Robot-assisted evacuation optimal intervention method and system based on crowd disorder degree

Abstract

The application discloses a robot-assisted evacuation optimal intervention method and system based on crowd chaos degree, and the method comprises the following steps: constructing a crowd network model, and quantifying the chaos degree of crowd movement through the crowd chaos degree; modeling the evolution process of crowd chaos by using a random differential equation and a marked time point process method, and constructing a crowd chaos degree evolution model; based on the crowd chaos degree evolution model, planning robot-assisted intervention as a random optimal control problem, and obtaining the optimal intervention rate of the robot by solving, and then determining the optimal intervention individual and intervention time. The application quantifies the chaos degree of crowd movement by introducing the crowd chaos degree, models the evolution process of crowd chaos, describes the robot-assisted intervention as a random optimal control problem, determines the optimal intervention individual and intervention time by solving the problem, and thus realizes the optimal intervention of the robot.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crowd evacuation intervention, and in particular to a robot-assisted evacuation optimal intervention method and system based on crowd chaos degree. BACKGROUND

[0002] With the continuous development of urbanization, crowd accidents are increasing, causing considerable casualties and economic losses. In crowd accidents, people exhibit complex individual and group behaviors, such as panic, pushing and stampede, and the interaction of these behaviors can increase the chaos degree of crowd movement, causing secondary casualties in crowd evacuation. Therefore, studying crowd evacuation regulation methods is of great significance to reduce or even avoid crowd chaos and save evacuation time.

[0003] In recent years, the rapid rise of intelligent robot technology has stimulated the development of robot-assisted applications. Robot-assisted crowd evacuation has become one of the popular applications due to its labor-saving, strong risk adaptability, easy deployment and other advantages. Researches on evacuation navigation, pedestrian flow control and evacuation route discovery have been widely studied. In the evacuation navigation method, mobile robots guide evacuees to less crowded exits, and pedestrian flow control provides activity methods for evacuees through sound, visual signals or movement of robots to avoid congestion near exits. In the evacuation route discovery, robots explore the area where the emergency occurs and find the best evacuation path from the accident site to the emergency exit through communication with pre-deployed sensors.

[0004] However, the existing methods of robot-assisted crowd evacuation do not consider the impact of crowd chaos on crowd evacuation, which is crucial for effective evacuation. When a crowd accident occurs, people try to escape from the dangerous place as soon as possible, which may lead to chaotic behavior or even stampede. Generally speaking, the more chaotic the crowd is, the slower the evacuation is. Therefore, considering the impact of crowd chaos on evacuation can provide new strategies for robot intervention during crowd evacuation. SUMMARY

[0005] To solve the above problems of the prior art, the present application provides a robot-assisted evacuation optimal intervention method and system based on crowd chaos degree, which quantifies the chaos degree of crowd movement by introducing crowd chaos degree, models the evolution process of crowd chaos, describes robot-assisted intervention as a stochastic optimal control problem, determines the optimal intervention individual and intervention time by solving the problem, and thus the robot identifies the most important individual for intervention instead of controlling the entire irrational crowd, achieving optimal intervention of the robot.

[0006] In a first aspect, the present disclosure provides a robot-assisted evacuation optimal intervention method based on crowd chaos degree:

[0007] An optimal intervention method for robot-assisted evacuation based on crowd disorder includes the following steps:

[0008] Construct a crowd network model and quantify the degree of disorder in crowd movement by measuring crowd disorder.

[0009] The evolution of crowd disorder is modeled using stochastic differential equations and labeled time-point processes, thus constructing a crowd disorder evolution model.

[0010] Based on the evolutionary model of crowd disorder, robot-assisted intervention is planned as a stochastic optimal control problem. By solving the problem, the optimal intervention rate of the robot can be obtained, and then the optimal intervention individual and intervention time can be determined.

[0011] A further technical solution, namely the construction of a crowd network model, which measures the degree of disorder in crowd movement through crowd disorder metrics, specifically includes:

[0012] Constructing a population network model using undirected graphs;

[0013] Based on a population network model, the local disorder degree between adjacent individuals is constructed.

