CN115081186A - Driving behavior simulation system supporting data driving and simulation method thereof - Google Patents
Driving behavior simulation system supporting data driving and simulation method thereof Download PDFInfo
- Publication number
- CN115081186A CN115081186A CN202210540780.3A CN202210540780A CN115081186A CN 115081186 A CN115081186 A CN 115081186A CN 202210540780 A CN202210540780 A CN 202210540780A CN 115081186 A CN115081186 A CN 115081186A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- data
- simulation
- driving behavior
- road network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000004088 simulation Methods 0.000 title claims abstract description 69
- 238000000034 method Methods 0.000 title claims abstract description 21
- 230000006399 behavior Effects 0.000 claims abstract description 56
- 238000012549 training Methods 0.000 claims description 15
- 230000002452 interceptive effect Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 claims description 4
- 230000007787 long-term memory Effects 0.000 claims description 4
- 230000007613 environmental effect Effects 0.000 claims description 2
- 238000000926 separation method Methods 0.000 claims description 2
- 230000015654 memory Effects 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 230000002045 lasting effect Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to a driving behavior simulation system supporting data driving and a simulation method thereof, wherein the system mainly comprises 3 modules: the general kernel is developed based on an Opendrive road network data format and is used for constructing a wide traffic environment; the method comprises the steps that a long-time memory network model, namely an SHRC-LSTM model, which is mixed with retraining constraints is used for realizing integrated modeling of various driving behaviors and simulating high-precision tracks; and realizing the independence and the connection of the driving behavior model and the general kernel by a customized interface. Compared with the prior art, the invention realizes the simulation reproduction of the traffic flow by providing a simulation system driven by complete data, can support the replacement of models in different scenes and different types, and simultaneously improves the precision, the flexibility and the expansibility of the microscopic traffic simulation.
Description
Technical Field
The invention relates to the technical field of microscopic traffic simulation, in particular to a driving behavior simulation system supporting data driving and a simulation method thereof.
Background
The microscopic traffic simulation technology can realize the multi-dimensional evaluation of road network traffic by simulating the driving behavior of vehicles, provides effective support for traffic management, and can also be used for evaluating a design scheme in a planning period. After about half a century of development, simulation systems are packaged as sophisticated simulation software, and have gained wide attention and applications in the industry. High simulation precision is an important guarantee for the reliability of a simulation result, and a simulation system is usually a complex structure with multi-module integration, wherein a driving behavior model is the most important module influencing precision. So far, the traditional driving behavior model spatially decomposes the motion into two non-interfering directions, namely a motion direction (longitudinal direction) and a vertical motion direction (transverse direction), and temporally divides the motion into various driving behaviors which are sequentially connected and respectively controlled by different models; and the model needs to be calibrated and verified according to the observation data to ensure that the model performance conforms to reality. However, the existing model greatly simplifies the actual behavior, and the complicated calibration process is accompanied by an error which is difficult to overcome, so that the simulation precision is not good. On the other hand, the rapidly developed data driving method has great potential in various fields, the data driving method is connected to a simulation system, the method is a feasible means for improving the simulation precision, and the application of the simulation technology in the industry is further promoted.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a driving behavior simulation system supporting data driving and a simulation method thereof, aiming at realizing the access of a driving behavior model driven by data by adjusting the structure and the connection of the prior simulation system so as to improve the precision and the compatibility of the simulation system.
The invention designs a simulation system structure and operation logic, reconstructs a simulation environment by reading road network and traffic flow information, and adopts a data driving model to simulate the motion of a vehicle, thereby realizing the independence and the connection between a driving behavior model and other modules.
