of the forward energy into torsional energy, thereby changing the posture of the human body and completing the turning behavior. Therefore, this application aims to analyze the change in walking speed of turning pedestrians from the perspective of human energy consumption. It should be noted that this application also assumes that the direction of movement is fixed, that is, the pedestrian maintains the original direction of movement when turning. At the same time, this application takes into account the psychological distance when calculating the turning angle, so it can be considered that the effect of the pedestrian turning angle on reducing the walking speed is the same in the three cases. Therefore, in this application, only the simplest scenario of crossing a narrow area is used to discuss the reduction in the walking speed of turning pedestrians. In addition, in order to simplify the calculation and simulation process, similar to previous studies, this application ignores the human skeletal structure, muscles, joints and other anatomical features. Although this method may limit the degree of freedom of the model, it has been proven to effectively simulate the behavior and movement of pedestrians.

[0097] According to the first law of thermodynamics, pedestrians who turn while maintaining the direction of movement need to convert part of their walking energy into torsional energy. On the other hand, according to the second law of thermodynamics, pedestrians' turning behavior also involves entropy, that is, part of the energy is consumed inside the human body. This part of energy is irreversible and can no longer be used and converted. This article simplifies it as internal consumption. Based on this, this application constructs an energy consumption model for pedestrians during normal movement and turning movement.

[0098] The force analysis of pedestrian motion is shown in Figure 2 (a), where m and v0 are set to be the pedestrian’s mass and initial walking speed, respectively. When the pedestrian is walking straight, the pedestrian will be under the driving force F i , the interaction force F from other pedestrian j ij and the repulsive force F from the wall w iw The final forward force F is formed on the basis of the speed, and its direction is consistent with the speed direction. At the same time, the pedestrian will encounter the ground friction f in the opposite direction of movement. In this case, the energy consumed by the movement is E m will be converted into forward kinetic energy E w , friction energy Ef and internal friction E i ,Right now:

[0099] (1).

[0100] At the same time, as shown in Figure 2 (b), the pedestrian's turning process is divided into two stages: changing body posture and restoring body posture. In the first stage, when the pedestrian turns, he will be affected by a pair of opposing balanced forces Fr acting on the two vertices of the long axis of the ellipsoid model, thereby generating a torque T, which enables the pedestrian to achieve a turning behavior with an angle of θ. It should be noted that although the friction force is only related to the roughness and pressure of the contact surface, when the pedestrian turns, the applied pressure will increase, and the friction force will theoretically change. In order to facilitate subsequent calculations, the change in friction force is incorporated into the change in forward force, and the walking speed will be reduced to .

[0101] Therefore, the energy consumption of a pedestrian turning By forward momentum , torsional energy Friction energy and internal friction Composition, the calculation formula is: (2).

[0102] in, is the moment of inertia of the ellipsoidal model, is the angular velocity of the pedestrian, a and b (a>b) are constants, representing the semi-major and semi-minor axes of the ellipse, respectively, and t represents the time step.

[0103] In addition, due to the internal consumption of the human body, it cannot be assumed that the energy consumption of the above two processes is equal. But it is certain that there is an inevitable relationship between the two, namely Combining formulas (1) and (2), we can get the relationship between walking speed and turning angle as follows:

[0104] .

[0105] Obviously, the internal friction of pedestrians is related to their weight, so Simplified to the constant term A. At the same time, in order to establish the equation between walking speed and turning angle, three parameters α, β and γ are introduced as the influencing factors of each item, and the interference factor ,Right now:

[0106] (3); The turning angle θ needs to be converted into the form of π. According to previous research results, t = 0.5 seconds is set.

[0107] It should be noted that during the stage of restoring body posture, pedestrians need to consume their own energy to generate counter-torque, turn around to restore body posture and increase walking speed. It is assumed that pedestrians will restore their speed to the required speed, and the acceleration process is affected by the physical fitness of pedestrians and environmental conditions, so the time required will vary.

[0108] In summary, formula (3) shows that the decrease in walking speed is proportional to the turning angle. When the pedestrian turns to a larger angle, the decrease in walking speed is more obvious. This is correct from the perspective of both biomechanics and pedestrian dynamics. The turning model based on the angular velocity relationship is as follows:

[0109] .

[0110] in, is the mass of pedestrian i, is the actual speed, J and W represent the number of pedestrians and walls respectively, is the expected speed at time t, is the unit vector pointing to the desired target at time t, τ is the characteristic time, Indicates driving force, represents the interaction force between pedestrian i and pedestrian j, represents the interaction force between pedestrian i and the wall, and They represent the torque and turning intention of pedestrian i, respectively, α, β and γ are the influencing factors of each item, is the interference factor, a and b (a>b) are constants, A is the constant term, represents the angular acceleration of pedestrian i at time t.

[0111] In the embodiment of the present disclosure, the method further includes the following steps:

[0112] Step S104: when it is determined that the current walking environment satisfies the surpassing condition, determining the regional utility score of each action area of ​​the target pedestrian;

[0113] Step S105: When it is determined that the target pedestrian meets the macro decision-making conditions, the movement trajectory of the target pedestrian is determined according to the target movement area corresponding to the maximum regional utility score.

[0114] In actual motion, pedestrians will judge whether there is enough space to complete continuous overtaking behavior based on the perceived overall density of the area, and determine the direction in which the overtaking behavior can be completed based on the local density of the surrounding area. At the same time, in order to avoid collisions with others, pedestrians will set up personal areas around themselves, which will change with the heterogeneity of pedestrians.

[0115] Based on this, this application sets multiple action areas for the target pedestrian. Here, it is assumed that the pedestrian only pays attention to the situation in front and ignores the perception from the back. On this basis, the following areas can be set for the target pedestrian: personal area, action area and decision area. For example, as shown in Figure 3 (a), Figure 3 (b) and Figure 3 (c), the smallest sector area is the personal area, which consists of a radius of r p The semicircular area is represented by the radius r. The larger sector area and the largest sector area are the action area and decision area for identifying local and global density, respectively. a , r d and the angle θ i , The fan-shaped area is represented by .

[0116] Normally, pedestrians have a viewing angle of 170° and a viewing depth of about 10m. However, studies have found that as the viewing depth increases, the viewing angle decreases, and pedestrians will change their viewing areas according to their needs when making different judgments. Therefore, this application selects r p =0.8m, r a =3m、r d =10m,θ i =170° and =120° is used as the parameter of the pedestrian density recognition area.

[0117] The present application may calculate the global density in the manner described above, and determine the regional utility scores of each action area of ​​the target pedestrian when it is determined based on the global density that the current walking environment of the target pedestrian satisfies the surpassing condition.

[0118] In the embodiment of the present application, a maximum density threshold is preset. and minimum density threshold Wherein, if it is determined that the global density is within the threshold range (ie, the global density is greater than the minimum density threshold and less than the maximum density threshold), it is determined that the current walking environment of the target pedestrian meets the surpassing condition.

[0119] In the embodiment of the present application, the pedestrian's action area is pre-divided into 5 non-overlapping sector areas, each sector area is used as an action area, as shown in Figure 3 (c). The coverage of each area (the angle occupied is shown above each area) corresponds to a set of walking directions pointing to the left (L), left front (FL), front (F), right front (FR) and right (R). .