[0014] Based on the local disorder between adjacent individuals, construct the global disorder between any two individuals;

[0015] Based on the global disorder between any two individuals, construct the individual disorder of any individual.

[0016] A further technical solution is that the crowd disorder evolution model describes the dynamic transition process between ordered and disordered states, including a crowd disorder evolution model without intervention and a crowd disorder evolution model under intervention.

[0017] A further technical solution involves constructing the non-interventional population disorder evolution model, which includes:

[0018] Based on the degree of individual disorder at the initial moment, individuals are divided into disordered or ordered states.

[0019] Based on the initial state of individuals, this study uses a time-point marking process method combined with stochastic differential equations to simulate the state transition process of individuals within a set time period, thus constructing an intervention-free population disorder evolution model.

[0020] A further technical solution involves a time-point marking process, which refers to a counting process of the number of events occurring within a set time period. Using this time-point marking process, combined with stochastic differential equations, a non-interventional population disorder evolution model is constructed. The specific process includes:

[0021] Define the transition events between disordered and ordered states, namely the transition events from ordered to disordered states and the transition events from disordered to ordered states.

[0022] The counting process of the number of occurrences of two transition events within a set time period is described by intensity. Based on stochastic differential equations, the expected number of occurrences of two transition events within a set time period is calculated, and a non-interventional population disorder evolution model is constructed.

[0023] A further technical solution involves constructing the population disorder evolution model under intervention, including:

[0024] Based on the non-interventional population disorder evolution model, a state variable of robot intervention is introduced to construct a population disorder evolution model under intervention.

[0025] A further technical solution is that the stochastic optimal control problem refers to finding the robot intervention rate that maximizes the expected value of the cumulative intervention benefit within a set time period.

[0026] Secondly, this disclosure provides a robot-assisted evacuation optimal intervention system based on crowd disorder, including:

[0027] The data processing module is used to build a crowd network model and quantify the degree of disorder in crowd movement by measuring crowd disorder.

[0028] The model building module is used to model the evolution of crowd disorder using stochastic differential equations and marked time point processes, and to build a crowd disorder evolution model.

[0029] The optimization solution module is used to plan robot-assisted intervention as a stochastic optimal control problem based on the population disorder evolution model. By solving the problem, the optimal intervention rate of the robot can be obtained, and then the optimal intervention individual and intervention time can be determined.

[0030] Thirdly, this disclosure also provides an electronic device, including a memory and a processor, and computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the steps of the method described in the first aspect.

[0031] Fourthly, this disclosure also provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, perform the steps of the method described in the first aspect.

[0032] The above one or more technical solutions have the following beneficial effects:

[0033] 1. This disclosure provides a robot-assisted stochastic optimal intervention method and system for crowd evacuation, which takes into account the impact of crowd chaos on evacuation, provides a new strategy for robot intervention in the crowd evacuation process, effectively determines the individuals involved in the chaos and the intervention time, and realizes targeted robot intervention.

[0034] 2. This disclosure provides a method and system for stochastic optimal intervention in crowd evacuation assisted by a robot. By introducing crowd disorder degree to quantify the degree of disorder in crowd movement and modeling the evolution process of crowd disorder, the robot-assisted intervention is described as a stochastic optimal control problem. By solving this problem, the optimal intervention individual and intervention time are determined. In this way, the robot identifies the most important individual for intervention, rather than controlling the entire irrational crowd, thus achieving optimal robot intervention. Attached Figure Description

[0035] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0036] Figure 1 This is an overall flowchart of the method described in Embodiment 1 of the present invention. Detailed Implementation

[0037] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0038] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0039] Evacuation interventions can prevent accidents and provide safety guarantees. Robotic intervention is a current research hotspot, but how to provide effective robotic intervention schemes based on the real-time situation of crowd evacuation in emergency situations is a highly challenging problem. However, existing studies have not considered the impact of crowd disorder, which is a fundamental factor determining the efficiency of crowd evacuation. This invention proposes a stochastic optimal intervention method based on crowd disorder, providing an optimal strategy for robot-assisted crowd evacuation. First, crowd disorder is introduced to quantify the degree of disorder in crowd movement; then, the evolutionary model of the crowd disorder state is studied using stochastic differential equations and labeled time-point processes; subsequently, robot-assisted intervention is described as a stochastic optimal control problem, seeking to maximize the intervention reward; finally, by solving this problem, an analytical solution to the stochastic optimal control problem can be obtained.