The purpose of the invention can be realized by the following technical scheme:
the invention divides the whole simulation system into three parts (see the attached figure 1): the method comprises the following steps that firstly, a universal kernel is formed by a plurality of submodules such as input/output, infrastructure, traffic control, vehicle generation, a road network and the like and is used for building a simulation environment; secondly, a data-driven driving behavior model, namely a long-time memory network model (model full name: Sim-Hybrid training Constrained LSTM, abbreviated as SHRC-LSTM), with mixed Retraining constraint is added in the model training process to realize safe behavior prediction; and finally, the customized interface for connecting the two parts is used for realizing the transmission of vehicle motion information and ensuring the independence and connection of all the modules.
The method comprises the following steps:
a driving behavior simulation system supporting data driving comprises a general kernel, a data-driven driving behavior model and a customized interface, wherein the customized interface is used for connecting the general kernel and the data-driven driving behavior model for data transmission;
the general kernel is used for constructing a traffic scene and generating a vehicle state and an environment state;
the data-driven driving behavior model is used for predicting the vehicle state of the next time step according to the time sequence data of the vehicle;
the customized interface is used for integrating the vehicle state and the environment state, constructing time sequence data and transmitting the time sequence data to a data-driven driving behavior model; and (4) applying constraint to the predicted vehicle state of the next time step to obtain a re-revised vehicle state, and feeding back the re-revised vehicle state to the general kernel to update the traffic scene.
Further, the general-purpose kernel includes:
the input module is used for initializing a road network and vehicles and analyzing road network information;
the output module is used for outputting simulation statistical parameters and analysis results and deriving a vehicle motion track;
the basic setting module is used for configuring corresponding basic facilities for different road types;
the traffic control module is used for configuring a control scheme of the signal lamp;
the vehicle generation module is used for setting the entering position of the vehicle, the initial state of the vehicle and the entering interval of the vehicle according to the initialization information of the input module and generating the vehicle state according to the data of the traffic control module;
the road network module is used for constructing a road network and generating an environmental state according to the information of the basic setting module and the traffic control module and based on a preset road network format;
and the vehicle control module is used for updating the vehicle state in real time according to the vehicle state which is revised again and feeding back to the road network module and the output module.
Further, the road network module constructs a road network according to an OpenDRIVE road network format, the road network format includes road segments and nodes, the road segments are a set of one or more lanes in the same driving direction, the road segments are connected by the nodes, and one or more connectors are arranged in one node.
Further, the input of the data-driven driving behavior model comprises continuous running characteristics of the vehicle, interactive characteristics with surrounding vehicles, lane line information and traffic rules;
the model structure of the data-driven driving behavior model is a pre-constructed and trained SHRC-LSTM model;
the output of the data-driven driving behavior model is a two-dimensional position coordinate.
Further, the training process of the SHRC-LSTM model includes:
the method comprises the steps of obtaining a training set of driving behaviors, dividing the training set into a data set D1 and a data set D2, firstly training a long-term memory network model by adopting the data set D1, then inputting the data set D2 for prediction, checking and correcting inconsistent prediction results according to preset safety constraints, taking the corrected data set D2 as the data set D3, and integrating the data set D1 and the data set D3 to continuously train the long-term memory network model.
Further, the SHRC-LSTM model is a double-layer LSTM model.
Further, the safety constraints include that the maximum speed must not be exceeded, the distance from the surrounding vehicle must not be less than the minimum distance, and no reversing is performed.
Further, the computational expression of the security constraint is as follows:
in the formula,
as predicted coordinates at time i, Y i For the real coordinate at time i, Δ v y At maximum speed, at is the prediction time interval,
is the coordinate of the nearest vehicle in the right direction,
the coordinates of the nearest vehicle in the left direction,
is the minimum inter-vehicle distance in the right direction,
is the minimum inter-vehicle distance in the left direction,
to predict the speed.
Further, the constraints imposed by the customization interface on the predicted vehicle state for the next time step include a maximum speed constraint, a minimum separation constraint, and a no-reverse constraint.