[0120] In the related art, a commonly used method is to calculate the comprehensive utility score of each area based on the speed difference and position of pedestrians in each area, the shortest path preference and social norms, so as to determine the walking direction. However, the technical solution of the present application introduces the factor of pedestrian heterogeneity, and pedestrians with different personalities do not necessarily choose their behavior completely according to the regional utility. For example, in order to reduce the detour distance, aggressive pedestrians may choose crowded areas and surpass others by detouring, which obviously does not take into account the utility of the area. Therefore, in the present application, local density recognition will only output the regional utility score of each area to the subsequent decision-making steps, and will not directly determine the walking direction of pedestrians.

[0121] Since the goal of this application is to pass in one-way traffic, only the speed at which pedestrians can fully utilize the space to complete the passing behavior is considered when calculating the zone utility score. Each zone will be scored based on the position and speed of pedestrians in that zone:

[0122] (4). Among them, is the score of each action zone (i.e., zone score), Z = {L, FL, F, FR, R}; n is a constant; k < 0 is a parameter, indicating that each pedestrian in the action zone will reduce the zone utility; is the speed of pedestrian j in zone Z; is the expected speed of target pedestrian i direction; represents the inner product of parameters; is the distance between pedestrians; Used to avoid undefined items.

[0123] In addition, pedestrians with different personalities have different compliance with density-based behavior choices. Therefore, this application introduces a compliance parameter C of pedestrian i: i , which is a discrete random variable with values ​​of 1 and 0, representing whether the pedestrian complies with or rejects the decision result based on local density. Complying with the decision result based on local density means that the macro decision condition is met, and rejecting the decision result based on local density means that the macro decision condition is not met. Here, P(C i ) is the probability of a random variable, which can be obtained through actual experiments. In summary, when it is determined that the target pedestrian meets the macro decision-making conditions, the pedestrian will choose the target action area with the largest regional utility score, that is: , and then determine the movement trajectory of the target pedestrian according to the target movement area.

[0124] In the embodiment of the present disclosure, the above steps of determining the regional utility score of each action area of ​​the target pedestrian specifically include the following steps:

[0125] Step S1041: determining a regional score of the target pedestrian in each of the action areas according to the personality parameters of the target pedestrian; wherein the regional score is a position and speed score of the target behavior in the action area;

[0126] Step S1042: Determine the region score as the region utility score corresponding to the action region.

[0127] After research, it was found that the influence of shortest path preference and social norms cannot be ignored. On this basis, additional evaluation criteria for pedestrians with different personality parameters are proposed as follows:

[0128] Aggressive pedestrians like to go straight to their destination and do not like to change direction, so it is necessary to add a specific score to formula (4), that is, ,in, is a constant, is a binary variable used to improve the regional utility score of the front area.

[0129] Impulsive pedestrians have a weaker perception of spatial utility. They do not care about the change of direction, but are more concerned about whether they can pass quickly. Therefore, the calculation method of regional utility score is improved to ,in, It is a monotonically decreasing exponential function, which can reduce the influence of pedestrian speed and position in the area.

[0130] Confident pedestrians are rarely affected by the environment, so their perception of the area utility rating does not need to change.

[0131] Shy pedestrians are more attracted to the sides of the area, so a specific score is added in the left and right directions: , where S∈{L,R}, is a constant based on pedestrian habits, is a binary variable used to improve the regional utility score of the left and right regions, respectively.

[0132] Nervous pedestrians are more sensitive to the availability of the area and more aware of other pedestrians in the area, so a positive index is added to enhance the impact of pedestrians. The utility score of the area with intense pedestrians is .

[0133] In the above implementation, by determining the area score according to the personality parameters of the target pedestrian, the action area where the overtaking behavior can be completed can be determined more accurately.

[0134] In the embodiment of the present disclosure, the above steps determine the target pedestrian's action trajectory according to the target action area corresponding to the maximum area utility score, specifically including:

[0135] First, the regional information of the target action area is input into the social force model for processing;

[0136] Secondly, the social force model is optimally solved, and the position and speed of the target pedestrian at the next moment are determined according to the optimal solution result.

[0137] In the disclosed embodiment, after the target action area is determined, the area information (e.g., area utility score) and global density of the target action area can be input into the social force model for processing, so as to obtain the position and speed of the target pedestrian at the next moment, and thus obtain the action trajectory of the target pedestrian. The specific model structure of the social force model will be introduced in the following implementation.

[0138] From the above description, we can see that when pedestrian i moves in the simulated environment, it will make a density-based macro decision, which is divided into two steps: first, according to the global density ( ) to determine whether the current environment requires or can implement transcendental behavior. On the one hand, when the global density is small enough ( ), pedestrians can walk freely. On the contrary, when the global density is high enough ( ), pedestrians cannot overtake. and are the global density thresholds for free flow and crowded flow, respectively.

[0139] When the global density is within the threshold range, the regional utility score based on the local density is calculated. ), make the next macro decision: when the pedestrian meets the calculation based on local density ( =1), it selects the optimal direction with the highest score (i.e., the target action area), calculates the final force, and updates the velocity and position; otherwise ( =0), micro-decisions based on energy consumption will be made.

[0140] It can be seen from the above description that when it is determined that the current walking environment meets the transcending condition, a target action metamodel that meets the energy consumption requirement can be determined in the at least one action metamodel, specifically including the following steps:

[0141] Step S11: calculating the total energy consumption of the target pedestrian when performing the overtaking behavior according to the actions indicated by each action meta-model;

[0142] Step S12: Determine that the action metamodel corresponding to the minimum total energy consumption is the target action metamodel that meets the energy consumption requirement.

[0143] When macro-decision cannot determine which direction the pedestrian should overtake, micro-decision comes into play, which introduces a new parameter: energy consumption. Pedestrians usually use the least amount of energy to achieve their goals. When the action element cannot be selected after considering the individuality and density factors, we can try to establish an energy consumption model of pedestrians under different action elements and select the action element with the least energy consumption to implement the overtaking behavior.

[0144] According to the principle of minimum effort, the energy consumed by a person moving along the path Π can be written as:

[0145] (5).

[0146] Among them, R, m, v and ω are the pedestrian's heat consumption rate, mass, instantaneous speed and instantaneous angular velocity, respectively. , and are three parameters, representing the efficiency of heat use, the energy consumption rate when stationary, and the difficulty of turning, and t is the time step. When the pedestrian does not change direction (detour) or body posture (turn around), the last item is 0 by default.

[0147] Therefore, when a pedestrian walks along a straight line at a time interval τ, the energy consumption is Obviously, when pedestrians overtake others through various actions, their speed, direction, or body posture will change, thus consuming additional energy. The additional energy consumption of each action element will be introduced below.

[0148] Detour: During this process, pedestrians will adjust their direction of movement and lose a certain amount of speed. becomes When the pedestrians have an extra energy consumption of

[0149] .

[0150] Combined with formula (5), the total energy consumption of detour behavior is: ; .