[0040] Example 1

[0041] This embodiment provides an optimal intervention method for robot-assisted evacuation based on crowd disorder, such as... Figure 1 As shown, it includes the following steps:

[0042] Construct a crowd network model and quantify the degree of disorder in crowd movement by measuring crowd disorder.

[0043] The evolution of crowd disorder is modeled using stochastic differential equations and labeled time-point processes, thus constructing a crowd disorder evolution model.

[0044] Based on the evolutionary model of crowd disorder, robot-assisted intervention is planned as a stochastic optimal control problem. By solving the problem, the optimal intervention rate of the robot can be obtained, and then the optimal intervention individual and intervention time can be determined.

[0045] First, a crowd network model is constructed using an undirected graph. The crowd network is modeled using an undirected graph G(V,E), where V represents the set of n individuals in the crowd, and E represents the set of edges between individuals. The crowd network consists of the positions and speeds of individuals captured in the surveillance video. An edge exists between a pair of individuals only when the geometric distance between them is less than a fixed range r. Two individuals with an edge are adjacent and are called neighbors.

[0046] Building upon this, the degree of disorder in crowd movement is quantified through crowd disorder measurement. Since individual behavior is often influenced by others rather than acting independently, it is difficult to accurately measure crowd disorder. Therefore, this embodiment introduces the definitions of local disorder and global disorder to accurately quantify crowd disorder.

[0047] The local disorder ε(i,j) represents the degree of disorder between adjacent individuals i and j, and is defined as follows:

[0048]

[0049] Where θ(i,j)∈[0,π] is the angle between the movement directions of individuals i and j, and the value of ε(i,j) increases monotonically with the increase of the degree of disorder. Specifically, when the angle θ(i,j) is small, the movement directions of the two individuals are more consistent and the local disorder is close to 0. Conversely, when the angle θ(i,j) is large, the local disorder is close to 1.

[0050] An individual's disorder is influenced not only by neighboring individuals but also by other distant individuals. Non-adjacent individuals can influence each other through the propagation of disorder. The disorder between any two individuals... Define the global disorder between individual i and individual j. The global disorder can be calculated through initialization steps and iterative processes:

[0051] In the initialization step, a symmetric matrix A∈R is first constructed. n×n Matrix A is an n x n matrix consisting of n x n numbers, where A ij =ε(i,j) represents the number in the i-th row and j-th column of matrix A, A ii = 0, to avoid self-reinforcing effects; then, matrix A is normalized to ensure the convergence of subsequent iterations, the normalized matrix is:

[0052]

[0053] Where D is the degree matrix, D = diag{D1, D2, ..., D} n}=diag{Σ j A 1j ,Σ j A 2j ,…,Σ j A nj}

[0054] During the iteration process, the propagation of disorder over time through the population network G(ν,ε) is calculated. Time is divided into multiple steps, denoted by t. Y(t) represents the disorder influence matrix at time t, which represents the disorder influence of the t-th order neighborhood. Initially, Y(0) = βB represents the influence of the first-order neighborhood, where β is the weight parameter. Similarly, BY(0) represents the influence of the second-order neighborhood, and so on. As the disorder of an individual propagates to its neighborhood, Y(t) can be calculated step by step using the following equation until convergence:

[0055] Y(t+1)=βBY(t)+(1-β)Y(0)

[0056] In the above equation, the chaotic effect between nodes i and j consists of two components: the effect from their higher-order neighbors (the first term) and the effect from their first-order neighborhood (the initial chaotic degree, the second term). Furthermore, since B is a symmetric matrix, the effect propagates symmetrically throughout the network. Through iteration, the chaos eventually propagates to the entire network, resulting in a convergent chaotic degree matrix:

[0057]

[0058] In the above formula, matrix I is the identity matrix, that is, a matrix in which all values ​​on the main diagonal are 1 and all others are 0.

[0059] Because global pairwise chaos accumulates the effects of all other individual chaos through topological relationships, it more accurately represents pairwise chaos.

[0060] Next, we define the disorder for each individual and compute its upper bound to determine the individual state for further processing. The individual disorder ε for any individual i is... g (i) is defined as the accumulation of global disorder between i and the remaining nodes:

[0061] ε g (i)=[Y * 1] i

[0062] In the above formula, 1 is a vector in which all elements are equal to 1.

[0063] After quantifying the degree of disorder among individuals in a crowd through crowd chaos measurement, and considering that crowd chaos dynamically evolves during crowd evacuation, it is impractical to track the chaos of crowd movement using surveillance video due to limitations in camera coverage and occlusion issues. Each individual in the crowd exhibits two distinct states: chaos and order, with one state dynamically transitioning to the other. Therefore, in this embodiment, a chaos evolution model is constructed to theoretically describe the basic dynamics of crowd chaos. In this model, two states are used to represent chaos, reducing its evolution to dynamic state transitions.

[0064] First, a non-interventional model of crowd disorder evolution is constructed. As the crowd moves, disorder may evolve spontaneously; for example, an ordered state may transform into a disordered state due to the influence of other disorderly movements, and a disordered state may re-enter an ordered state through self-ordering. In the crowd disorder evolution model, the states of individuals are first defined, and then the dynamics of state transitions are studied.

[0065] At time t, the state of individual i is determined by Z. i (t) represents Z, where Z i(t) = 1 / 0 represents a disordered / ordered state. Using a state binarization method, individuals are divided into two states, disordered and ordered, based on the global disorder level.

[0066] The state of each node at the initial time t0 is as follows:

[0067]

[0068] Where 'a' is the discount parameter, and ε* is the threshold for determining whether the crowd is chaotic or not.

[0069] An individual's state may change over time. Given a small time interval dt, i.e., a time period, the evolution of state transitions is described by the following formula:

[0070] Z i (t+dt)=Z i (t)+dZ i (t)

[0071] In the formula, dZ i (t)∈{0,1,-1} is Z i The derivative of (t) is 0, where 0 indicates that the state remains unchanged, 1 indicates that the state changes from ordered to disordered, and -1 indicates that the state changes from disordered to ordered.

[0072] Due to the differential dZ i The state transition (t) depends on the number of state transitions occurring during time dt. Therefore, this embodiment proposes a method based on a marked time-point process to simulate the state transition process. The marked time-point process consists of a series of discrete events located in time, typically represented by a counting process that counts the number of events occurring before time t. The counting process is characterized by an intensity λ(t|H(t)), representing the event occurrence rate for a given history H(t), i.e., this intensity value equals the number of events occurring per unit time. For simplicity, λ(t|H(t)) is referred to as λ(t).

[0073] The method based on marked time points includes the following two steps. First, two types of state transition events are defined: 1) OC, the transition from an ordered state to a disordered state; 2) CO, the transition from a disordered state to an ordered state. When individual i transitions from an ordered state to a disordered state at time t, the event e(i,OC,t) occurs. Similarly, when individual i transitions from a disordered state to an ordered state at time t, the event e(i,CO,t) occurs.

[0074] Secondly, the strength is defined as and Two counting processes L i (t) and M i(t) is used to calculate the number of times events e(i,OC,t) and e(i,CO,t) occur for any individual i before t. To quantify the intensity, this embodiment uses η and μ to represent the disorder diffusion rate of the link and the disorder recovery rate of nodes in the population network G(ν,ε). Specifically, η represents the speed at which a node in a disordered state influences its neighborhood over time, and μ represents the speed at which a node in a disordered state spontaneously transitions to an ordered state. A node in an ordered state may transition to a disordered state under the influence of its neighborhood in a disordered state, i.e.:

[0075]

[0076] in, Z(t) represents the number of neighborhoods of a disordered state. n (t)) T , It is the i-th row of the adjacency matrix of a directed graph, and its result is multiplied by 1-Z. i (t), because the event only occurs when the nodes are in order.

[0077] Similarly, nodes in a disordered state transition to an ordered state at a rate of μ, that is:

[0078]

[0079] According to the definition of intensity, the expected number of occurrences of events e(i,OC,t) and e(i,CO,t) within dt can be obtained by the following formula:

[0080]

[0081]

[0082] Among them, E[dL i (t)] and E[dM i [(t)] respectively represent solving dL i (t) and dM i (t) Expected degree, differential dL i (t)=L i (t+dt)-L i (t) and dM i (t)=M i (t+dt)-M i (t) represents the number of OC / CO events occurring within dt, where dL i (t)∈{0,1},dM i (t)∈{0,1}. Considering dZ i(t) is determined by the difference in the number of OC and CO events; therefore, the dynamics of the state transition can be described by the following stochastic differential equation (SDE) with jumps:

[0083] dZ i (t)=dL i (t)-dM i (t)

[0084] Furthermore, the evolutionary model of population disorder under no-intervention conditions is specifically manifested as follows:

[0085] Z i (t+dt)=Z i (t)+dL i (t)-dM i (t)

[0086] Secondly, based on the non-interventional population disorder evolution model, a state variable of robot intervention is introduced to construct a population disorder evolution model under intervention.

[0087] Currently, intelligent robots are widely used in public places. Robots can help guide people to evacuate more orderly. Therefore, with the intervention of robots, individuals can transition from a chaotic state to an orderly state more quickly. In this embodiment, a new state variable is introduced to represent whether a node is subject to intervention, and then the state change process under robot intervention is studied.

[0088] Let W i (t) represents the state variable of the robot intervention, where W i (t) = 1 / 0 indicates whether individual i is under robot intervention. The state change process is described as follows:

[0089] W i (t+dt)=W i (t)+dW i (t)

[0090] Among them, dW i (t)∈{0,1,-1} is the differential of W, where 0 / 1 / -1 represent the state remaining unchanged, the state changing from non-intervention to intervention, and the state changing from intervention to non-intervention, respectively.

[0091] In order to derive the differential dW i (t) defines two new types of state transition events and a new counting process: 1) NI, a non-interventional state transitioning to an interventional state; 2) IN, an interventional state transitioning to a non-interventional state; 3) N i (t), intervention intensity is The counting process calculates the number of times event e(i,NI,t) occurs before t. Since event e(i,NI,t) only occurs when the state of i is chaotic (i.e., Z...), the counting process is repeated.i (t)=1) and no intervention (i.e., W) i It occurs when (t) = 0, therefore the intensity of the intervention can be described as:

[0092]

[0093] Where, ν i (t) represents the rate of robot intervention on individual i.

[0094] Next, the expected number of times event e(i,NI,t) occurs within the set time period, i.e., the time interval dt, can be calculated as follows:

[0095]

[0096] Wherein, differential dN i (t)=N i (t+dt)-N i (t) represents the number of times the event occurs during the time interval dt.

[0097] With the help of robot intervention, the state of a node can transition from chaos to order more quickly. Let δ be the increase in the recovery rate of a node under robot-assisted intervention, then the CO event intensity of individual i becomes:

[0098]

[0099] Event e(i,NI,t) occurs only when node i transitions from disorder to order with the help of robot intervention. In this case, event e(i,CO,t) also occurs. Therefore, the number of times event e(i,NI,t) occurs in dt is W. i (t)dM i (t).

[0100] Finally, the differential dW i (t) can be described by the following stochastic differential equation (SDE) with jumps:

[0101] dW i (t)=dN i (t)-W i (t)dM i (t)

[0102] Furthermore, the specific manifestation of the population disorder evolution model under intervention is as follows:

[0103] W i (t+dt)=W i (t)+dN i (t)-W i (t)dM i (t)

[0104] Finally, the crowd disorder evolution model provides a basis for how to regulate the degree of robot intervention: a higher intervention rate can reduce crowd disorder and the risk of stampedes, but at the same time incurs higher costs; conversely, a lower intervention rate can effectively save costs, but may lead to uncontrollable crowd disorder and increase the risk of stampedes.

[0105] To balance the above scenarios, this invention presents an optimal intervention strategy. Specifically, robot-assisted intervention is formulated as a stochastic optimal control problem, finding the optimal robot intervention rate that maximizes the expected value of the cumulative intervention benefit within the time window:

[0106]

[0107] Wherein, the expected value x Indicates from t0 to t f The counting process yields all possible outcomes, where u(t) is the payoff function that measures the return on robot intervention at time t:

[0108]

[0109] The intervention benefit term is described by the weighted reduction in the number of individuals in a state of confusion within the population from the initial time to a later time period, ω. i =ηA i* 1 represents the weight of node i (i.e., individual i), indicating that the more nodes in the neighborhood, the greater the contribution to reducing chaos; the intervention cost is related to the cost Ω of intervening in each individual. i and the corresponding robot intervention rate ν i (t) proportional; without loss of generality, the intervention cost for each individual is expressed as

[0110] Next, the optimal stochastic control problem is solved to obtain the time window t∈(t0,t... f The optimal intervention rate for each individual in the [], i.e., the n optimal intervention rates Λ(t)=(ν1(t),…,ν n (t)) T .

[0111] Due to the initial disorder Σ i ω i Z i (t0) is a constant that can be ignored when searching for the optimal solution. Therefore, the above optimization problem can be simplified to:

[0112]

[0113] stν i (t)≥0, i∈ν,t∈(t0,t)f ]

[0114] in,

[0115] Then, by solving the above problem, we obtain the closed-form optimal solution for the intervention rate during robot-assisted crowd evacuation, which is the optimal intervention rate for each individual.

[0116] Based on the above formula for calculating intervention intensity, the optimal intervention intensity for each individual is calculated using the obtained optimal intervention rate. Therefore, the total intensity of all intervention events can be calculated as follows:

[0117]

[0118] Next, the probability of each individual intervention event occurring is calculated. The individual that will be the subject of the intervention at the current moment is selected according to the roulette wheel mechanism, which is the optimal intervention individual.

[0119] This embodiment provides a robot-assisted stochastic optimal intervention method and system for crowd evacuation, which takes into account the impact of crowd chaos on evacuation, provides a new strategy for robot intervention during crowd evacuation, effectively determines the individuals involved in the chaos and the intervention time, and realizes targeted robot intervention.

[0120] Example 2

[0121] This embodiment provides a robot-assisted evacuation optimal intervention system based on crowd disorder, including:

[0122] The data processing module is used to build a crowd network model and quantify the degree of disorder in crowd movement by measuring crowd disorder.

[0123] The model building module is used to model the evolution of crowd disorder using stochastic differential equations and marked time point processes, and to build a crowd disorder evolution model.

[0124] The optimization solution module is used to plan robot-assisted intervention as a stochastic optimal control problem based on the population disorder evolution model. By solving the problem, the optimal intervention rate of the robot can be obtained, and then the optimal intervention individual and intervention time can be determined.

[0125] Example 3

[0126] This embodiment provides an electronic device, including a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the computer instructions are executed by the processor, they complete the steps in the optimal intervention method for robot-assisted evacuation based on crowd disorder as described above.

[0127] Example 4

[0128] This embodiment also provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps in the robot-assisted evacuation optimal intervention method based on crowd disorder as described above.

[0129] The steps and methods involved in Embodiments 2 to 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0130] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0131] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0132] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A robot-assisted evacuation optimal intervention method based on crowd disorder, characterized in that, Includes the following steps: Construct a crowd network model to quantify the degree of disorder in crowd movement through crowd disorder measurement, including: The specific steps for quantifying the degree of disorder in crowd movement through crowd disorder measurement include: Constructing a population network model using undirected graphs; Based on a population network model, the local disorder degree between adjacent individuals is constructed. Based on the local disorder between adjacent individuals, construct the global disorder between any two individuals; Based on the global disorder between any two individuals, construct the individual disorder of any individual; The evolution of crowd disorder is modeled using stochastic differential equations and labeled time-point processes, thus constructing a crowd disorder evolution model. The crowd disorder evolution model describes the dynamic transition process between ordered and disordered states, including a crowd disorder evolution model without intervention and a crowd disorder evolution model under intervention. The construction process of the non-interventional population disorder evolution model includes: Based on the degree of individual disorder at the initial moment, individuals are divided into disordered or ordered states. Based on the individual's state at the initial moment, the method of marking time points and combining stochastic differential equations is used to simulate the change process of the individual's state within a set time period, and to construct an intervention-free population disorder evolution model. The process of constructing the population disorder evolution model under the intervention includes: Based on the non-interventional crowd disorder evolution model, a state variable of robot intervention is introduced to construct a crowd disorder evolution model under intervention. Based on the evolution model of crowd disorder, robot-assisted intervention is planned as a stochastic optimal control problem. By solving the problem, the optimal intervention rate of the robot can be obtained, and then the optimal intervention individual and intervention time can be determined. Among them, local disorder Indicates adjacent individuals i and j The degree of confusion between them is defined as: ； in, E Represents the set of edges between individuals. Individual i and j The angle between the directions of motion, The value of increases monotonically with increasing disorder, i.e., angle. The smaller the angle, the more consistent the movement directions of the two individuals, and the local disorder degree approaches 0; conversely, the larger the angle... The larger the value, the closer the local disorder is to 1; Utilizing the chaos between any two individuals Define individual i and individuals j The global disorder is calculated through initialization steps and iterative processes, including: In the initialization step, a symmetric matrix is ​​constructed. Matrix A is n OK n A matrix of columns, by n×n Composition of numbers, Represents the first... i Line 1 j Number of columns, For the matrix Normalization is performed, resulting in: ； In the above formula, It is a degree matrix. ; During the iteration process, computational chaos occurs over time through the population network. The spread of the virus divides time into multiple steps, using... It means that, with Indicates time t The disordered influence matrix, that is, the matrix representing the first... t The chaotic effect of the order neighborhood, Indicates the influence of the first-order neighborhood. For weight parameters, It is a symmetric matrix. To represent the influence of the second-order neighborhood, calculate step by step. Until convergence; through iteration, the chaos eventually propagates to the entire network, resulting in a converged chaos matrix, which is: ； In the above formula, matrix I is the identity matrix, that is, a matrix in which all values ​​on the main diagonal are 1 and all others are 0. Define the disorder for each individual and compute its upper bound to determine the individual state; for any individual Individual confusion Defined as i The cumulative global disorder among all nodes is: ； In the above formula, It is a vector in which all elements are equal to 1.

2. The optimal intervention method for robot-assisted evacuation based on crowd disorder as described in claim 1, characterized in that, The process of marking time points refers to the counting process of the number of times an event occurs within a set time period; By using a time-point marker process method combined with stochastic differential equations, a non-intervention-based population disorder evolution model is constructed. The specific process includes: Define the transition events between disordered and ordered states, namely the transition events from ordered to disordered states and the transition events from disordered to ordered states. The counting process of the number of occurrences of two transition events within a set time period is described by intensity. Based on stochastic differential equations, the expected number of occurrences of two transition events within a set time period is calculated, and a non-interventional population disorder evolution model is constructed.

3. The optimal intervention method for robot-assisted evacuation based on crowd disorder as described in claim 1, characterized in that, The stochastic optimal control problem refers to finding the robot intervention rate that maximizes the expected value of the cumulative intervention benefit over a set time period.

4. A robot-assisted evacuation optimal intervention system based on crowd disorder, characterized in that, The robot-assisted evacuation optimal intervention method based on crowd disorder as described in any one of claims 1-3 includes: The data processing module is used to build a crowd network model and quantify the degree of disorder in crowd movement by measuring crowd disorder. The model building module is used to model the evolution of crowd disorder using stochastic differential equations and marked time point processes, and to build a crowd disorder evolution model. The optimization solution module is used to plan robot-assisted intervention as a stochastic optimal control problem based on the population disorder evolution model. By solving the problem, the optimal intervention rate of the robot can be obtained, and then the optimal intervention individual and intervention time can be determined.

5. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, complete the steps of an optimal intervention method for robot-assisted evacuation based on crowd disorder as described in any one of claims 1-3.

6. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, complete the steps of an optimal intervention method for robot-assisted evacuation based on crowd disorder as described in any one of claims 1-3.

Images