RU2601133C2 - Simulation model of traffic and pedestrian flows in urban conditions based on agent-oriented approach - Google Patents

Abstract

FIELD: training means.

SUBSTANCE: invention relates to production of a simulation model of traffic and pedestrian flows used in training simulators. Multiple independent agents of a road network are generated, wherein each said agent contains a physical and logical models; then road users are simulated using obtained agents. Unique identifiers of the readjustment of the state of the agents for each road traffic participant are determined. Identifiers are processed by means of discrete-event simulation (DES). Enviroment for agents is generated and its visual image is created. Information on the road network used in constructing the traffic simulation model is loaded from the database.

EFFECT: technical result consists in creation of high-precision simulation model of traffic with the possibility of flexible adjustment of relationships between a great number of road users.

16 cl, 8 dwg

Description

FIELD OF TECHNOLOGY

The claimed invention relates to a method, system and computer-readable medium for creating a simulation model of the movement of traffic and pedestrian flows, used in simulators for driving training.

BACKGROUND

Today, there are many different systems and methods for building a simulation model for use in driving simulators.

The TRANSIMS [1] package (TRansportation ANalysis and SIMulation System) is part of the Los Alamos National Research Center model upgrade program for the multi-lane road transport, paid for by the US Department of Transportation and the Environmental Protection Agency. TRANSIMS models each transport unit and evaluates its movement over its movement along the designated route. Thus, TRANSIMS predicts the behavior of each participant in the transport stream.

This system has a modular structure.

- The Demand for Traffic module generates individual vehicles and their movement through the creation of settlement models, the data on which are taken from the census or other sources. In this case, the module generates the movement of each vehicle.

- The intermodular “Route planner” uses cost-based indicators of traffic determined on the basis of demographic and other data individually for each consumer of transport demand. It presents the optimality data of the selected traffic mode for each vehicle, made during the preparation of the route plan.

- The block “Micromodeling of motion” reproduces trips along the transport network to predict the characteristics of individual vehicles and the transport system as a whole. He is trying to reproduce the route of each vehicle on a city-wide scale. So, for example, each passenger vehicle has a driver who implements its own logic of movement according to its route, accelerating or slowing down the movement of the vehicle.

- The block “Air Quality” receives information from the block “Micromodeling of motion” and evaluates environmental indicators. The emission model evaluates environmental parameters for both vehicles in motion and those awaiting movement. Modules have various feedbacks. So, unplanned trips naturally affect weekly activity. Those trips that during the micromodeling process will be longer than it actually turned out to require adjustment (model calibration). That is, the model must be constantly adjusted to the real transport situation. The disadvantages of the model is the lack of the ability to simulate pedestrian flows and recreate emergency situations.

AIMSUN [2] is a microscopic model developed at the Polytechnic Institute of Barcelona, designed to simulate urban and suburban traffic. Includes proprietary design and build models. The road network is based on reference nodes and links describing sections of roads between them. The model includes representations of traffic lights, signs and road markings. The conceptual architecture of the AIMSUN simulation system is shown in the figure below.

This model does not provide the ability to simulate pedestrian flows, as well as the ability to recreate emergency situations.

VISSIM Model [3].

VISSIM is a discrete, stochastic time-step model developed at Karlsruhe University in Germany. Movement participants are represented as separate entities. The model contains a psychophysical model of repetition for longitudinal movement and a rule-based transverse motion algorithm. The applied model of repetition is based on the Wiedmann model, which is based on the assumption that the driver can be in one of 4 states.

- Free movement - there is no influence in front of the going car, the driver seeks to maintain his own speed.

- Approach - the driver adapts its own speed to the speed of the vehicle in front by braking to a safe distance.

- Following - the driver follows the car without acceleration or braking.

- Braking - the driver brakes when approaching a distance less than a safe distance.

In the case of driving along a multi-lane road, a lot of hierarchically organized rules form a model of rebuilding.

This model does not provide the ability to simulate pedestrian flows, as well as the ability to recreate emergency situations.

Model Paramics [4].

Developed at the Edinburgh Parallel Computing Center, Paramics Simulation System uses a combination of microscopic and macroscopic models to simulate city-wide traffic. In addition to the leader following model and rebuilding model, the system includes an overtaking model in the oncoming lane and a parking model.

Distinctive features of the system are two modules: dynamic route selection and emergency situations. The first on the basis of the current traffic situation builds a route for each car. The second imitates emergency situations with the help of fluctuations in the behavior of the driver in the traffic stream.