[0151] Following: Assume that the walking speed is reduced to the speed of the pedestrian in front (V f ) are the same and keep the direction unchanged, so the energy consumed by deceleration can be simply calculated as the difference in kinetic energy, that is, the total energy consumption of the following behavior is expressed as follows:

[0152] .

[0153] Pushing: It should be noted that the pushing behavior mentioned in the technical solution of this disclosure does not include using arms or shoulders to deliberately collide with others, but is a behavior of quickly filling the space gap. In this process, pedestrians will ignore "courtesy" and social distance and complete the overtaking behavior when contacting others. This behavior can maintain the ideal speed before contact, and only needs to increase the driving force at the moment of contact to offset the impact force, thereby offsetting the speed loss and body posture changes caused by the impact (such as Figure 4 As shown in Figure 2). Therefore, the additional energy consumption of the pushing behavior is used to maintain walking speed and body posture. In addition, the energy consumption of body posture change is related to the impact point. For the convenience of calculation, the energy consumption of body posture change is simplified to kinetic energy loss. Therefore, when a pedestrian needs to push n pedestrians to overtake, the total energy consumption is:

[0154] ;in, .

[0155] As mentioned above, assuming that the pedestrian is a rigid body, according to the momentum conservation theorem of rigid body collision, it can be roughly assumed that the speed of the pedestrian after the collision is the same, that is:

[0156] , .

[0157] Turning: By considering the change in speed during turning, the additional energy consumed by turning can be calculated. According to the turning model based on the angular velocity relationship, the total energy consumption of turning is:

[0158] , , .

[0159] After calculating the total energy consumption of each overtaking action, the action element with the minimum total energy consumption can be selected to perform the overtaking action. At the same time, the speed calculated based on the minimum energy consumption will be used to update the speed and position of the next time step.

[0160] In the embodiment of the present disclosure, determining the target pedestrian's action trajectory through the target action meta-model and the social force model specifically includes:

[0161] First, the model information of the target action meta-model is input into the social force model for processing;

[0162] Secondly, the social force model is optimally solved, and the position and speed of the target pedestrian at the next moment are determined according to the optimal solution result.

[0163] In the embodiment of the present disclosure, after the target action meta-model is determined, the target action meta-model can be input into the social force model for processing, so as to obtain the position and speed of the target pedestrian at the next moment, and thus obtain the movement trajectory of the target pedestrian. In the embodiment of the present disclosure, the social force model includes:

[0164]

[0165] in, is the mass of pedestrian i, is the actual speed of pedestrian i, is the actual speed of pedestrian j, is the expected speed at time t, is the unit vector pointing to the desired target at time t, τ is the characteristic time, , are constants, representing the intensity and range of social interaction forces, , is the sum of the radii of pedestrian i and pedestrian j and the distance between their centers of mass, is the distance between pedestrian i and the center of mass of wall w, , are the normalized vectors from pedestrian j and wall w to pedestrian i, , are the body compression coefficient and sliding friction coefficient, , is the tangent direction, J and W represent the number of pedestrians and walls, and g(x)=max(x, 0) indicates whether pedestrians are not touching each other or touching the wall; and are the detour coefficient and anisotropy coefficient respectively; is the normalized distance vector The direction of pedestrian i The angle between and are the correction coefficients to avoid unreasonable huge forces between pedestrians and between pedestrians and close obstacles; H, , ,and is the correction parameter of the following action element, is the specified constant; is the strength of the social interaction force that pushes the action element; and They represent the torque and turning intention of pedestrian i, respectively. Indicates driving force, represents the interaction force between pedestrian i and pedestrian j, represents the interaction force between pedestrian i and the wall; and are the global density and the regional utility score calculated from the local density, respectively; For pedestrians at a speed of The minimum energy consumption when represents the angular acceleration of pedestrian i at time t, represents the actual position of pedestrian i at time t, and are the shoulder width and chest thickness of pedestrian i, θ∈[-π / 2,π / 2] is the turning angle, is the turning time of pedestrian i, represents the friction between pedestrian i and the ground, represents the congestion flow threshold, Indicates the free flow threshold.

[0166] The following will be combined Figure 5 The technical solution of the present disclosure is introduced. Figure 5 The figure shows an action-based macro-to-micro decision-making (AMMDM) framework for simulating fine-grained heterogeneous pedestrian overtaking behavior. The simulation flow chart is shown in Figure 5 As shown. Specifically, the AMMDM framework is used to simulate the decision-making process of overtaking behavior and includes two modules. The AM module models the action elements in the overtaking process and constructs the mapping relationship between these action elements and multiple personality features. Therefore, the AM module is used to generate personality-based heterogeneous pedestrians in the simulation environment and output the action elements that pedestrians with different personalities can choose in the overtaking process to the MMDM module. The MMDM module makes density-based macro decisions and energy-based micro decisions in turn, and obtains the pedestrian behavior of the next time step based on pedestrian attributes and environmental factors.

[0167] like Figure 5 As shown, first select trait factors and personality parameters. Trait factors include psychoticism, extraversion, and neuroticism. Personality parameters include: aggressive (Ag), impulsive (I), confident (As), shy (S), and nervous (T). Secondly, construct the association between trait factors and personality parameters. Next, construct an action metamodel and determine the mapping relationship between personality parameters and action metamodel. Among them, action metamodels include: turn (R), detour (D), follow (F), and push (P).

[0168] In the disclosed embodiment, the target personality parameters of the target pedestrian can be determined based on the correlation between the pre-constructed trait factors and the personality parameters of the surpassed pedestrian, and at least one action metamodel matching the target pedestrian can be determined based on the mapping relationship between the personality parameters and the pre-constructed action metamodel. In specific implementation, the start of the target pedestrian is detected. After the start of the target pedestrian is detected, it is determined whether to make a macro decision based on density based on the global density; wherein, if it is determined that the global density is within the threshold range (that is, the global density is greater than the minimum density threshold and less than the maximum density threshold), it is determined that the current walking environment of the target pedestrian meets the surpassing condition. At this time, it can be determined to make a macro decision, otherwise a micro decision is made based on energy consumption. If it is determined that a macro decision needs to be made based on density, the regional utility score of the action area is calculated, and then it is determined whether the target pedestrian meets the calculation based on local density.

[0169] If it is determined that the calculation based on local density is met, the highest utility direction is determined. For example, the direction corresponding to the action area corresponding to the maximum regional utility score (i.e., the target action area) can be determined as the highest utility direction. Next, the social force is calculated, wherein the regional information of the target action area can be input into the social force model for calculation processing, thereby outputting the calculation result, and updating the position and speed of the target pedestrian according to the calculation result. Afterwards, it is determined whether the target pedestrian has reached the destination. If it is determined that it has, the process is terminated, otherwise the step of detecting the start of the target pedestrian is executed.

[0170] If it is determined that the calculation based on local density is not met, the total energy consumption of each action element corresponding to the target pedestrian is calculated, and after the action element corresponding to the minimum total energy consumption is determined, the position and speed of the target pedestrian can be updated. After that, it is determined whether the target pedestrian has reached the destination. If it is determined that it has, the process ends, otherwise the step of detecting the start of the target pedestrian is executed.

