Pedestrian node selection behavior simulation method based on unreal engine

Abstract

The invention relates to the technical field of public transport stations and crowd behavior simulation, and discloses a pedestrian node selection behavior simulation method based on an unreal engine. The method comprises the following steps: firstly, constructing a dynamic three-dimensional scene containing interactive equipment logic and environment elements by utilizing an unreal engine; defining pedestrian node alternative schemes and dividing queue groups to construct a nested structure; then objective attributes and finite rationality factors influencing node selection are determined; constructing a queue group and a utility function of a specific node level based on a nested structure and finite rationality factors; calculating a pedestrian selection probability through a hybrid nested Logit model; and finally running simulation in the unreal engine and outputting a result. The method solves the problems that an existing simulation method is difficult to reflect pedestrian heterogeneity and limited cognition and is poor in applicability, has wide applicability, comprehensive influence factor consideration and high visualization ability, and can be used for scenes such as station layout optimization, emergency evacuation and intelligent scheduling.

Description

Technical Field

[0001] This invention relates to the field of public transportation station and crowd behavior simulation technology, specifically to a multi-node, multi-level pedestrian node selection behavior simulation method based on Unreal Engine, which can be applied to scenarios such as layout optimization of public transportation stations such as stations and airports, emergency evacuation drills, and intelligent scheduling evaluation. Background Technology

[0002] In public places such as train stations and airports, pedestrians (such as passengers) exhibit diverse choices when passing through turnstiles and transfer facilities (elevators, escalators, etc.). These behaviors are influenced by factors such as distance, queue length, equipment status, personal preferences, and cognitive abilities, and their rationality directly relates to the operational efficiency of the station and the safety of pedestrian traffic.

[0003] Existing pedestrian node selection behavior simulation technologies have the following shortcomings: Model assumptions are out of touch with reality: Traditional simulation models often use static rules (such as fixed path selection), single probability models (such as the standard Logit model), or completely rational assumptions, without considering the limitations of pedestrians' cognitive abilities, such as the inability to obtain all node information and errors in the perception of queue length, resulting in a large deviation between simulation results and actual behavior. Insufficient heterogeneity representation: Existing models fail to reflect the differences in preferences among different pedestrians, such as some pedestrians prioritizing nearby nodes and others prioritizing short queues, and cannot accurately simulate the diverse choices of the crowd; Poor applicability and visualization: Most models are designed for a single type of node, such as only turnstile design, and are difficult to extend to other nodes such as elevators and escalators; moreover, the simulation scenarios are mostly two-dimensional or low-precision three-dimensional, which cannot intuitively show the pedestrian flow and node selection process, which is not conducive to the decision-making of site planners.

[0004] In recent years, although hybrid nested Logit models can capture multi-level decision structures and stochastic preferences, they have not fully incorporated the bounded rationality factors of pedestrians, such as perceptual errors and heuristic decision rules. At the same time, although game engines such as Unreal Engine have the ability to build high-fidelity 3D scenes, they have not yet been deeply integrated with pedestrian behavior models to achieve accurate node selection simulation. Summary of the Invention

[0005] This invention aims to overcome the shortcomings of existing technologies and provide a pedestrian node selection behavior simulation method based on Unreal Engine. It solves the problems of simulation results being detached from reality, having narrow applicability, and poor visualization, and achieves accurate, universal, and highly visualized simulation of pedestrian node selection behavior in complex station environments.

[0006] To achieve the above objectives, this invention provides a method for simulating pedestrian node selection behavior based on Unreal Engine, comprising the following steps: S1: Scene modeling. Using the Unreal Engine Datasmith plugin, the Revit / IFC format BIM model is converted into an Unreal Engine mesh, and the mesh is lightweighted. Based on the Unreal Engine Blueprint system, a dynamic 3D scene containing interactive equipment logic such as elevators, turnstiles, and escalators is constructed. The on / off, operation, and stop interactive behaviors of the equipment are realized through custom blueprint nodes and event-driven mechanisms. Environmental elements simulating the impact of temperature and rain / snow weather on walking are generated. S2: Define the selection set and nested structure, determine all available pedestrian nodes as alternatives, divide the nodes into multiple queue groups according to the physical location and functional characteristics of the nodes, and construct a nested structure that first selects the queue group and then selects the specific node within the group. S3: Identify influencing factors, including key objective attributes such as time, space, and node type; consider pedestrian perception error, limited information, heuristic rules, and bounded rationality factors that are satisfactory but not optimal, and reflect these bounded rationality factors by adding a random utility term to the utility function, defining a visible set, introducing attribute simplification, threshold effect, anchoring, and adjustment; S4: Utility function construction. Based on the nested structure and bounded rationality factors, construct pedestrian selection node utility functions at the queue group level and specific node level. The utility functions include the influence of key attributes and random parameters on selection behavior. S5: Calculate the selection probability. Based on the utility function, use a hybrid nested Logit model to calculate the probability that a pedestrian selects each queue group and a specific node within the group, taking into account the randomness and heterogeneity of the parameters, and calculate the probability using software. S6: Unreal Engine simulation. Set the pedestrian's initial position and destination parameters, run the simulation to simulate the pedestrian's node selection behavior, and output simulation results including pedestrian path selection and node traffic distribution.