This model is the closest in its technical essence to the claimed invention, however, its disadvantages are the lack of the possibility of deliberate creation of certain emergency situations, as well as the modeling of pedestrian flows.

SUMMARY OF THE INVENTION

The task to which the present invention is directed is to create a new type of traffic simulation model based on an agent-based approach, containing many different agents - participants of the road traffic, which allows for flexible adjustment of interaction parameters between agents within the model, thereby maximizing the simulation model for driving training in real conditions.

The technical result is the creation of a high-precision simulation model of traffic with the ability to flexibly customize the relationships between many road users. The claimed technical result is achieved due to the implementation of the method of constructing a simulation model of traffic, which contains stages in which:

- download information about the road network from the database;

- generate many autonomous agents of the road network, and each of the mentioned agent contains a physical and logical model;

- using the mentioned agents model road users;

- determine the unique identifiers of events recalculating the state of agents for each participant in the road traffic;

- process the mentioned identifiers using discrete event modeling (DSM);

- generate an environment for the functioning of agents;

and

- visualize the resulting environment for the functioning of agents.

In a private embodiment of the claimed invention, simulation model agents are: cars, trucks, public transport, road service equipment, pedestrians, animals, motorcyclists, slow-moving vehicles, cyclists, subway rolling stocks, or combinations thereof.

In a private embodiment of the claimed invention, the physical model of the agents contains information about the geometric shape of the road user, information about his position in space, speed and acceleration parameters.

In a private embodiment of the claimed invention, the agent’s logical model contains the rules of behavior of road users, which include navigation methods, maneuvering methods, methods of controlling the dynamics of the physical model, and script execution methods.

In a particular embodiment of the claimed invention, the physical and logical models recalculate their state after a given period of time.

In a particular embodiment of the claimed invention, state recounting is carried out using agent state recount event identifiers.

In a particular embodiment of the claimed invention, at least one agent is non-autonomous and is assigned an identifier for executing a scenario creating an extreme situation in a simulation model.

In a private embodiment of the claimed invention, the script is loaded from the database.

In a particular embodiment of the claimed invention, one or more agents executing a scenario creating an extreme situation in a virtual model do not participate in the general movement with other agents until the script is completed.

In a particular embodiment of the claimed invention, after execution of the aforementioned scenario, at least one agent continues to participate in the general movement with other agents of the simulation model. In a particular embodiment of the claimed invention, data are generated on the position, speed, acceleration, direction of motion and angular velocity of all road users. The claimed technical result is also achieved by implementing the claimed invention as a device, system and computer-readable medium.

Particular embodiments of the claimed invention will be described in detail below.

BRIEF DESCRIPTION OF THE DRAWINGS

In FIG. 1 shows the sequence of steps of a method for constructing a simulation model of traffic.

In FIG. 2 - algorithm of the component of maneuvering the logical model of a road user.

In FIG. Figure 3-6 shows a graphical representation of the simulation model.

In FIG. 7 shows an example implementation of a device for constructing a simulation model of traffic.

In FIG. 8 shows an example implementation of a system for constructing a simulation model of traffic.

DETAILED DESCRIPTION OF THE INVENTION

In FIG. 1 depicts a method (100) that is implemented when building a simulation model of traffic.

The method for constructing a simulation model of traffic (100) begins with loading information about the road network from the database at step (110). To simulate a road network, two graphs are used that have intersecting sets of nodes, namely a topographic graph and a logical graph.

A topographic graph is a graph whose nodes are points in three-dimensional space, and the edges contain information about the presence of the road, its width and geometric shape (straight line or arc with a given radius). The logical graph uses as its own nodes the nodes of the topographic graph where the intersection of several directions for movement occurs (more than one incoming edge, or more than one outgoing edge), or only one incoming / outgoing edge. It contains information about the types of road users who can move along the corresponding edges of the topographic graph (car, pedestrian, train). The nodes of the logical graph contain information about the possibility of a transition between logical arcs and the type of such a transition (straight, right, left, turn).

The road network also contains embedded objects having their own road networks, for example, subway lines, trains, level crossings, tram lines, etc. These objects are necessary for the possibility of modeling objects along which there is a movement of road users (public transport stops, parking lots, railway stations), objects containing fragments of the road network are used in the road network model, which simplifies the development of models of road networks such as routes the movement of buses or subway lines.