[0171] The verification process of the technical solution disclosed in the present invention is briefly introduced below.

[0172] 1. Design of actual experiment and data collection

[0173] (1) Overall situation of data collection.

[0174] First, several volunteers were selected, and the specific information of each volunteer was collected, including gender (male / female), age, height (cm), weight (kg), walking speed (m / s), shoulder width (m), chest thickness (m), and personality.

[0175] (2) Questionnaire setting and data collection.

[0176] The questionnaire of this study is divided into two parts: the first part uses a mature personality questionnaire based on the PEN model to score the personality traits of the subjects and determine the personality characteristics of the subjects. The second part of the questionnaire contains 4 simulated videos in different density scenes, which are randomly extracted from the video library to test the obedience of pedestrians with different personalities.

[0177] 2. Data processing and analysis of actual experiments

[0178] On the one hand, according to the results of the questionnaire survey, the proportion of pedestrians with different personalities was roughly calibrated, that is, Ag:I:As:T:S = 2:5:7:2:4. On the other hand, the obedience probability of pedestrians with different personalities was obtained by combining the results of the two parts of the questionnaire. By normalizing the data, the average value of the results was selected to calibrate the obedience parameter. It can be seen that pedestrians with aggressive and impulsive personalities have lower obedience, which is consistent with the actual situation. However, the obedience of pedestrians with shy personalities fluctuates greatly.

[0179] 2.1 Rotation experiment

[0180] (1) Setup and data collection

[0181] First, a narrow channel was constructed, and the width of the narrow channel was set to L = 0.3m, 0.4m, 0.5m, and 0.6m. When the volunteers passed through the narrow channel, the rotation parameters of the pedestrians, such as rotation angle and angular velocity, were collected. Among them, 30 sets of data were obtained for each L, a total of 120 data sets. Subsequently, the parameters of the rotation model based on the angle-velocity relationship were calibrated through 120 sets of video data and sensor data.

[0182] (2) Data processing and parameter calibration

[0183] The collected data were processed, and the processing results determined the invalid data in the 120 data sets. For example, 33 data sets without rotation behavior and 75 data sets with rotation behavior were obtained. By analyzing the 33 data sets without rotation behavior, it can be seen that different channel widths have no significant effect on the change of walking speed. Therefore, it can be proved that the change of walking speed in these scenarios can be regarded as normal fluctuations, and the change of walking speed in rotation behavior is less affected by psychological factors.

[0184] By analyzing the data with rotation behavior, it can be seen that according to the horizontal comparison results of different pedestrians in each scene, as L decreases, the rotation angle increases, and the walking speed decreases more, proving that the walking speed is affected by the rotation angle. On the other hand, the longitudinal comparison results of each pedestrian in different scenes prove that although there are differences in the recovery ability of pedestrians, they will eventually return to the desired speed.

[0185] The study also found that the start time of deceleration for each pedestrian is different in each scene, and all pedestrians will decelerate in advance, and it is found that the duration of deceleration is almost equal to the rotation time of the pedestrian to the maximum angle. Therefore, the deceleration start time is set to the rotation start time here.

[0186] Since the flow velocity always fluctuates, the average flow velocity before and after deceleration is selected to calibrate the model parameters. A multiple regression analysis was performed and the fitting results were: Note that in order to keep the magnitude of the parameters as consistent as possible, the value of θ is converted from degrees to radians, i.e. At the same time, In addition, the fitting surface is Figure 6 As shown, it can be seen that the R-Squared and adjusted R-Squared of the fitting results are 0.906 and 0.902 respectively, which proves that the fitting results are valid and the obtained parameters can be used for subsequent simulation studies.

[0187] 2.2 Beyond Experimentation

[0188] (1) Setup and data collection

[0189] Here, if Figure 7 As shown, a 11m x 6m recording area can be set on the ground with red tape, and this area is evenly divided into 5 2.2m x 6m rectangular areas. The camera is directly above the center line of the recording area. At the same time, the volunteers are divided into crowds and overtaking pedestrians (OP). The role of the crowd is to create pedestrian traffic at different densities, randomly distributed in the waiting area, and pass through the recording area at a normal walking speed; the number of crowds ranges from 2 to 22, with an increment of 2. Here, the changes in the acceleration and rotation angle of the OP also need to be recorded, and they are required to pass through the recording area at the fastest speed from the starting line without violating their walking habits, and the OP is required to start 3 seconds later than the crowd.

[0190] Assumption A: For the local density of pedestrian perception, the sector area is roughly simplified to a rectangular area (see Figure 8 ), whose area and sensing range are similar to sectors. The density of the area outside the OP is taken as the local density.

[0191] Assumption B: Approximate the density of the waiting area as the global density.

[0192] (2) Data processing and parameter calibration

[0193] Here, pedestrian head motion trajectories are obtained from 330 video records of the transcendence experiment, and the changes in pedestrian walking speed are calculated.

[0194] By comparing the walking behaviors of 30 volunteers at different densities, it was found that when the LOS belongs to the three situations A, B, and C, the pedestrians are in the free flow stage and can freely complete the overtaking behavior according to their own attributes without the need for a complex decision-making process, and the pedestrians' walking speed hardly changes. On the other hand, when the global density is greater than 1.75 p / m2, the gap for overtaking is very small, and even radical and impulsive pedestrians have to slow down and follow after failing to overtake. This phenomenon proves the authenticity of the decision-making process proposed by the technical solution of the present disclosure. Specifically, when the OP is in a free flow state, there is no need to impose too many constraints on the overtaking behavior, while the overtaking behavior cannot be implemented in a crowded state.

[0195] Therefore, combined with the experimental results, the free flow threshold and congestion flow threshold in the simulation model are set to 0.71 p / m2 and 1.75 p / m2, respectively.

[0196] 2.3. AMMDM framework rationality verification

[0197] (1) The rationality of personality classification and action elements.

[0198] Here, the number of times each personality type of pedestrian performs overtaking behavior using different actions is counted. Note that natural overtaking behavior (i.e., no obstacles in front of the pedestrian during the entire movement) will not be counted.

[0199] In addition, a qualitative analysis of the walking behaviors and trajectories of OPs with different personalities was conducted, and the results are as follows:

[0200] Aggressive pedestrians usually have higher walking speeds and larger body sizes. Due to their size advantage, these pedestrians are not afraid of colliding with others, and at the same time, their faster walking speed consumes more energy to slow down, so they usually choose to change their body posture (rotate) to keep themselves in a straight line or detour slightly.

[0201] Impulsive pedestrians are the type of pedestrians that generate the most rotations and detours. As analyzed before, impulsive pedestrians want to reach their destination quickly without colliding with others, so they often have rotation or detour motions, but no pushing motions. On the other hand, this type of pedestrian has the fastest walking speed among the participants in the experiment and does not have a large body size, which ensures that they have enough flexibility to complete the rotation and detour motions.

[0202] Confident pedestrians walk at a more stable speed, and even when they are tasked to reach the destination as quickly as possible, they still walk steadily and rarely accelerate or decelerate. Under different global densities, the trajectories of these pedestrians can be clustered as walking and following on the original route, or trying to take a detour and follow others.