The invention also provides a simulation method of the driving behavior simulation system supporting data driving, which comprises the following steps:
s1: initializing a road network according to the input module and the road network module;
s2: updating the road network state through a road network model;
s3: generating vehicles through a vehicle generation module according to the road network state;
s4: calculating the vehicle state;
s5: generating track coordinates of vehicles according to the vehicle states, extracting surrounding vehicles and roads according to the road network states to obtain a vehicle time sequence data set, and integrating state information;
s6: predicting the vehicle track according to the integrated state information through a data-driven driving behavior model;
s7: adding constraints to the vehicle trajectory prediction result;
s8: updating the vehicle state;
s9: judging whether the vehicle reaches a trip end point, if so, deleting the vehicle, otherwise, executing a step S10;
s10: generating a track record of the vehicle;
s11: judging whether the vehicle is the last vehicle at the current moment, if so, executing the step S12, otherwise, returning to the step S4;
s12: and judging whether the preset simulation termination time is reached, if so, ending the simulation, and otherwise, returning to the step S2.
Compared with the prior art, the invention has the following advantages:
(1) the SHRC-LSTM model provided by the invention extracts the space-time operation rule from the trajectory data, can realize high-precision vehicle trajectory prediction, and improves the simulation precision, thereby promoting the application of simulation to a certain extent.
(2) The invention is developed and adapted based on the Opendrive road network format, provides an adaptation algorithm of a road network and traffic flow model, is convenient for standardizing a road network structure and is beneficial to establishing a universal simulation system tool.
(3) According to the universal data-driven simulation framework provided by the invention, the driving behavior model can support two forms of a data-driven model and an analytic model, and the customized interface can ensure the universality of a system kernel structure on one hand and can maintain the safety of vehicles in the system on the other hand.
Drawings
FIG. 1 is a system framework diagram of a system for supporting data-driven driving behavior simulation according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of the control logic of a system for supporting data-driven driving behavior simulation according to an embodiment of the present invention;
FIG. 3 is a schematic illustration of an interactive vehicle inspection area provided in an embodiment of the present invention;
fig. 4 is a schematic diagram of a SHRC-LSTM model structure and training provided in the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
Example 1
As shown in fig. 1, the present embodiment provides a driving behavior simulation system supporting data driving, which includes a general kernel, a driving behavior model driven by data, and a customized interface, where the customized interface is used to connect the general kernel and the driving behavior model driven by data for data transmission;
the general kernel is used for constructing a traffic scene and generating a vehicle state and an environment state;
the data-driven driving behavior model is used for predicting the vehicle state of the next time step according to the time sequence data of the vehicle;
the customized interface is used for integrating the vehicle state and the environment state, constructing time sequence data and transmitting the time sequence data to the data-driven driving behavior model; and the method is also used for applying constraint on the predicted vehicle state of the next time step, and feeding back the re-revised vehicle state to the general kernel to update the traffic scene.
Each portion is described in detail below.
1. Building a common kernel portion
1.1, an input/output module. The module mainly realizes the initialization of the road network and the vehicles. At the beginning of the operation of the simulation system, no simulation individuals exist in the system. The traffic defined in the road network initialization stage needs to gradually enter the road network according to a certain rule, and three main problems need to be solved: the location of the vehicle entry, the initial state of the vehicle, and the interval of vehicle entry will invoke the vehicle generation module implementation.
The input module also performs a road network information analysis operation, and the analyzed information is transmitted to the correlation module. The OpenDRIVE format is a universal standard for describing roads and road networks based on extensible markup language (XML). The data stored in the OpenDRIVE file describes the geometry of the road and features along the line, and defines the underlying traffic signs and road infrastructure, such as lanes and signal lights.
The output module can output related statistical parameters and analysis results according to the requirements of the simulation scene; and after the motion trail of the vehicle is derived, generating a simulation animation.
1.2, infrastructure modules. Different infrastructures will be considered for different road types. If the road is a high expressway, considering a service area connected with the ramp at the entrance and the exit; if the urban road is the urban road, considering a parking lot; and the urban and rural roads are considered to be connected by roads and bridges.