The database is preferably a Microsoft SQL Server relational database management system for storing the road network. All elements of the road network, such as nodes and edges of the road graph, information about objects of traffic, subway stations, are stored in the corresponding tables, and the links between these elements are stretched as links between the tables. Also, emergency scenarios are stored in separate tables.

At step (120), a plurality of simulation model agents — road users — are generated. To achieve realism in the behavior of the simulation model of a road user, an agent-based approach is used. The traffic model is implemented as a multi-agent system. At step (130), each road user is modeled by an autonomous agent without a long-term planning function for their actions, containing a physical model and a logical model that interacts with a fixed set of rules of behavior.

Agents in the simulation model are: cars, trucks, public transport, road service equipment, pedestrians, animals, motorcyclists, slow-moving vehicles, cyclists, subway rolling stocks, or combinations thereof.

The physical model of the agent contains information about the boundaries of the object represented in the form of a certain geometric primitive (a figure describing the shape of the road user), as well as a set of data on the position and orientation of the agent in space, its speed and acceleration.

After each fixed period of time, the physical model recalculates its own state. The agent’s logical model contains thematic components containing sets of rules of behavior, and contains semantic information about the traffic situation. The rules of behavior are divided into sets of thematic methods: navigation methods, maneuvering methods, methods of controlling the dynamics of a physical model, script execution methods. The logical model recalculates the state of its own components after a period of time, the value of which depends on the current speed of the agent. The recalculation of the model of their states is carried out using the identifiers of events for recalculating the state of agents.

Navigation methods specify the rules for determining the desired orientation and the desired position in space on a logical map. Maneuvering methods set the parameters for the rules of interaction of agents with each other in accordance with the rules of the road (SDA) and allow you to implement such maneuvers, such as rebuilding, overtaking. The methods of controlling the dynamics of the physical model allow you to determine the necessary acceleration and angle of rotation of the wheels to achieve the desired position in space and orientation. Script execution methods allow using agents to recreate extreme (emergency) situations.

At step (140), unique identifiers of each model module and all agent components (call destinations) are determined. Unique identifiers of all their states that need recounting are also determined. The events generated in step (150) contain the specified identifiers and the time the event was called. These events at the same stage are placed in the DSM for their processing. When the time of the event call is reached, the DSM finds the necessary addressee by its identifier and initiates the function of recalculating the desired state from the addressee by the status identifier, then this event is excluded from the DSM. Recalculation is performed relative to the recipient's parameters. If it is necessary to process any state of any addressee periodically (for example, recalculating the position, speed), then upon completion of processing this state and deleting the event, a new event is created in the DSM, including the addressee identifier, the event identifier and the new call time.

Next, at step (160), data about the agents is transferred to the environment where the agents function. The environment for the functioning of agents is an intermediate link between a model for simulating traffic and external modules relative to it.

At step (170), a visualization environment of the functioning of agents is carried out, containing many agents in virtual space. Visualization can be carried out on all computing systems that display virtual space visualization on monitors, or projectors, or any other suitable type of device for presenting data.

In FIG. 2 shows the algorithm of the component of maneuvering the logical model of a road user.

Data about the agents is processed by the DSM, which initiates the recalculation of the state parameters of the agents, in particular, each agent analyzes its state in relation to other agents located in its coverage area. For this, each agent has the function of constructing a sensor — an area for calculating the parameters of other agents that have fallen into the mentioned area, or those whose sensors have fallen into the mentioned area. The search for agents inside the sensor is carried out using the module - dispatcher. Using data from the dispatcher, agents prioritize travel, such as intersections, crossings, traffic lights, pedestrian crossings, etc. Agents recount their condition and, based on the data received, depending on the traffic situation, perform one or another action that does not contradict traffic rules. After performing a reaction to one or more agents that have fallen into the sensor, the agent recounts its status and sends the corresponding identifiers to the DSC to plan further recounting.

Each agent recounts its position in the road network many times over a certain period of time and assesses the need for interaction with other road users. The various components of the road user model are responsible for this: the “navigation” component and the “maneuvering” component that are part of the logical module of the road participant model. During agent initialization in the DSM, these components place the identifiers of the methods for recalculating the necessary parameters with the indicated call time. During the call of these methods, if necessary, the method identifiers are re-placed in the DSCM with a new call time (the call time of the order of 0.01 sec depends on the speed of the agent). The generation of data on the position, speed, acceleration, direction of motion and the angular velocity of all road users is carried out.