[0203] Nervous pedestrians show the highest proportion of following actions. They tend to avoid overtaking after approaching the crowd, but choose to slow down and adapt to the speed of the crowd. After research, it was found that some nervous pedestrians repeatedly searched for overtaking routes while walking, but ultimately failed to overtake and chose to follow others. Therefore, it can be speculated that the reason for this phenomenon is that in real life, such pedestrians rarely overtake and it is difficult to find a suitable path to overtake in the first place. This also proves the rationality of the mapping relationship between action elements and personality proposed in this application.

[0204] Shy pedestrians will adjust their walking direction in the original density selection area to move closer to the edge of the channel. Interestingly, this seemingly roundabout way is the most effective way to implement overtaking behavior, because there are fewer pedestrians at the edge, and they can complete the overtaking behavior with the least amount of action. Moreover, by integrating the trajectories of all shy pedestrians, it can be seen that when walking halfway or approaching the destination, there are still many detours to the edge, which is consistent with the psychological attributes of shy pedestrians.

[0205] In general, in this application, the classification of personality and action elements is reasonable, and the mapping relationship between the two is supported by real experimental data.

[0206] (2) The rationality of the rotation model based on angular velocity relationship

[0207] After statistics, there are 37 groups of OPs who completed overtaking behaviors through rotation action elements, of which the data are almost all obtained from aggressive and impulsive pedestrians, only 3 groups of data are from confident pedestrians, and shy and nervous pedestrians did not have rotation elements. This is consistent with the results of TPA mapping.

[0208] Subsequently, the walking speed before and after the rotation behavior and the maximum rotation angle of the OP were extracted from the above 37 sets of data. In addition, the real date was substituted into , find the speed. Obviously, all the fitted speeds after the rotation are smaller than the true value. The reason for this may be that in the preliminary rotation experiment, the pedestrians can reach the destination very quickly after completing the rotation behavior without much acceleration, while in the overtaking experiment, the OP needs to complete the subsequent walking process at a faster speed after completing the rotation behavior. Therefore, they will consume more internal energy to ensure that the speed reduction is as small as possible.

[0209] In addition, by calculating the error between the fitted and true speeds after the rotation behavior, it can be seen that the average error is 7.85% and the maximum speed difference is 0.149 m / s. These results once again prove the rationality of the rotation modeling of the ABS module.

[0210] (3) Rationality of MMDM module

[0211] After research, it was found that in all the overtaking experiments, only one group showed pushing behavior. This was an active pedestrian. Therefore, the rationality of the MMDM module can be roughly verified through this special set of data.

[0212] Specifically, the walking speed before the OP's pushing behavior was calculated. (ID = 26, v 26 = 1.398m / s) and two pushed pedestrians (ID = 7&28, v7 = 0.862m / s, v 28 = 0.833m / s, the distances between volunteers 7 and 28 are ( ), the straight-line distance between OP and the two promoted persons ( ), and the walking angle (θ 26 = 59.7°). On the one hand, the aggressive pedestrian will not choose to slow down to follow others, and on the other hand, the rotation behavior is not possible under the current conditions. Therefore, according to the energy consumption determination stage of the DM module, the energy consumption of the OP with detour and push behavior is E d = 347.66 J and E p = 215.91 J. Obviously, the aggressive pedestrian does not obey the macroscopic decision based on density, but at the same time, he unconsciously chooses the action element with the least energy consumption under this condition to complete the overtaking behavior, which is consistent with the microscopic decision based on energy consumption proposed in this application.

[0213] From the perspective of pedestrian perception, when the walking speed is too fast and the distance to other pedestrians is relatively close, it is difficult for pedestrians to perform other actions, so they choose to push the action element. From the perspective of energy consumption, although pedestrians cannot calculate their energy consumption during the movement, they still subconsciously choose the action element with lower energy consumption to complete the overtaking behavior. Therefore, this special case proves the rationality of the MMDM module to a certain extent.

[0214] 3. Simulation environment settings.

[0215] As an improvement of the original SFM model, the AMMDM model can simulate various tasks of heterogeneous pedestrians more realistically. After comparing the operability of various SFM-based simulation software for modifying the underlying logic, Anylogic Professional 8.7.0 and JAVA 2.0 were selected as the basic software, and a simulation model based on the AMMDM framework was established. The simulation was run on an Intel(R) Core(TM) i5-10210U1.60GHZ PC with 4GB of memory.

[0216] First, the pedestrian's body model is changed from the original circle to an ellipse through Anylogic's agent module. At the same time, according to the classification of pedestrian personality, 6 types of subjects are constructed: aggressive (Ag), impulsive (I), confident (As), nervous (T), shy (S) and normal (N). It should be noted that this application does not discuss the impact of gender on overtaking behavior, and the differences between genders will be reflected in physiological attributes such as weight, shoulder width, and chest thickness. In addition, in order to facilitate subsequent analysis, when simulating a single personality environment, the OP is uniformly set to male; when simulating a mixed personality environment, the ratio of male to female is set to 1:1 to randomly generate pedestrians of different genders.

[0217] Then, the decision region and action region are added to each pedestrian agent to identify the global and local density, respectively, as Fig. 9 As shown. This application sets the radius of the global density range to 5 m. In addition, in order to ensure the clarity and simplicity of the simulation results, the two sectors are hidden during the simulation and are only used to make logical judgments.

[0218] In addition, the orbit, follow, and push action elements are modeled by using the social force parameter setting module, and the rotation action element is modeled by using the state diagram module. The modeling of the first three action elements is relatively simple, mainly achieved through parameter settings, which will not be repeated here. The rotation action element involves changes in body posture, which is more complicated. Specifically, first add the rotation state and rotation conditions to the agents of Ag, I, and As. After that, set the conditions for the rotation angle and link the angle to the body posture. There are two key settings:

[0219] (a) Based on The change of walking speed during the rotation behavior is added, while the process of gradual speed change is ignored. When the body posture changes, the walking speed is reduced to the appropriate speed after the rotation, and this speed will continue until the rotation behavior is completed. And after the body posture returns to the original walking state, the pedestrian will try to accelerate to the desired speed according to the constraints of the surrounding environment, which is the same as the original SFM; (b) The decision process based on energy consumption is added to the conditions of the rotation behavior, that is, the pedestrian will only perform the rotation and overtaking behavior when the current environment meets the original rotation conditions and the energy consumption of the rotation behavior is minimized.

[0220] Finally, the state diagram module is used to establish the motion logic of all agents according to the AMMDM framework as follows: (a) Determine whether the global density is in a free flow or a crowded flow. If this condition is true, the state will not change and the speed and position of the OP will be updated. (b) For pedestrians that are able to perform overtaking behavior, an evaluation is made as to whether pedestrians with different personalities comply with the overtaking behavior based on the local density, and those pedestrians who comply will select the best direction with the highest score and simulate to the next time step. (c) For pedestrians that do not obey, the energy consumption of the action element is calculated according to the personality classification, and then the execution with the minimum energy consumption is selected to update the body posture, walking speed, and position of the next time step. According to real experimental results, the following can be obtained respectively. Fig.10 Simulation parameters in the calibration table shown.