1.3, a traffic control module. Mainly an adjustable signal lamp and a signal lamp timing scheme.
1.4, a vehicle generation module. The position of the vehicle entry is controlled by the following parameters: { Global abscissa G x Global ordinate G y Link, Link code id Lane code lane id (wherein, Link code Link) id Lane code lane id Which may be selected according to a simulated coordinate system), the parameters support both specification and random function generation. The initial state of the vehicle includes { speed v, acceleration a, direction angle γ, vehicle type T, in addition to the position p The parameters have randomness and correlation, and can be realized by adopting a random function according to requirements; the vehicles enter the interval and are expressed by adopting Poisson distribution. The poisson distribution is suitable for the condition that the traffic flow density is moderate or high, the mutual influence of the arrival conditions among individuals is weak, and the arrival is random. The poisson distribution characterizes the distribution of the number of occurrences over a period of time, with important properties being that both the mathematical expectation and the variance are equal to the parameter λ. Assuming that the number of arriving vehicles in a signal cycle is X, possible values for X are 0,1,2, …, n. The value probability is:
wherein,
p (X ═ k) is the probability of reaching k vehicles within one signal period;
λ is the average arrival rate of the vehicles per cycle, and k can be 0,1,2 …, n.
Since the arrival of vehicles follows a poisson distribution, the time intervals generated by the vehicles follow a negative exponential distribution. The probability that the time interval between the arrival of the vehicles is at least t seconds is:
P(h≥t)＝e -λt
after the arrival interval is calculated, the motor vehicles can be controlled to enter the simulated road network according to the arrival time of the Poisson distribution.
1.5, a road network module. The road network is mainly composed of road sections and nodes. A road segment is a collection of one or more lanes in the same direction of travel, while the number of lanes in a road segment does not change. The road sections are connected by nodes, and a plurality of connectors are possible in one node. The road segment attributes include{ coding link id Type, lane, front link be Posterior segment link af Center line coordinate xy cen Line coordinates xy si }；
Wherein the lane attribute comprises { code lane id Adjacent left lane le Adjacent right lane ri Attribute of lane dividing line, center line coordinate xy, side line coordinate xy si }。
1.6, a vehicle control module. Updating the vehicle state in real time according to the parameters calculated by the driving behavior model, wherein the parameters comprise speed v, acceleration a, direction angle gamma and vehicle type T p Global abscissa G x Global ordinate G y Matching the vehicles in the road network, namely updating the following states { the Lane where the vehicles are located } id Link of the road section id Global abscissa G x Global ordinate G y Mapping lane center line position IN y }。
2. Establishing data-driven driving behavior model
2.1 model input
And selecting the position as a representative index of the driving characteristics of the vehicle, and selecting the continuous Li/s historical track as input. Assuming that the vehicle only interacts with the vehicle in the detection area, selecting the historical track of the interactive vehicle lasting Li/s as input to indicate interactive characteristics; here, a rectangular range surrounded by the lane where the vehicle is located and the adjacent left and right lanes is defined as a detection area, and as shown in fig. 3, the nearest vehicles in six directions in the detection area are interactive vehicles, specifically, vehicles at left front, right front, left rear, and right rear positions. And finally, calculating the position of the current speed lasting for n/s, and taking the coordinate positions mapped to the lane lines on the two sides as input to indicate the influence of the lane lines. To sum up, the inputs at each time are:
wherein x is t Is the input of the model time point,
is a two-dimensional coordinate position, t indicates time, ind indicates different vehicles, and i is a time interval;
the model will last L i The historical track with delta t/s and time interval delta t is used as input, and then the model inputs X t Comprises the following steps:
2.2 model Structure
Selecting a Long Short-Term Memory (LSTM) network model as a basic structure, wherein each unit of the LSTM network is a Memory block with three self-parameterized control gates: input gate, forget gate and output gate, the model adds or removes information to the cell state through the fine structure of the three "gates".