In FIG. 3-6 are examples of display of the claimed simulation model. To simulate busy traffic around special points, agents are emulated that simulate, for example, cars and pedestrians. To ensure the realism of the simulation, the creation of machines occurs beyond a given horizon of visibility from these points and depends on the type of terrain and the type of point of visibility. To ensure system performance, machines that leave the line of sight are removed. Various objects can be used as generation points, depending on the application of the model, for example, in driving simulators - this is a model of a vehicle controlled by a real crew. To recreate emergencies for training people, scenarios of real extreme situations are used. Agents that recreate an emergency do not participate in the general road traffic during the execution of the scenario, and only after its execution (optionally after a specified period of time) do they follow the SDA. These agents are not autonomous; to execute the given scenarios, the agents are assigned the corresponding identifier, which allows the agents to download the required script from the database, until the completion of which the mentioned agents will not participate in the general movement along with the autonomous agents. Moreover, at least one agent who created an extreme situation, or a group of agents, can continue to participate in the general movement. These scripts are user configurable. All agents within the simulation model move along the road network in accordance with the traffic rules, according to which navigation methods and maneuvering methods were compiled. Also, taking into account the traffic rules, the dispatcher module works. Traffic agents interact with traffic control elements such as traffic lights, traffic signs, traffic controllers, etc. User of the model - an instructor or user, can adjust the flux density depending on the recreation of the desired situation. An agent that is not controlled by the logic of the model, for example, the agent - user, will also be read by the sensor of each agent in the environment of the agents to obtain information on the need to perform one or another action. In this case, it is possible to adjust the intensity for each type of road user individually, and indicate a fraction of the maximum number of road users for traffic in general.

If conflicts arise during the interaction of agents, the identifiers of such agents are passed to the dispatcher, which resolves conflicts between the agents. Also, the dispatcher is contacted when two agents determine the intersection passage priority. The dispatcher module contains the rules for determining the priorities of travel, the rules for resolving conflict situations, as well as all agents register their position in it, which is necessary for the speed of searching for agents inside the sensors by other agents.

The claimed invention is also implemented using a device for constructing a simulation model of the traffic shown in FIG. 7, and a computer readable medium. The device (200) contains one or more processors (210) connected via a serial bus to one or more memory devices (220), input / output devices (I / O) (230) and I / O interfaces (240) ) The memory (230) of the device (200) contains instructions that, when executed by one or more processors (210), implement the method for constructing a simulation model of the road traffic described above. The device (200) can be a stationary computer, a mainframe, a server or server cluster, a laptop, and any other suitable type of device that allows implementing the method for constructing a simulation model according to the instructions. The storage medium is a read-only memory (ROM), or random access memory (RAM), or a hard disk (HDD), or an external computer-readable storage medium, or a combination thereof. An external computer-readable storage medium is selected from the group: USB flash drive, memory card, optical disk, mini-disk, external HDD disk or other suitable type of medium, with the possibility of its use in the device (200). The database (250) containing the necessary data for building a simulation model of traffic can be an external data storage connected via a wired or wireless data network to device (200), for example, the Internet, intranet, LAN, WAN, WLAN, etc. .p., as well as contained directly in the device (200) (this option is not shown in Fig. 7).

I / O interfaces (240) are, but are not limited to, for example, serial ports, parallel ports, universal serial bus (USB), IEEE-1394 (i. Link or Fire Ware), LAN, or any other type of interface that uses a particular particular embodiment of the device (200).

A computer-readable storage medium is a storage medium containing computer-readable instructions. Which, when executed, prompt one or more processors to perform the steps of a method for constructing a simulation model of traffic. A computer-readable medium may be any of the aforementioned memory means or media suitable for storing computer-readable instructions for execution by a processor.

In FIG. Figure 8 shows an exemplary view of a system for constructing a simulation model (300). System (300) is an environment of devices connected via a data network (310). The device (320), which is the main computing device - the processor, in particular the server, implements the steps of the method disclosed in the present materials above. Devices (330) to (360) are user devices or workstations, such as desktop computers or laptops, connected via a data network (310) or directly to a server (320). Devices (330) - (360) can be networked together. In a particular embodiment, groups of devices can organize clusters that can individually be connected to a server device (320), this is used as an option for generating a scenario for executing extreme situations or simulating a tuple, tracking, etc. The device (370) is a means for displaying data, for example, a display, a monitor, a projector, which is interfaced by appropriate means with each of the aforementioned user workstations.