[0221] In order to verify the authenticity and effectiveness of the proposed simulation model, a simulation environment was constructed based on the above-mentioned overtaking experiment (6m×20m channel) and the overtaking behavior of a single-personality pedestrian (Case A) and the walking behavior of a mixed-personality pedestrian (Case B) were simulated. The specific settings are as follows:

[0222] Case A: If Fig.11 As shown, an OP is sequentially generated from five agents Ag, I, As, T, and S at the midpoint of the left edge of the channel, and a crowd is generated from agent N at a random position on the waiting line. The number of OPs is 1, and the number of crowds is set from 0.25 p / m2 to 2 p / m2 according to the global density, increasing by 0.25 p / m2, that is, the number of crowds is 7, 13, 20, 26, 33, 39, 46, and 52, respectively.

[0223] Case B: In the same simulation environment, 5 types of pedestrians are generated according to the proportion of the questionnaire survey, where pedestrians with different personalities are randomly generated from the left edge of the channel. At the same time, there are 20 pedestrians generated from agent N in the channel. All pedestrians move from the left edge to the right edge of the channel and then re-enter the channel from the left edge to ensure a stable pedestrian flow. In addition, a 6m×4m measurement area can be set in the middle of the channel to measure the speed and density of the pedestrian flow. The overall density of the channel is controlled by adjusting the total number of heterogeneous pedestrians generated, and the total number of heterogeneous pedestrians in this application ranges from 20 to 300.

[0224] 4. Qualitative results of simulation experiments.

[0225] Here, the above two situations are simulated using a simulation model based on the AMMDM framework and the original SFM respectively. The model can effectively and realistically simulate the overtaking behavior of pedestrians with heterogeneity, and successfully simulates the overtaking behavior of different action elements. In contrast, the simulation results of the model of the technical solution of the present application are closer to the real situation, and effectively reduce the drawbacks of pedestrian overlap, unreasonable collision, etc. in the original SFM. Therefore, from a qualitative point of view, it is believed that the model of the technical solution of the present application is more in line with reality.

[0226] 5. Quantitative results of simulation experiments.

[0227] Case A Analysis

[0228] Here, a set of OP trajectories are randomly extracted from 20 simulation results (800 simulated videos), such as Fig.12 It can be seen intuitively that the model of the technical solution of the present application can realistically simulate the overtaking behavior of pedestrians with different personalities at different densities.

[0229] In addition, it is also possible to quantify the frequency with which pedestrians with different personalities use various action elements to achieve overtaking behaviors, e.g. Fig.13 shown. Fig.13 The frequency of achieving transcendental behavior for each action element under the aggressive, impulsive, confident, nervous and shy types is included. Fig.13 Each group of data corresponds to the frequency of overtaking under turning, detouring and following. For example, the 0.5 data of the aggressive type includes the frequency of overtaking under turning and detouring, and the 1.5 data of the aggressive type includes the frequency of overtaking under turning, detouring and following. Therefore, for the bar graph corresponding to each group of data, the first bar graph represents the frequency of overtaking under turning, the second bar graph represents the frequency of overtaking under detouring, and the third bar graph represents the frequency of overtaking under following. If the third bar graph is empty, it means that there is no corresponding data.

[0230] When the global density is low, pedestrians are more inclined to choose detour action elements to overtake; as the global density increases, pedestrians need to detour a longer distance, resulting in increased energy consumption, and pedestrians begin to use rotation action elements and detour action elements to complete the overtaking; when the density increases further, it is difficult to completely complete the overtaking behavior after rotation or detour, so the number of follow-up action elements gradually increases. This conclusion is consistent with the actual situation. At the same time, the simulation results show the same trend as the real data collected in the experiment. Therefore, it is believed that after the introduction of action elements, the model of the technical solution of this application can more realistically simulate the fine-grained overtaking behavior of heterogeneous pedestrians.

[0231] However, in all simulation results, the occurrence rate of the push action is 0, which is not in line with expectations. Therefore, it is speculated that the reason for this is that the push action requires too much energy compared to other behaviors, and it is challenging to use it in simulation unless the OP has a considerable difference in walking speed from the crowd.

[0232] To verify the conjecture, more than 20 simulations were performed under the condition that the expected speed of Ag pedestrians was set to a fixed value (increased by 0.1m / s from 1.3m / s to 1.7m / s) while other parameters were kept constant. Obviously, when the speed increased to 1.6m / s, the push action element began to appear, and as the density increased, the pedestrian began to perform continuous push actions, followed by the rotation action element. At the same time, as the expected speed increased, the energy consumption of the rotation action element would be greater than that of the push action element, so the rotation action element was replaced as the push action element appeared, which is consistent with the analysis of the push action element in the real experiment. This result proves that the AMMDM framework can simulate push actions, although their implementation conditions are quite strict. At the same time, it also proves that the model is reasonable and the push action element should not be used frequently.

[0233] Therefore, the simulation results of Case A prove that the proposed model can enhance the realism of simulation while ensuring the rationality of the results.

[0234] Case B Analysis

[0235] By the measurement method proposed in the related art, the average speed and density of pedestrians in the time interval are obtained, and the result of the relationship between speed and density is shown in Figure 14 (a) (see asterisk). The straight line is the fitting curve of the result (R2 = 0.96597). In addition, some basic graphs of the one-way pedestrian flow previously studied are also shown in Figure 14 (a). Compared with these, it is obvious that the results of the technical solution of the present disclosure have the same trend as the previous research, which proves that the model of the technical solution of the present disclosure can effectively simulate the walking behavior of pedestrians in the one-way flow.

[0236] In addition, the movements of 300 heterogeneous pedestrians (generated according to the questionnaire ratio of 5 types of agents, Ag, I, As, T and S) and 300 homogeneous pedestrians (all generated by N agents) can be simulated for 20 hours, and the pedestrian flow per hour can be extracted, as shown in Figure 14 (b). Obviously, the channel capacity simulated by the AMMDM-based model is significantly higher than the original simulation result. At the same time, the width of the channel is 6 m, so the channel capacity in the two simulations is 4990 p / (h·m) and 3172 p / (h·m), respectively. According to the current "Metro Design Code" (GB 50157-2013) of my country, the design capacity of a one-way channel is 5000 p / (h·m). Therefore, the model of the technical solution disclosed in the present invention can more realistically simulate the pedestrian flow in a one-way channel.

[0237] Those skilled in the art will appreciate that, in the above method of specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of the steps should be determined by their functions and possible internal logic.

[0238] Based on the same inventive concept, an action-element-based overtaking behavior simulation device corresponding to the action-element-based overtaking behavior simulation method is also provided in the embodiment of the present disclosure. Since the principle of solving the problem by the device in the embodiment of the present disclosure is similar to the above-mentioned action-element-based overtaking behavior simulation method in the embodiment of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0239] Reference Fig.15 As shown, it is a schematic diagram of a device for simulating overtaking behavior based on action elements provided by an embodiment of the present disclosure, the device comprising: a first determining unit 10, a judging unit 20, and a second determining unit 30; wherein,

[0240] A first determination unit is used to determine a target personality parameter of a target pedestrian based on an association relationship between a pre-constructed trait factor and a personality parameter of a surpassing pedestrian, and to determine at least one action meta-model matching the target pedestrian based on a mapping relationship between the personality parameter and a pre-constructed action meta-model;

[0241] A judging unit, used for judging whether the current walking environment of the target pedestrian meets the overtaking condition during the movement of the target pedestrian;

[0242] The second determination unit is used to determine a target action metamodel that meets the energy consumption requirements in the at least one action metamodel when it is determined that the current walking environment does not meet the surpassing condition, and determine the action trajectory of the target pedestrian through the target action metamodel and the social force model.