In order to reduce the conventional violation results in practical application, the invention adds position and speed constraints into the model design, proposes the SHRC-LSTM model, divides the training set into two data sets D1 and D2 as shown in FIG. 4, and firstly inputs the data set D1 to train the model; inputting a data set D2 for prediction after training is finished, sequentially checking and correcting inconsistent prediction results according to three constraints, and calling a corrected data set D2 as a data set D3; finally, the data sets D1 and D3 are integrated as a training set of models, and the training of the models is continued.
The three conditions that need to be corrected are: the maximum speed is not exceeded, the distance between the vehicle and the surrounding vehicle is not less than the minimum distance, and the vehicle does not back:
in the formula:
Δv y is the maximum speed;
Δ t is the prediction time interval;
The main structure of the model adopts a double-layer LSTM model, the number of neurons in a hidden layer is 250, the Mean Square Error (MSE) is selected as a loss function, the activation function adopts tanh, the optimizer adopts Adam, the Batch size (Batch size) is selected to be 32, and the setting period (Epoch) is selected to be 50.
2.3 model output
The model outputs the predicted future position of the vehicle at time intervals Δ t as follows:
3. establishing customized interface to realize model access
The information to be transferred from the general kernel to the driving behavior model specifically includes: and 4s lane boundary coordinates are predicted for the current vehicle speed by the history track of the vehicle with the duration of 8s and the history track of 6 interactive vehicles.
Note that all vehicle information is transferred after integration, i.e., only once per simulation step.
The information to be transferred from the driving behavior model to the general kernel specifically includes: corrected model predicted trajectory
The three correction conditions are respectively: maximum speed, minimum spacing, no backing:
wherein:
Δv y is the maximum speed;
Δ t is the prediction time interval;
4. simulation system universal framework and control logic
As shown in fig. 2, the simulation starts with initializing the road network and vehicles, and building a simulation environment; the outer-layer cycle is simulation time, the inner-layer cycle is carried out on all vehicles at each moment, the motion state of each vehicle is calculated, and the vehicles are updated into a road network after safety constraint check. And when the calculation of the last vehicle in the road network is finished, adding 1 to the simulation step length, and continuing the simulation. The sign of the vehicle deleting from the road network is the arrival of the trip end point, and the sign of the simulation end is the arrival of the simulation end time.
The specific process is as follows:
s1: initializing a road network according to the input module and the road network module;
s2: updating the road network state through a road network model;
s3: generating vehicles through a vehicle generation module according to the road network state;
s4: calculating the vehicle state;
s5: generating track coordinates of vehicles according to the vehicle states, extracting surrounding vehicles and roads according to the road network states to obtain a vehicle time sequence data set, and integrating state information;
s6: predicting the vehicle track according to the integrated state information through a data-driven driving behavior model;
s7: adding constraints to the vehicle trajectory prediction result;
s8: updating the vehicle state;
s9: judging whether the vehicle reaches a trip destination, if so, deleting the vehicle, and if not, executing the step S10;
s10: generating a track record of the vehicle;
s11: judging whether the vehicle is the last vehicle at the current moment, if so, executing the step S12, otherwise, returning to the step S4;
s12: and judging whether the preset simulation termination time is reached, if so, finishing the simulation, and otherwise, returning to the step S2.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (10)
1. A driving behavior simulation system supporting data driving is characterized by comprising a general kernel, a data-driven driving behavior model and a customized interface, wherein the customized interface is used for connecting the general kernel and the data-driven driving behavior model for data transmission;
the general kernel is used for constructing a traffic scene and generating a vehicle state and an environment state;
the data-driven driving behavior model is used for predicting the vehicle state of the next time step according to the time sequence data of the vehicle;
the customized interface is used for integrating the vehicle state and the environment state, constructing time sequence data and transmitting the time sequence data to a data-driven driving behavior model; and (4) applying constraint to the predicted vehicle state of the next time step to obtain a re-revised vehicle state, and feeding back the re-revised vehicle state to the general kernel to update the traffic scene.