[0007] Furthermore, in step S1, the lightweighting process involves removing redundant details that do not affect the simulation results but increase the computational burden, in order to improve the loading and running efficiency of the model in the simulation environment.

[0008] Furthermore, in step S1, the interactive device logic includes: elevator door opening and closing, operating status and passenger call response logic; gate ticket information recognition, valid ticket triggering barrier opening, invalid / unswapped card keeping closed and alarm logic.

[0009] Furthermore, in step S3, the key objective attributes also include node queue length, waiting time, node fault status, queue movement speed, and queue length variance; the heuristic rules include selecting the shortest queue within the visual range, selecting the nearest node, following the pedestrian in front, and avoiding low-speed nodes.

[0010] Furthermore, in step S4, the node utility function is: ; A specific constant term for queue group k is used to capture unobserved group characteristics; These are the preference parameters for passenger n pairs: queue length, queue length variance, queue movement speed, and average distance of queue groups, respectively. For the corresponding attributes of queue group k; To measure the inclusion value of the expected maximum utility of choices within group k, ,in Let be the utility function of the lower-level gate j; This parameter measures intragroup correlation and its value ranges from 0 to 1. =1 indicates that the groups are completely independent, which is equivalent to the standard Logit model. The smaller the value, the stronger the correlation within the group; The random error term for the queue group utility follows a Gumbel distribution.

[0011] Furthermore, in step S4, the utility function at the specific node level is: ;in, For a specific constant term of node j, , These are the random preference parameters for passenger n pairs of node distances and node types, respectively. Let be the distance or time from the passenger to node j. A dummy variable of node type. It is a random error term that follows a Gumbel distribution and is independent and identically distributed within the group.

[0012] Furthermore, in step S5, the software is Stata. After statistically analyzing the values ​​of each parameter through video data, the magnitude of the preference parameter is calculated in Stata.

[0013] Furthermore, in step S6, the simulation process is as follows: the hybrid nested Logit model, parameters and attribute data are integrated into the Unreal Engine, the model is called at each time step to calculate the pedestrian selection probability, the pedestrian selection behavior is determined according to the probability distribution, and the pedestrian position and status are updated until the pedestrian completes the travel task or the simulation end time is reached.

[0014] Furthermore, the pedestrian nodes include turnstiles, elevators, escalators, and stairs, and are applicable to station layout optimization, emergency evacuation simulation, and intelligent scheduling evaluation scenarios in stations, airports, and subway stations.

[0015] Compared with the prior art, the present invention has the following beneficial effects: Wide applicability: This method is not only applicable to the selection of turnstile nodes, but can also be extended to various node types such as elevators, escalators, and stairs; and can be applied to different stations such as stations, airports, and subway stations. It can be adapted by adjusting the node type and utility function parameters, and has strong versatility and portability. Accurate simulation results: This method considers both objective attributes, such as distance and queue length, and bounded rationality factors, such as perceptual error and heuristic rules. It also uses a hybrid nested Logit model to reflect pedestrian heterogeneity, making the simulation results more consistent with actual pedestrian behavior and reducing the bias rate. High visualization capabilities: Based on Unreal Engine, high-fidelity 3D scenes can be built to intuitively display pedestrian movement trajectories, node selection processes, and congestion conditions. Site planners can understand the simulation results without professional knowledge, providing intuitive decision support for layout optimization, such as increasing the number of wide turnstiles, and emergency evacuation, such as simulating evacuation routes during fires. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the behavior simulation method for pedestrian nodes based on Unreal Engine; Figure 2 This is a schematic diagram of a train station scene built using Unreal Engine. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Figure 1 A pedestrian node selection behavior simulation method based on Unreal Engine is provided for embodiments of the present invention. The method flow is as follows: S1: The 3D scene construction of the station in Unreal Engine S11: Obtain the RVT format BIM model of the station. This model contains detailed information about the station's architectural structure, including elements such as the concourse, platforms, stairs, passageways, and various entrances and exits. Convert the model into Unreal Engine meshes using the Datasmith plugin. Simultaneously, perform lightweight model processing, removing redundant details that do not affect the overall simulation results but increase computational burden, such as... Figure 2 .