[0243] In a possible implementation, the device is further used to: determine the regional utility scores of each action area of ​​the target pedestrian when it is determined that the current walking environment satisfies the surpassing condition; and determine the action trajectory of the target pedestrian based on the target action area corresponding to the maximum regional utility score when it is determined that the target pedestrian meets the macro decision-making conditions.

[0244] In a possible implementation, the device is further used to: determine the regional score of the target pedestrian in each of the action areas based on the personality parameters of the target pedestrian; wherein the regional score is the position and speed score of the target behavior in the action area; and determine the regional score as the regional utility score of the corresponding action area.

[0245] In a possible implementation, the device is further used to: input the area information of the target action area into a social force model for processing; perform optimal solution on the social force model, and determine the position and speed of the target pedestrian at the next moment based on the optimal solution result.

[0246] In a possible implementation, the second determination unit is further used to: calculate the total energy consumption of the target pedestrian performing the overtaking behavior according to the actions indicated by each of the action metamodels; and determine that the action metamodel corresponding to the minimum total energy consumption is the target action metamodel that meets the energy consumption requirements.

[0247] In a possible implementation, the second determination unit is further used to: input the model information of the target action meta-model into the social force model for processing; perform optimal solution on the social force model, and determine the position and speed of the target pedestrian at the next moment according to the optimal solution result.

[0248] In a possible implementation, the social force model includes:

[0249]

[0250] in, is the mass of pedestrian i, is the actual speed of pedestrian i, is the actual speed of pedestrian j, is the expected speed at time t, is the unit vector pointing to the desired target at time t, τ is the characteristic time, , are constants, representing the intensity and range of social interaction forces, , is the sum of the radii of pedestrian i and pedestrian j and the distance between their centers of mass, is the distance between pedestrian i and the center of mass of wall w, , are the normalized vectors from pedestrian j and wall w to pedestrian i, , are the body compression coefficient and sliding friction coefficient, , is the tangent direction, J and W represent the number of pedestrians and walls, and g(x)=max(x, 0) indicates whether pedestrians are not touching each other or touching the wall; and are the detour coefficient and anisotropy coefficient respectively; is the normalized distance vector The direction of pedestrian i The angle between and are the correction coefficients to avoid unreasonable huge forces between pedestrians and between pedestrians and close obstacles; H, , ,and is the correction parameter of the following action element, is the specified constant; is the strength of the social interaction force that pushes the action element; and They represent the torque and turning intention of pedestrian i, respectively. Indicates driving force, represents the interaction force between pedestrian i and pedestrian j, represents the interaction force between pedestrian i and the wall; and are the global density and the regional utility score calculated from the local density, respectively; For pedestrians at a speed of The minimum energy consumption when represents the angular acceleration of pedestrian i at time t, represents the actual position of pedestrian i at time t, and are the shoulder width and chest thickness of pedestrian i, θ∈[-π / 2,π / 2] is the turning angle, is the turning time of pedestrian i, represents the friction between pedestrian i and the ground, represents the congestion flow threshold, Indicates the free flow threshold.

[0251] In a possible implementation manner, the device is further used to: construct a mathematical model of at least one action element of overtaking a pedestrian to obtain an action element model; wherein the at least one action element includes at least one of the following: bypassing, following, pushing, and turning, and the action element model of turning includes:

[0252]

[0253] in, is the mass of pedestrian i, is the actual speed, J and W represent the number of pedestrians and walls respectively, is the expected speed at time t, is the unit vector pointing to the desired target at time t, τ is the characteristic time, Indicates driving force, represents the interaction force between pedestrian i and pedestrian j, represents the interaction force between pedestrian i and the wall, and They represent the torque and turning intention of pedestrian i, respectively, α, β and γ are the influencing factors of each item, is the interference factor, a and b (a>b) are constants, A is the constant term, represents the angular acceleration of pedestrian i at time t.

[0254] For descriptions of the processing flow of each module in the device and the interaction flow between each module, reference may be made to the relevant descriptions in the above method embodiment, which will not be described in detail here.

[0255] Corresponds to Figure 1 The embodiment of the present disclosure further provides an electronic device 1600, such as Fig.16 FIG. 1 is a schematic diagram of the structure of an electronic device 1600 provided in an embodiment of the present disclosure, including:

[0256] Processor 161, memory 162, and bus 163; memory 162 is used to store execution instructions, including internal memory 1621 and external memory 1622; the internal memory 1621 is also called internal memory, which is used to temporarily store the operation data in the processor 161 and the data exchanged with the external memory 1622 such as a hard disk. The processor 161 exchanges data with the external memory 1622 through the internal memory 1621. When the electronic device 1600 is running, the processor 161 communicates with the memory 162 through the bus 163, so that the processor 161 executes the following instructions:

[0257] Determine a target personality parameter of a target pedestrian based on a correlation relationship between a pre-constructed trait factor and a personality parameter of a surpassing pedestrian, and determine at least one action meta-model matching the target pedestrian based on a mapping relationship between the personality parameter and a pre-constructed action meta-model;

[0258] During the movement of the target pedestrian, determining whether the current walking environment of the target pedestrian meets the overtaking condition;

[0259] When it is determined that the current walking environment does not satisfy the surpassing condition, a target action metamodel that satisfies the energy consumption requirement is determined in the at least one action metamodel, and the action trajectory of the target pedestrian is determined through the target action metamodel and the social force model.

[0260] The embodiment of the present disclosure also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method for simulating the overtaking behavior based on action elements described in the above method embodiment are executed. The storage medium can be a volatile or non-volatile computer-readable storage medium. The embodiment of the present disclosure also provides a computer program product, which carries a program code, and the instructions included in the program code can be used to execute the steps of the method for simulating the overtaking behavior based on action elements described in the above method embodiment. For details, please refer to the above method embodiment, which will not be repeated here.

[0261] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is implemented as a computer storage medium. In another optional embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).