2. A system for supporting data-driven driving behavior simulation according to claim 1, wherein the generic kernel comprises:
the input module is used for initializing a road network and vehicles and analyzing road network information;
the output module is used for outputting simulation statistical parameters and analysis results and deriving a vehicle motion track;
the basic setting module is used for configuring corresponding basic facilities for different road types;
the traffic control module is used for configuring a control scheme of the signal lamp;
the vehicle generation module is used for setting the entering position of the vehicle, the initial state of the vehicle and the entering interval of the vehicle according to the initialization information of the input module and generating the vehicle state according to the data of the traffic control module;
the road network module is used for constructing a road network and generating an environmental state according to the information of the basic setting module and the traffic control module and based on a preset road network format;
and the vehicle control module is used for updating the vehicle state in real time according to the revised vehicle state and feeding the vehicle state back to the road network module and the output module.
3. The system of claim 2, wherein the road network module constructs the road network according to an OpenDRIVE road network format, the road network format comprising road segments and nodes, the road segments being a set of one or more lanes in the same driving direction, the road segments being connected by the nodes, and one or more connectors being provided in one of the nodes.
4. The simulation system for supporting data-driven driving behavior according to claim 1, wherein the input of the data-driven driving behavior model comprises continuous running characteristics of the vehicle, interactive characteristics with surrounding vehicles, lane line information and traffic rules;
the model structure of the data-driven driving behavior model is a pre-constructed and trained SHRC-LSTM model;
the output of the data-driven driving behavior model is a two-dimensional position coordinate.
5. A system for supporting data-driven driving behavior simulation as recited in claim 4, wherein the training process of the SHRC-LSTM model comprises:
the method comprises the steps of obtaining a training set of driving behaviors, dividing the training set into a data set D1 and a data set D2, firstly training a long-term memory network model by adopting the data set D1, then inputting the data set D2 for prediction, checking and correcting inconsistent prediction results according to preset safety constraints, taking the corrected data set D2 as the data set D3, and integrating the data set D1 and the data set D3 to continuously train the long-term memory network model.
6. A support data driven driving behavior simulation system according to claim 5, characterized in that the SHRC-LSTM model is a double-layer LSTM model.
7. A data driven driving behaviour simulation system according to claim 5, wherein said safety constraints include not exceeding a maximum speed, not being spaced from the surrounding vehicles by less than a minimum spacing and not backing up.
8. A system for supporting data-driven driving behavior simulation according to claim 7, wherein the computational expression of the safety constraints is:
in the formula,
as predicted coordinates at time i, Y i As true coordinates of time i, Δ v y At maximum speed, at is the prediction time interval,
is the coordinate of the nearest vehicle in the right direction, Y l i The coordinates of the nearest vehicle in the left direction,
is the minimum inter-vehicle distance in the right direction,
is the minimum inter-vehicle distance in the left direction,
to predict the speed.
9. The system of claim 1, wherein the constraints imposed by the customization interface on the predicted vehicle state for the next time step include a maximum speed constraint, a minimum separation constraint, and a no reverse constraint.