[0019] S12: Based on the Unreal Engine's Blueprint system, construct the logic for interactive devices in the station. Taking the elevator as an example, create an elevator Blueprint component, defining its door opening and closing, operating status, and logic for responding to passenger calls, enabling it to perform corresponding actions in the simulation based on passenger instructions. For the turnstiles, construct logic to recognize pedestrian ticket information; the gate opens when a valid ticket approaches, and remains closed and triggers an alarm when an invalid ticket or no card is swiped.

[0020] S13: Use the Unreal Engine physics engine to generate station environment elements, such as simulating temperature regulation inside the station, adjusting the ground friction coefficient during rainy weather, and affecting pedestrian walking speed and stability.

[0021] S2: Define selection sets and nested structures S21: In the station's turnstile area, identify all available turnstiles as alternatives, numbered j = 1, 2,..., J.

[0022] S22: Based on the physical location of the turnstiles and the station layout, divide adjacent turnstiles whose queues may affect each other into a single queue group, numbered k = 1, 2, ..., K. For example, divide a row of 4 turnstiles into 2 groups (1-2, 3-4). The specific division can be determined based on the actual situation of the station and observation data.

[0023] S3: Identify key attributes and bounded rationality factors S31: Collect the location information of each turnstile, calculate the walking distance and estimated walking time from the passenger's current location to the turnstile; record the turnstile type (wide turnstile, narrow turnstile) and whether it is malfunctioning; and statistically analyze the average queue length (number of people or estimated waiting time), the variance of the queue length, the queue movement speed (average processing time), and the average physical distance between the turnstiles in the group for each queue group.

[0024] S32: Consider passenger perception errors, such as overestimating or underestimating queue length; limited information for passengers, such as only being able to observe part of the turnstile queue within their line of sight; heuristic rules used by passengers, such as “choosing the shortest-looking queue,” “choosing the queue closest to me,” “following the person in front of me,” “avoiding turnstiles that look slow,” etc.; and the behavioral characteristics of passengers seeking satisfactory rather than optimal solutions, i.e., as long as they find a turnstile that meets certain conditions (such as waiting time being below a certain threshold).

[0025] S4: Constructing the utility function S41: Construct the utility function for each node.

[0026] ;

[0027] in, A specific constant term for queue group k is used to capture unobserved group characteristics; These are the preference parameters for passenger n pairs: queue length, queue length variance, queue movement speed, and average distance of queue groups, respectively. For the corresponding attributes of queue group k; To measure the inclusion value of the expected maximum utility of choices within group k, ,in Let be the utility function of the lower-level gate j; This parameter measures intragroup correlation and its value ranges from 0 to 1. =1 indicates that the groups are completely independent, which is equivalent to the standard Logit model. The smaller the value, the stronger the correlation within the group; The random error term for the queue group utility follows a Gumbel distribution.

[0028] S42: Utility function within the turnstile.

[0029] ; in, For a specific constant term of node j, , These are the random preference parameters for passenger n pairs of node distances and node types, respectively. Let be the distance or time from the passenger to node j. A dummy variable of node type. It is a random error term that follows a Gumbel distribution and is independent and identically distributed within the group.

[0030] S5: Calculate the selection probability By statistically analyzing the values ​​of each parameter through video data, the magnitude of each parameter is calculated in the professional software Stata.

[0031] S6: Simulation based on Unreal Engine The hybrid nested Logit model constructed above, along with its related parameters and attribute data, is integrated into Unreal Engine. By writing scripts or using the development tools provided by the engine, the model can be used for real-time calculation and application during the simulation process.

[0032] The simulation program is started, allowing pedestrians in the virtual scene to move and select nodes under the drive of the engine. At each time step, based on the pedestrian's position and surrounding environment information, a hybrid nested Logit model is invoked to calculate the probability of the pedestrian selecting different turnstiles, and the pedestrian's selection behavior is randomly determined according to the probability distribution. At the same time, the pedestrian's position and status information are updated until all pedestrians complete their travel tasks or the set simulation end time is reached.

[0033] The simulation results are analyzed to evaluate the rationality of the existing gate node layout and facility configuration at the station, and to predict potential congestion points and bottleneck locations. Based on the simulation results, optimization suggestions are proposed, such as adjusting the gate layout or changing its location and number. The implementation effects of these optimization measures are simulated to provide a scientific basis for the planning and management of actual stations.