[0262] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0263] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, each functional unit in each embodiment of the present disclosure may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0264] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present disclosure, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0265] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present disclosure, which are used to illustrate the technical solutions of the present disclosure, rather than to limit them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed in the present disclosure, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A method for simulating transcending behavior based on action elements, characterized in that: include: Based on the association relationship between the pre-constructed trait factors and the personality parameters of the surpassing pedestrian, the target personality parameters of the target pedestrian are determined, and based on the mapping relationship between the personality parameters and the pre-constructed action meta-models, at least one action meta-model matching the target pedestrian is determined; wherein the mapping relationship between the personality parameters and the pre-constructed action meta-models includes: , , , and , OA AG Indicates the enterprising personality in the personality parameters, OA I Indicates the impulsive personality in the personality parameters, OA AS Indicates the self-confident personality in the personality parameters, OA S Indicates the shy personality in the personality parameters, OA T represents the nervous personality in the personality parameter, P represents the turning action element in the action element, R represents the pushing action element in the action element, D represents the detour action element in the action element, and F represents the following action element in the action element; During the movement of the target pedestrian, determining whether the current walking environment of the target pedestrian meets the overtaking condition; When it is determined that the current walking environment does not satisfy the surpassing condition, determining a target action metamodel that meets the energy consumption requirement in at least one action metamodel matching the target pedestrian, and determining the action trajectory of the target pedestrian through the target action metamodel and the social force model; The step of determining a target action metamodel that meets the energy consumption requirement from at least one action metamodel matching the target pedestrian includes: Calculating the total energy consumption of the target pedestrian performing the overtaking behavior according to the actions indicated by each action meta-model matching the target pedestrian; Determining, in at least one action metamodel matching the target pedestrian, that an action metamodel corresponding to a minimum total energy consumption is the target action metamodel that meets the energy consumption requirement; Determining the target pedestrian's action trajectory by using the target action meta-model and the social force model includes: inputting the speed of the target action meta-model into the social force model for processing; The social force model is optimally solved to obtain the position and speed of the target pedestrian at the next moment, and the movement trajectory of the target pedestrian is determined according to the position and speed of the target pedestrian at the next moment.

2. The method according to claim 1, characterized in that The method further comprises: In the case where it is determined that the current walking environment satisfies the surpassing condition, determining a regional utility score of each action area of ​​the target pedestrian; When it is determined that the target pedestrian meets the macro decision-making conditions, the movement trajectory of the target pedestrian is determined according to the target action area corresponding to the maximum regional utility score.

3. The method according to claim 2, characterized in that The determining of the regional utility score of each action area of ​​the target pedestrian includes: Determining a regional score of the target pedestrian in each of the action areas according to the personality parameters of the target pedestrian; wherein the regional score is a position and speed score of the target behavior in the action area; The region score is determined as a region utility score corresponding to the action region.

4. The method according to claim 1, characterized in that Determining the target pedestrian's action trajectory by using the target action meta-model and the social force model includes: Inputting model information of the target action meta-model into a social force model for processing; The social force model is optimally solved, and the position and speed of the target pedestrian at the next moment are determined according to the optimal solution result.

5. The method according to claim 1, characterized in that The social force model includes: in, is the mass of pedestrian i, represents the actual speed of pedestrian i at time t, represents the actual speed of pedestrian j at time t, is the expected speed of pedestrian i at time t, is the unit vector pointing to the desired target by pedestrian i at time t, τ is the characteristic time, , are constants, representing the intensity and range of social interaction forces, , are the sum of the sight radius of pedestrian i and pedestrian j and the distance between their centers of mass, is the distance between pedestrian i and the center of mass of wall w, represents the normalized vector from pedestrian j to pedestrian i, is the normalized vector from wall w to pedestrian i, , are the body compression coefficient and sliding friction coefficient, represents the tangent direction between pedestrians i and j, represents the tangent direction between pedestrian i and wall w, J and W represent the number of pedestrians and walls respectively, and g(x)=max(x, 0) indicates whether pedestrians are not touching each other or touching the wall; and are the detour coefficient and anisotropy coefficient respectively; is the normalized distance vector The direction of pedestrian i The angle between and are the correction coefficients to avoid unreasonable huge forces between pedestrians i and j and between pedestrian i and close-range obstacle w; H, , ,and is the correction parameter of the following action element, is the specified constant; is the strength of the social interaction force that pushes the action element; and They represent the torque and turning intention of pedestrian i, respectively. represents the driving force of pedestrian i, represents the interaction force between pedestrian i and pedestrian j, represents the interaction force between pedestrian i and the wall; and are the global density and the regional utility score calculated from the local density, respectively; For pedestrians at a speed of The minimum energy consumption when represents the angular acceleration of pedestrian i at time t, represents the actual position of pedestrian i at time t, and are the shoulder width and chest thickness of pedestrian i, ∈[-π / 2,π / 2] is the turning angle of the i-th pedestrian, is the turning time of pedestrian i, represents the friction between pedestrian i and the ground, represents the congestion flow threshold, represents the free flow threshold, C i represents the compliance parameter of pedestrian i.

6. The method according to claim 1, characterized in that The method further comprises: A mathematical model of at least one action element of overtaking a pedestrian is constructed to obtain an action element model; wherein the at least one action element includes at least one of the following: bypassing, following, pushing, and turning, and the action element model of turning includes: in, is the mass of pedestrian i, is the actual speed of pedestrian i at time t, J and W represent the number of pedestrians and walls respectively, is the expected speed of pedestrian i at time t, is the unit vector pointing to the desired target by pedestrian i at time t, τ is the characteristic time, represents the driving force of pedestrian i, represents the interaction force between pedestrian i and pedestrian j, represents the interaction force between pedestrian i and wall w, and They represent the torque and turning intention of pedestrian i, respectively, α, β and γ are the influencing factors of each item, is the interference factor, a and b are constants, a > b, A is a constant term, represents the angular acceleration of pedestrian i at time t.

7. A transcending behavior simulation device based on action elements, characterized in that: include: The first determination unit is configured to determine a target personality parameter of a target pedestrian based on an association relationship between a pre-constructed trait factor and a personality parameter of a surpassing pedestrian, and to determine at least one action meta-model matching the target pedestrian based on a mapping relationship between the personality parameter and a pre-constructed action meta-model; wherein the mapping relationship between the personality parameter and the pre-constructed action meta-model includes: , , , and , OA AG Indicates the enterprising personality in the personality parameters, OA I Indicates the impulsive personality in the personality parameters, OA AS Indicates the self-confident personality in the personality parameters, OA S Indicates the shy personality in the personality parameters, OA T represents the nervous personality in the personality parameter, P represents the turning action element in the action element, R represents the pushing action element in the action element, D represents the detour action element in the action element, and F represents the following action element in the action element; A judging unit, used for judging whether the current walking environment of the target pedestrian meets the overtaking condition during the movement of the target pedestrian; A second determination unit is used to determine a target action metamodel that meets the energy consumption requirement in at least one action metamodel matched with the target pedestrian when it is determined that the current walking environment does not meet the surpassing condition, and determine the action trajectory of the target pedestrian through the target action metamodel and the social force model; The second determining unit is further configured to: Calculating the total energy consumption of the target pedestrian performing the overtaking behavior according to the actions indicated by each action meta-model matching the target pedestrian; Determining, in at least one action metamodel matching the target pedestrian, that an action metamodel corresponding to a minimum total energy consumption is the target action metamodel that meets the energy consumption requirement; inputting the speed of the target action meta-model into the social force model for processing; The social force model is optimally solved to obtain the position and speed of the target pedestrian at the next moment, and the movement trajectory of the target pedestrian is determined according to the position and speed of the target pedestrian at the next moment.

8. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the action-element-based transcending behavior simulation method as described in any one of claims 1 to 6 are performed.

Images