10. A simulation method supporting a data-driven driving behavior simulation system according to claim 2, characterized by comprising the steps of:
s1: initializing a road network according to the input module and the road network module;
s2: updating the road network state through a road network model;
s3: generating vehicles through a vehicle generation module according to the road network state;
s4: calculating the vehicle state;
s5: generating track coordinates of vehicles according to the vehicle states, extracting surrounding vehicles and roads according to the road network states to obtain a vehicle time sequence data set, and integrating state information;
s6: predicting the vehicle track according to the integrated state information through a data-driven driving behavior model;
s7: adding constraints to the vehicle trajectory prediction result;
s8: updating the vehicle state;
s9: judging whether the vehicle reaches a trip destination, if so, deleting the vehicle, and if not, executing the step S10;
s10: generating a track record of the vehicle;
s11: judging whether the vehicle is the last vehicle at the current moment, if so, executing the step S12, otherwise, returning to the step S4;
s12: and judging whether the preset simulation termination time is reached, if so, finishing the simulation, and otherwise, returning to the step S2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210540780.3A CN115081186B (en) | 2022-05-17 | 2022-05-17 | Driving behavior simulation system supporting data driving and simulation method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210540780.3A CN115081186B (en) | 2022-05-17 | 2022-05-17 | Driving behavior simulation system supporting data driving and simulation method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115081186A true CN115081186A (en) | 2022-09-20 |
| CN115081186B CN115081186B (en) | 2023-09-26 |
Family
ID=83250136
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210540780.3A Active CN115081186B (en) | 2022-05-17 | 2022-05-17 | Driving behavior simulation system supporting data driving and simulation method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115081186B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2590151A1 (en) * | 2011-11-03 | 2013-05-08 | Deutsche Telekom AG | A framework for the systematic study of vehicular mobility and the analysis of city dynamics using public web cameras |
| CN103345548A (en) * | 2013-06-27 | 2013-10-09 | 同济大学 | Integrated network experiment platform for integrating microcosmic traffic simulator with multiple driving simulators |
| CN110427682A (en) * | 2019-07-26 | 2019-11-08 | 清华大学 | A kind of traffic scene simulation experiment platform and method based on virtual reality |
| CN111680377A (en) * | 2020-06-11 | 2020-09-18 | 杭州海康威视数字技术股份有限公司 | Traffic situation simulation method and system and electronic equipment |
| CN112417756A (en) * | 2020-11-13 | 2021-02-26 | 清华大学苏州汽车研究院（吴江） | Interactive simulation test system of automatic driving algorithm |
| WO2021077725A1 (en) * | 2019-10-21 | 2021-04-29 | 南京航空航天大学 | System and method for predicting motion state of surrounding vehicle based on driving intention |
| CN113223293A (en) * | 2021-05-06 | 2021-08-06 | 杭州海康威视数字技术股份有限公司 | Road network simulation model construction method and device and electronic equipment |
| CN113643542A (en) * | 2021-10-13 | 2021-11-12 | 北京理工大学 | Vehicle track prediction method and system under multi-vehicle interaction environment based on ensemble learning |
| CN114021330A (en) * | 2021-10-28 | 2022-02-08 | 武汉中海庭数据技术有限公司 | Simulated traffic scene building method and system and intelligent vehicle control method |
| WO2022082476A1 (en) * | 2020-10-21 | 2022-04-28 | 华为技术有限公司 | Simulated traffic scene file generation method and apparatus |
-
2022
- 2022-05-17 CN CN202210540780.3A patent/CN115081186B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2590151A1 (en) * | 2011-11-03 | 2013-05-08 | Deutsche Telekom AG | A framework for the systematic study of vehicular mobility and the analysis of city dynamics using public web cameras |
| CN103345548A (en) * | 2013-06-27 | 2013-10-09 | 同济大学 | Integrated network experiment platform for integrating microcosmic traffic simulator with multiple driving simulators |
| CN110427682A (en) * | 2019-07-26 | 2019-11-08 | 清华大学 | A kind of traffic scene simulation experiment platform and method based on virtual reality |