[0034] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for simulating pedestrian node selection behavior based on Unreal Engine, characterized in that, Includes the following steps: S1: Scene modeling, using the Unreal Engine Datasmith plugin to convert the Revit / IFC format BIM model into an Unreal Engine mesh, and then performing lightweight processing on the mesh; The Unreal Engine-based Blueprint system constructs a dynamic 3D scene containing the logic of interactive devices such as elevators, turnstiles, and escalators. The system implements the on / off, operation, and stop interactive behaviors of the devices through custom blueprint nodes and event-driven mechanisms. Generate environmental elements that simulate the effects of temperature, rain, and snow on walking; S2: Define the selection set and nested structure, determine all available pedestrian nodes as alternatives, divide the nodes into multiple queue groups according to the physical location and functional characteristics of the nodes, and construct a nested structure that first selects the queue group and then selects the specific node within the group. S3: Identify influencing factors, including key objective attributes such as time, space, and node type; consider pedestrian perception error, limited information, heuristic rules, and bounded rationality factors that are satisfactory but not optimal, and reflect these bounded rationality factors by adding a random utility term to the utility function, defining a visible set, introducing attribute simplification, threshold effect, anchoring, and adjustment; S4: Utility function construction. Based on the nested structure and bounded rationality factors, construct pedestrian selection node utility functions at the queue group level and specific node level. The utility functions include the influence of key attributes and random parameters on selection behavior. S5: Calculate the selection probability. Based on the utility function, use a hybrid nested Logit model to calculate the probability that a pedestrian selects each queue group and a specific node within the group, taking into account the randomness and heterogeneity of the parameters, and calculate the probability using software. S6: Unreal Engine simulation. Set the pedestrian's initial position and destination parameters, run the simulation to simulate the pedestrian's node selection behavior, and output simulation results including pedestrian path selection and node traffic distribution.

2. The method according to claim 1, characterized in that, In step S1, the lightweighting process involves removing redundant details that do not affect the simulation results but increase the computational burden, in order to improve the loading and running efficiency of the model in the simulation environment.

3. The method according to claim 1, characterized in that, In step S1, the interactive device logic includes: elevator door opening and closing, operating status and passenger call response logic, and gate ticket information recognition, valid ticket triggering barrier opening, invalid / unswapped card keeping closed and alarm logic.

4. The method according to claim 1, characterized in that, In step S3, the key objective attributes also include node queue length, waiting time, node fault status, queue movement speed, and queue length variance; the heuristic rules include selecting the shortest queue within the visual range, selecting the nearest node, following the pedestrian in front, and avoiding low-speed nodes.

5. The method according to claim 1, characterized in that, In step S4, the node utility function is: ; A specific constant term for queue group k is used to capture unobserved group characteristics; These are the preference parameters for passenger n pairs: queue length, queue length variance, queue movement speed, and average distance of queue groups, respectively. For the corresponding attributes of queue group k; To measure the inclusion value of the expected maximum utility of choices within group k, ,in Let be the utility function of the lower-level gate j; This parameter measures intragroup correlation and its value ranges from 0 to 1. =1 indicates that the groups are completely independent, which is equivalent to the standard Logit model. The smaller the value, the stronger the correlation within the group; The random error term for the queue group utility follows a Gumbel distribution.

6. The method according to claim 5, characterized in that, In step S4, the utility function at the specific node level is: ;in, For a specific constant term of node j, , These are the random preference parameters for passenger n pairs of node distances and node types, respectively. Let be the distance or time from the passenger to node j. A dummy variable of node type. It is a random error term that follows a Gumbel distribution and is independent and identically distributed within the group.

7. The method according to claim 1, characterized in that, In step S5, the software is Stata. After statistically analyzing the values ​​of each parameter through video data, the magnitude of the preference parameter is calculated in Stata.

8. The method according to claim 1, characterized in that, In step S6, the simulation process is as follows: the hybrid nested Logit model, parameters and attribute data are integrated into the Unreal Engine, the model is called at each time step to calculate the pedestrian selection probability, the pedestrian selection behavior is determined according to the probability distribution, and the pedestrian position and status are updated until the pedestrian completes the travel task or the simulation ends.

9. The method according to claim 1, characterized in that, The pedestrian nodes include turnstiles, elevators, escalators, and stairs, and are applicable to station layout optimization, emergency evacuation simulation, and intelligent dispatch evaluation scenarios in stations, airports, and subway stations.

Images