The examples of implementation of the claimed invention given in these materials were successfully implemented in the following systems:

- a set of LiAZ-5256/6212 simulators installed in the Mosgortrans State Unitary Enterprise;

- VAZ-2110 simulators (ROSTO (DOSAAF));

- simulators Ural-4320, (Minsk Military Academy of the Republic of Belarus);

- KAMAZ-4350/5350 simulators (ROSTO (DOSAAF));

- a complex of simulators VAZ-2110 (Federal Security Service of the Russian Federation).

The information on the implementation of the claimed invention set forth in these materials of the application should not be interpreted as limiting other, private embodiments of the claimed invention that do not go beyond the disclosure of information of the application, and which should be obvious to specialists in the field of technology having the usual qualification, which calculated the claimed technical solution.

Information sources

1. Angshuman G., Introduction to TRANSIMS.

(URL: https://t-square.gatech.edu/access/content/group/28974.201002/Introduction_to_TRANSIMS_Part1.pdf)

2. Barcelo J. Microscopic traffic simulation: a tool for the design, analysis and evaluation of intelligent transport systems // Journal of intelligent and robotic systems. - 2005. - Vol. 41, No. 2-3. - P. 173-203.

3. Lownes N.E. VISSIM: a multi-parameter sensitivity analysis / N.E. Lownes, R.B. Machemehl // Proceedings of the 38th conference on Winter simulation. - 2006 .-- P. 1406-1413.

4. Cameron G. PARAMICS - moving vehicles on the connection machine / G. Cameron, B. Wyline, D. McArthur // Proceedings of the conference on supercomputing. - IEEE computer society press, 1994. - P. 291-300.

Claims (16)

1. A method of constructing a simulation model of traffic, containing stages in which:
- download information about the road network from the database;
- generate many autonomous agents of the road network, and each of the mentioned agent contains a physical and logical model;
- using the mentioned agents model road users;
- determine the unique identifiers of events recalculating the state of agents for each participant in the road traffic;
- process the mentioned identifiers using discrete event modeling (DSM);
- generate an environment for the functioning of agents;
and
- visualize the resulting environment for the functioning of agents. 2. The method according to p. 1, characterized in that the agents in the simulation model are: cars, trucks, public transport, road service equipment, pedestrians, animals, motorcyclists, slow-moving vehicles, cyclists, subway rolling stock, or a combination thereof. 3. The method according to p. 1, characterized in that the physical model of the agents contains information about the geometric shape of the road user, information about his position in space, speed and acceleration parameters. 4. The method according to p. 1, characterized in that the logical model of the agent contains the rules of behavior of road users, which include navigation methods, maneuvering methods, methods of controlling the dynamics of the physical model, script execution methods. 5. The method according to p. 1, characterized in that the physical and logical models recalculate their state after a given period of time. 6. The method according to p. 5, characterized in that the recalculation of the state is carried out using the identifiers of events recalculating the state of agents. 7. The method according to claim 1, characterized in that at least one agent is non-autonomous and is assigned an identifier for executing a script that creates an extreme situation in the simulation model. 8. The method according to p. 7, characterized in that the said script is loaded from the database. 9. The method according to p. 7, characterized in that one or more agents that execute the script that creates an extreme situation in the simulation model do not participate in the general movement with other agents until the script is completed. 10. The method according to p. 7, characterized in that after the execution of the mentioned scenario, at least one agent continues to participate in the general movement with the other agents of the simulation model. 11. The method according to p. 1, characterized in that the generation of data on the position, speed, acceleration, direction of motion and the angular velocity of all road users. 12. A device for constructing a simulation model of traffic containing one or more processors, devices and input / output interfaces and at least one memory tool, and the memory contains instructions executed by at least one processor, which, when executed, execute The method according to any one of paragraphs. 1-11. 13. The device according to p. 12, characterized in that the memory means is a read-only memory (ROM), or random access memory (RAM), or hard disk drive (HDD), or an external computer-readable storage medium, or combinations thereof. 14. The device according to p. 13, characterized in that the external computer-readable storage medium is selected from the group: USB flash drive, memory card, optical disk, mini-disk, external HDD. 15. A system for constructing a simulation model of traffic comprising a server, a database and one or more workstations, the server comprising one or more processors and at least one memory means, the memory containing executable by at least one processor instructions that, when executed, perform the method according to any one of paragraphs. 1-11, and each of the aforementioned workplace additionally contains a means of visualizing information received from the server. 16. A computer-readable storage medium containing computer-readable instructions executable by at least one processor, which, when executed, perform a method for constructing a simulation model of traffic according to any one of paragraphs. 1-11.

Images