| WO2021077725A1 (en) * | 2019-10-21 | 2021-04-29 | 南京航空航天大学 | System and method for predicting motion state of surrounding vehicle based on driving intention |
| CN111680377A (en) * | 2020-06-11 | 2020-09-18 | 杭州海康威视数字技术股份有限公司 | Traffic situation simulation method and system and electronic equipment |
| WO2022082476A1 (en) * | 2020-10-21 | 2022-04-28 | 华为技术有限公司 | Simulated traffic scene file generation method and apparatus |
| CN112417756A (en) * | 2020-11-13 | 2021-02-26 | 清华大学苏州汽车研究院（吴江） | Interactive simulation test system of automatic driving algorithm |
| CN113223293A (en) * | 2021-05-06 | 2021-08-06 | 杭州海康威视数字技术股份有限公司 | Road network simulation model construction method and device and electronic equipment |
| CN113643542A (en) * | 2021-10-13 | 2021-11-12 | 北京理工大学 | Vehicle track prediction method and system under multi-vehicle interaction environment based on ensemble learning |
| CN114021330A (en) * | 2021-10-28 | 2022-02-08 | 武汉中海庭数据技术有限公司 | Simulated traffic scene building method and system and intelligent vehicle control method |
Non-Patent Citations (3)
| Title |
|---|
| 王祺;胡坚明;慕春棣;王易之;: "一种车路协同系统微观仿真平台的实现", 公路交通科技, no. 04 * |
| 袁娜;史昕;赵祥模;: "基于改进人工鱼群算法的车辆轨迹规划方法", 计算机应用, no. 10 * |
| 马寿峰, 贺国光, 刘豹: "一种通用的城市道路交通流微观仿真系统的研究", 系统工程学报, no. 04 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115081186B (en) | 2023-09-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112216108B (en) | Traffic prediction method based on attribute-enhanced space-time graph convolution model | |
| CN112215337B (en) | Vehicle track prediction method based on environment attention neural network model | |
| JP7075366B2 (en) | Methods, devices, equipment and media for classifying driving scene data | |
| CN103593535B (en) | Urban traffic complex self-adaptive network parallel simulation system and method based on multi-scale integration | |
| CN112487954B (en) | Pedestrian crossing behavior prediction method for plane intersection | |
| US12037027B2 (en) | Systems and methods for generating synthetic motion predictions | |
| CN113643528B (en) | Signal lamp control method, model training method, system, device and storage medium | |
| Ghodsi et al. | Generating and characterizing scenarios for safety testing of autonomous vehicles | |
| CN114638148A (en) | Safe and extensible model for culture-sensitive driving of automated vehicles | |
| CN114170789B (en) | Intelligent network link lane change decision modeling method based on space-time diagram neural network | |
| Sun et al. | Vehicle turning behavior modeling at conflicting areas of mixed-flow intersections based on deep learning | |
| CN114495486B (en) | Microcosmic traffic flow prediction system and microcosmic traffic flow prediction method based on hierarchical reinforcement learning | |
| WO2023082726A1 (en) | Lane changing strategy generation method and apparatus, computer storage medium, and electronic device | |
| CN117521389A (en) | Vehicle perception test method based on vehicle-road collaborative perception simulation platform | |
| CN113642682B (en) | Trajectory primitive extraction and analysis method and system under multi-vehicle interaction environment | |
| Bartin | Use of learning classifier systems in microscopic toll plaza simulation models | |
| WO2021258847A1 (en) | Driving decision-making method, device, and chip | |
| CN114372627A (en) | Urban vehicle travel time estimation method based on hybrid deep learning framework | |
| CN112947466A (en) | Parallel planning method and equipment for automatic driving and storage medium | |
| Lu et al. | DeepQTest: Testing Autonomous Driving Systems with Reinforcement Learning and Real-world Weather Data | |
| CN116580549A (en) | Dynamic three-dimensional traffic simulation scene collaborative operation method | |
| CN115081186B (en) | Driving behavior simulation system supporting data driving and simulation method thereof | |
| CN116080681A (en) | Zhou Chehang identification and track prediction method based on cyclic convolutional neural network | |
| CN115981302A (en) | Vehicle following lane change behavior decision-making method and device and electronic equipment | |
| Chen et al. | Platoon separation strategy optimization method based on deep cognition of a driver’s behavior at signalized intersections |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |