Method and system for generating and verifying digital twin scenarios of autonomous vehicles based on logical scenarios
By transforming real-world scenarios into meta-scenarios and generating twin scenarios, and utilizing formal modeling and time automata verification, the challenge of assessing the expected functional safety in virtual testing of autonomous vehicles is solved, improving assessment efficiency and accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- EAST CHINA NORMAL UNIV
- Filing Date
- 2023-06-05
- Publication Date
- 2026-06-19
AI Technical Summary
Existing technologies struggle to effectively assess and verify the expected functional safety of highly automated vehicles in complex scenarios, especially in virtual testing where it is difficult to simulate and compare the interaction between the vehicle and other virtual traffic participants.
The method uses OpenDRIVE and OpenSCENARIO standards to convert real-world scenes into .xodr and .xosc files, extracts attributes and parameters from the meta-scene, generates twin scenarios, and verifies the effectiveness of the twin scenarios through formal modeling and time automata.
It enables the assessment of the expected functional safety of autonomous vehicles in complex scenarios, improves the efficiency and accuracy of virtual testing, and ensures the responsible and ethical compliance of artificial intelligence.
Smart Images
Figure CN117313305B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of expected functional safety assessment in the context of expected functional safety, and in particular to a method and system for generating and verifying digital twin scenarios for autonomous vehicles based on logical scenarios. Background Technology
[0002] The research and application of artificial intelligence (AI) has accelerated the understanding and realization of highly automated vehicles (HAVs). However, the uncertainty of AI algorithms also brings safety risks. Currently, there is still a serious lack of concrete and effective methods to address these challenges. Therefore, Expected Functional Safety (SOTIF) aims to supplement Functional Safety (FuSa) and Information Security to address the challenges posed by the uncertainty of vehicle AI. Responsible and ethical AI helps reduce the risk of unintended dangers and is an effective way to ensure expected functional safety. The unreplicable nature of real-world scenarios is one of the well-known limitations of physical testing, but now automakers can use digital twin technology based on real-time data to make it possible to reproduce test scenarios. In virtual testing, the closed-loop interactions between the HAV and other virtual traffic participants make it difficult to directly compare experimental results with simulated evidence, as sudden behavior of the HAV can amplify subtle differences between environments. Therefore, scenario-based testing is considered the most advanced testing method for understanding and validating HAVs. Summary of the Invention
[0003] To address the shortcomings of existing technologies, the present invention proposes a method for generating and verifying digital twin scenarios for autonomous vehicles based on logical scenarios, specifically comprising the following four steps:
[0004] Step 1: Converting from the actual scene to the meta scene (MSA). First, use XML language based on the OpenDRIVE and OpenSCENARIO standards to convert the original real scene into .xodr and .xosc files. Then, extract the attributes and parameters from the meta scene based on the .xodr and .xosc files.
[0005] Step 2: Extract one or more meta-scenes (MSEs) from the meta-scene. A meta-scene is a slice of a meta-scene at a specific point in time. The specific time point refers to the point in time when the vehicle's driving state changes during operation. Driving states include lane changing, braking, acceleration, deceleration, etc. For example, when an accident occurs...When unexpected traffic participants suddenly appear, vehicles encounter small, slippery sections of road, or the lead vehicle in a convoy reaches a fork in the road, their driving status changes compared to a previous period. The point in time when these changes occur is considered a specific time point. A meta-scenario contains truck convoy information, environmental information, road information, and traffic participant information at a fixed time point. Extracting the meta-scenario prepares for the next step of generating a digital twin scenario.
[0006] Step 3: Digital twin scene generation based on meta-scenarios. Based on the meta-scenarios extracted from the meta-scenarios, specific meta-scenarios that change the driving state are transformed and generated into twin scenarios; the twin meta-scenarios and constraint variables are combined to form unverified twin scenarios.
[0007] In this invention, the constraint variables refer to the settings that restrict vehicles in a scenario, including traffic rules that vehicles need to follow in the scenario, specifically including speed limits, traffic direction, road markings, etc.
[0008] Step Four: Twin Scenario (TS) Validation. When generating virtual test twin scenarios for the automated truck platoon based on meta-scenarios and meta-contexts, the generated twin scenarios consist of twin meta-contexts. A scenario is the dynamic change process of the research object over a period of time, while a context is a specific moment in the scenario (the time when the driving state changes). Each context evolves over time, forming a new context. Therefore, when validating the generated twin scenarios, the twin scenario is constructed as a time automaton considering conditional constraints, and the twin scenario is abstracted into a meta-context as a state. The validity of the generated twin scenario is verified by judging the satisfiability of the time constraints and conditional constraints of the constructed time automaton.
[0009] In Step 1, the meta-scene is the real-world scenario used, and the meta-context is a slice of the meta-scene at a specific point in time; the meta-scene is typically used to expand upon the original real-world scenario. A meta-scene is usually a driving process that lasts for a period of time; it is dynamic and continuous. Therefore, the dynamic process of the meta-scene can be described by creating several key meta-contexts at specific points in time. A meta-scene contains primary vehicle information, the start and end coordinates of the basic trajectory, and matrix information consisting of n meta-context segments. Meta-context segments are composed of tuples, and a meta-scene (MSA) can be represented as MSA = {platoons, truck, n, scoord, ecoord, [nMSE]}, where platoons represent information about the entire truck platoon, truck represents information about the autonomous truck, n represents the number of meta-context segments, scoord and ecoord represent the start and end of the meta-context (scoord is the first selected meta-context, and ecoord is the last selected meta-context), and [nMSE] represents the MSE matrix. The MSE matrix is a formalized method, containing information from each MSE multiplied by n MSEs. Specifically, the interaction model between cars and other traffic participants, like other vehicles, is not in the meta-scene, but in the meta-context.
[0010] In step one, the conversion from the actual scene to the meta-scene is based on OpenDRIVE and OpenSCENARIO. ASAMOpenDRIVE provides the foundation for describing road networks using Extensible Markup Language (XML), while the OpenSCENARIO format allows for the creation of detailed scenes that can be used to test and validate autonomous driving algorithms and systems. In the meta-scene conversion algorithm, the conversion process from the original real-world scene to the meta-scene depends on the OpenDRIVE and OpenSCENARIO standards.
[0011] The attributes and parameters in the meta-scene mentioned in Step 1 include vehicle information, the start and end coordinates of the basic trajectory, the matrix information composed of n meta-scene fragments, and the specific parameters corresponding to the above attributes.
[0012] In step two, the meta-scenario is the state of a meta-scene at a key specific point in time. Each meta-scenario is a static scene description. Therefore, a meta-scenario includes road information, environmental information, truck queue information, and third-party traffic participant information at a static moment. Truck queue information in a meta-scenario mainly includes the queue state, the self-driving truck state (set speed and / or throttle state, vehicle driver and / or passenger state), etc., where queue state includes queue leader, queue length, queue space, etc. Road information includes lane information and road infrastructure information. Lane information includes basic road information such as curvature, slope, altitude, road friction, number of lanes, etc. Road infrastructure information includes traffic lights, roadside green belts, traffic signs, streetlights, etc. Environmental information includes two factors: weather and illumination. Weather is represented by wind, clouds, rain, fog, temperature, etc., and illumination is represented by light intensity, angle, and light source, etc. Information on third-party traffic participants includes expected traffic participants, such as pedestrians, motor vehicles, and non-motorized vehicles on the road. Furthermore, this invention also proposes an unexpected traffic participant. Unexpected traffic participants include those who suddenly enter the subject vehicle from an angle or whose actions are disrupted by factors affecting the main vehicle's normal driving, such as a rock suddenly falling from the sky and hitting the windshield, or an animal suddenly rushing in front of the main vehicle. Meta-situations are defined through formal modeling based on the description of the meta-situation.
[0013] Formal modeling is as follows:
[0014] In step two, define It is an n-ary predicate, (t1, t2, t3, t4) is a term, and this invention uses atomic formulas. (t1, t2, t3, t4) represents the meta-scenario. In this atomic formula, each term represents an element that constitutes the meta-scenario, where t1 represents queue information, t2 represents road information, t3 represents environmental information, and t4 represents information about traffic participants, corresponding to... Figure 2 The four branches.
[0015] In step two, according to the definition of a meta-scenario, each meta-scenario can be represented as an atomic formula. Therefore, extracting meta-scenarios from meta-scenarios is equivalent to extracting atomic formulas from pre-constructed formulas. In the formal model, a meta-scenario is divided into i time points, which means that the formula of a meta-scenario contains i atomic formulas: There is no specific action. A∧A: For example, in an overtaking scenario, there are acceleration and steering, so there are two MSEs, one for acceleration and one for steering. A∨A: In a scenario, there might be a choice between two options, A→A: one action leads to the other. In this formula, is a constraint variable, where the value of j ranges from [0, m], and m is the upper limit value set in the constraint variable. `j` is a free variable, where the value of `j` ranges from (a, b], where `b` > `a`. The values of `a` and `b` are determined according to the actual meaning and corresponding requirements of the variables. Here's an example to illustrate the values of constrained and free variables: For instance, if the speed limit on a highway is 60 to 120 km / h, then if the second `j` represents speed, then `a` and `b` would be 60 and 120 respectively. Similarly, if the first `j` represents acceleration, then it would be a number starting from 0 and reaching a specified value (since reversing is not allowed on a highway, acceleration is always greater than or equal to 0). These two variables mean that all variable values in the meta-scene (such as speed, acceleration, distance between vehicles, road friction coefficient, etc.) can be completely represented by these two variables; no variable value is unrepresentable.
[0016] In step three, a twin scenario is constructed based on the term substitution theorem. The term substitution theorem states that in an algebraic expression, identical terms can be replaced by the algebraic sum of their sums, differences, or products. It is generally used to simplify and optimize complex algebraic expressions. The meta-scenario L is a first-order language, I is the implementation of L, and A(x) i )∈F(L) is a formula, x i It is A(x) i Let t be a free variable of x. i Let v be a free variable, v' be the task of L in I, v' be the equivalent task of v, and v'(x) i ) = v(t).
[0017] In step three, the key characteristic of the term substitution theorem is that its intrinsic properties remain unchanged, regardless of whether the scenarios represent simple or complex variables. In other words, no matter how the free variables of the specific scenario change, the integrity of the scenario remains unaffected. Therefore, the twin scenarios generated by changing the free variables based on the term substitution theorem are theoretically complete.
[0018] In step four, let k be the transition in the timed automaton TA. If the time constraint of k can never be valid, then the time constraint of k is said to be unsatisfied. Otherwise, the time constraint of k can be satisfied. The generated twin scenario is valid only when the time constraint of k is satisfied. This twin scenario is constructed as a transition sequence of a timed automaton, that is, assuming that the path starts from the initial time k0 and passes through k... i ,…,k j Migrate to k, <k0,k t ,…,k j ,k> is a migration sequence, so <k0,k t ,…,k j ,k> can represent a twin scenario. Assume k i and k j It is a transition sequence <k0,k1,…,kn The two transitions in >, if k i and k j Is it adjacent or transition k? i or k j If a migration from k0 to k is a time-dependent migration where the clock value in the time constraint is set to 0 at the end of the migration sequence, then k0 to k is called a time-dependent migration; otherwise, k0 to k is called a non-time-dependent migration.
[0019] In step four, a complete scene consists of timed automata, a network of timed automata on a common clock and action set, and a complete scene consists of n timed automata.
[0020] In step four, the transition relationships between states constitute the process of traffic changes in the scene, including changes in the positions of traffic participants, changes in weather, changes in traffic rules, etc.
[0021] The present invention also provides a system for implementing the above method, the system comprising a meta-scene conversion module, a meta-scene extraction module, a twin scene generation module, and a twin scene verification module.
[0022] The beneficial effects of this invention are as follows: This invention designs a method for assessing the expected functional safety of autonomous vehicles based on the construction of digital twin scenarios. The main design feature of this invention is that meta-scenarios can be formed by transforming real-world scenarios and extracting key parameter clusters based on meta-scenarios (the parameter clusters contain some parameters needed for meta-scenario extraction). The meta-scenarios are redefined for end-point evaluation to generate numerous twin scenarios for virtual testing of autonomous vehicles. It addresses the requirements of numerous scenarios in virtual testing of autonomous vehicles and ensures that the artificial intelligence in autonomous vehicles is responsible and ethical, thereby improving efficiency and hazard detection in the expected functional safety assessment. Attached Figure Description
[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0024] Figure 1 This is a flowchart of the safety assessment for achieving the intended function of this invention.
[0025] Figure 2 It is a general framework that combines digital twin vehicles and digital twin scenarios.
[0026] Figure 3 This is a schematic diagram illustrating the idea behind generating test scenarios in an embodiment of the present invention. Detailed Implementation
[0027] The invention will be further described in detail below with reference to the specific embodiments and accompanying drawings. Except for the contents specifically mentioned below, the processes, conditions, and experimental methods for implementing the invention are all common knowledge and general knowledge in the art, and the invention does not have any particular limitations.
[0028] Figure 1 As shown, a method for generating and verifying digital twin scenarios for autonomous vehicles based on logical scenarios consists of four steps: Step 1: Converting the actual scenario to a meta-scenario (MSA). First, the original real scenario is converted into .xodr and .xosc files using XML language based on the OpenDRIVE and OpenSCENARIO standards. Then, attributes and parameters are extracted from the meta-scenario based on the .xodr and .xosc files. Step 2: Extracting meta-contexts (MSEs) from the meta-scenario. A meta-context is a slice of a meta-scenario at a specific point in time. The meta-context contains truck queue information, environmental information, road information, and traffic participant information at a fixed point in time. Extracting the meta-context prepares for the next step of generating digital twin scenarios. Step 3: Generating digital twin scenarios based on meta-contexts. Based on the meta-contexts extracted from the meta-scenario, specific meta-contexts are converted and generated into twin scenarios. The twin meta-contexts and constraint variables are combined to form an unverified twin scenario. Step 4: Twin scenario verification. When generating a virtual test twin scenario for an autonomous truck queue based on the meta-scenario and meta-contexts, the generated twin scenario consists of twin meta-contexts. A scenario is the dynamic process of a research object changing over a period of time, while a situation is a specific moment within a scenario. Each situation evolves over time to form a new situation. Therefore, when validating the generated twin scenarios, the twin scenario is constructed as a time automaton considering conditional constraints, and the twin scenario is abstracted as a meta-situation as a state. The validity of the generated twin scenario is verified by judging the satisfiability of the time constraints and conditional constraints of the constructed time automaton.
[0029] Figure 1As shown, in step one, the meta-scene is typically the original real-world scenario used for expansion. A meta-scene is usually a driving process that lasts for a period of time; it is dynamic and continuous. Therefore, the dynamic process of the meta-scene is described by creating several key time points. The meta-scene contains primary vehicle information, the start and end coordinates of the basic trajectory, and matrix information consisting of n meta-scene fragments. Meta-scene fragments are composed of tuples, and the meta-scene (MSA) is represented as MSA = {platoons, truck, n, scoord, ecoord, [nMSE]}, where platoons represents information about the entire truck platoon, truck represents information about the autonomous truck, n represents the number of meta-scene fragments, scoord and ecoord represent the start and end of the meta-scene, and [nMSE] represents the MSE metric. Specifically, the behavior models of trucks and other traffic participants, like other vehicles, are not in the meta-scene itself, but are in the meta-scene.
[0030] Figure 1 As shown, in step one, the conversion from the actual scene to the meta-scene is based on OpenDRIVE and OpenSCENARIO. OpenDRIVE provides the foundation for describing road networks using Extensible Markup Language (XML), while the OpenSCENARIO format allows for the creation of detailed scenes that can be used to test and validate autonomous driving algorithms and systems. In the meta-scene conversion algorithm, the process of converting from the original real-world scene to the meta-scene depends on the OpenDRIVE and OpenSCENARIO standards.
[0031] Figure 1As shown, in step two, the meta-scenario is the state of the meta-scene at key time points. Each meta-scenario is a static scene description. Therefore, a meta-scenario includes road information, environmental information, truck queue information, and third-party traffic participant information at static moments. Truck queue information in a meta-scenario mainly includes the queue status, self-driving truck status, set speed or throttle status, and the status of vehicle drivers and passengers. Road information includes lanes and road infrastructure. Lanes include basic road information such as curvature, slope, altitude, road friction, and number of lanes. Road infrastructure includes traffic lights, roadside green belts, traffic signs, and streetlights. Environmental information includes two factors: weather and lighting. Weather is represented by wind, clouds, rain, fog, and temperature, while lighting is represented by light intensity, angle, and light source. Information on third-party traffic participants includes expected traffic participants, such as pedestrians, motor vehicles, and non-motorized vehicles on the road. Additionally, an unexpected traffic participant is also considered. Such scenarios involve traffic participants suddenly entering the subject vehicle's field of vision or the sudden appearance of factors affecting the main vehicle's normal driving, such as a rock suddenly falling from the sky and hitting the windshield, or an animal suddenly rushing in front of the main vehicle. Meta-scenarios are defined through formal modeling based on the description of the meta-scenario.
[0032] Figure 1 As shown, in step two, the definition is... It is an n-ary predicate, (t1, t2, t3, t4) is a term, and this invention uses atomic formulas. (t1, t2, t3, t4) represents the meta-scenario. In this atomic formula, the terms represent the elements that constitute the meta-scenario, where t1 represents queue information, t2 represents road information, t3 represents environmental information, and t4 represents traffic participant information.
[0033] Figure 1 As shown, in step two, according to the definition of a meta-scenario, each meta-scenario can be represented as an atomic formula. Therefore, extracting meta-scenarios from meta-scenarios is equivalent to extracting atomic formulas from pre-constructed formulas. In the formal model, a meta-scenario is divided into i time points, which means that the formula of a meta-scenario contains i atomic formulas: In this formula, is a constrained variable, where the value of j ranges from [0, m], while j is a free variable, where the value of j ranges from (a, b], and b > a.
[0034] Figure 1 As shown in the diagram, in step three, a twin scenario is constructed based on the term substitution theorem. The meta-scenario L is a first-order language, I is the implementation of L, and A(x) i )∈F(L) is a formula, x i It is A(x)i Let t be a free variable of x. i Let v be a free variable, v' be the task of L in I, v' be the equivalent task of v, and v'(x) i ) = v(t).
[0035] Figure 1 As shown in step three, the characteristic of the term substitution theorem is that its intrinsic properties remain unchanged, regardless of whether the scenario represents a simple or complex variable. In other words, no matter how the free variables of the specific scenario change, the integrity of the scenario remains unaffected. The twin scenarios generated by changing the free variables based on the term substitution theorem are theoretically complete.
[0036] Figure 1 As shown, in step four, let k be the transition in the timed automaton TA. If the time constraint of k can never be valid, then the time constraint of k is said to be unsatisfied. Otherwise, the time constraint of k can be satisfied. The generated twin scenario is valid only when the time constraint of k is satisfied. This twin scenario is constructed as a transition sequence of a timed automaton, that is, assuming that the path starts from the initial time k0 and passes through k... i ,…,k j Migrate to k, <k0,k i ,…,k j ,k> is a migration sequence, so <k0,k i ,…,k j ,k> can represent a twin scenario. Assume k i and k j It is a transition sequence <k0,k1,…,k n The two transitions in >, if k i and k j Is it adjacent or transition k? i or k j If a migration is at the end of the migration sequence and the clock value in the time constraint is set to 0, then these two migrations are called time-related migrations; otherwise, they are called non-time-related migrations.
[0037] Figure 1 As shown, in step four, a complete scenario consists of timing automata, a network of timing automata on a common clock and action set, consisting of n timing automata.
[0038] Figure 1 As shown in step four, the transition relationships between states constitute the process of traffic changes in the scene, including changes in the positions of traffic participants, changes in weather, changes in traffic rules, etc.
[0039] Example
[0040] This embodiment provides an example illustrating the approach to generating test scenarios, such as... Figure 3 :
[0041] Left 1: Meta-scene; 1-2: Dividing the meta-scene into meta-contexts, comprising three slices; 2-3: The digital twin process. In this embodiment, differences from the meta-scene include lane positions, vehicle positions, etc. Firstly, this scenario has four 't's: queue information, road information, environmental information, and traffic participant information. Then, digital twinning is performed according to the method mentioned in this invention to obtain a new test scenario. The three scenarios in the lower right figure are all test scenarios generated by digital twins; the formal modeling process is the verification process of the scenarios.
[0042] The scope of protection of this invention is not limited to the above embodiments. Any variations and advantages that can be conceived by those skilled in the art without departing from the spirit and scope of the inventive concept are included in this invention and are protected by the appended claims.
Claims
1. A method for generating and verifying digital twin scenarios for autonomous vehicles based on logical scenarios, characterized in that, Includes the following steps: Step 1: Transform from real-world scene to meta-scene. First, use XML language based on OpenDRIVE and OpenSCENARIO standards to convert the original real-world scene into .xodr and .xosc files. Then, extract the attributes and parameters from the meta-scene based on the .xodr and .xosc files. Step 2: Extract one or more meta-scenes from the meta-scene to prepare for generating a digital twin scene; the meta-scene is a slice at the time point of change of driving state in the meta-scene; the meta-scene contains truck queue information, environmental information, road information and traffic participant information at a fixed time point; Step 3: Digital twin scene generation based on meta-scenario. Based on the meta-scenario extracted from the meta-scenario, the meta-scenario that changes the driving state is transformed and generated into a twin scenario; the twin meta-scenario and constraint variables are combined to form an unverified twin scenario. Step 4: Twin scenario verification. Based on the meta-scenario and meta-context, a virtual test twin scenario for the automated truck queue is generated. The twin scenario is constructed as a time automaton that considers conditional constraints, and the twin scenario is abstracted as a meta-context as a state. The validity of the generated twin scenario is verified by judging the satisfiability of the time constraints and conditional constraints of the constructed time automaton.
2. The twin scene generation and verification method according to claim 1, characterized in that, In step one, the meta-scene is a duration-driven process, which describes the dynamic process of the meta-scene by creating meta-scenes at time points when the driving state changes. The meta-scene contains key vehicle information, the start and end coordinates of the basic trajectory, and a matrix of information consisting of n meta-scene fragments. Each meta-scene fragment is composed of tuples. The meta-scene is represented as MSA = {platoons, truck, n, scoord, ecoord, [nMSE]}, where platoons represents the information of the entire truck platoon, truck represents the information of the autonomous truck, n represents the number of meta-scene fragments, scoord and ecoord represent the start and end of the meta-scene, and [nMSE] represents the MSE matrix. The interaction model between trucks and other traffic participants is contained within the meta-scene.
3. The twin scene generation and verification method according to claim 1, characterized in that, In step one, the conversion from real-world scenarios to meta-scenarios is based on OpenDRIVE and OpenSCENARIO. ASAMOpenDRIVE provides the foundation for describing road networks using Extensible Markup Language; the OpenSCENARIO format allows the creation of detailed scenarios for testing and validating autonomous driving algorithms and systems.
4. The twin scene generation and verification method according to claim 1, characterized in that, In step two, the meta-scenario is the state of the meta-scene at the point in time when the driving state changes; each meta-scenario is a static scene description. Meta-contexts include road information, environmental information, truck queue information, and information about third-party traffic participants at static times; The truck queue information in the meta-scenario includes the queue status and the self-service truck status; the queue status includes the queue leader, queue length, and queue space; the self-service truck status includes the set speed and / or throttle status, and the status of the vehicle driver and / or passenger. The road information includes lane information and road infrastructure information; Lane information includes road curvature, slope, altitude, road friction, and number of lanes; Road infrastructure information includes traffic lights, road green belts, traffic signs, and street light information; Environmental information includes two factors: weather and light. Weather is represented by information including wind, clouds, rain, fog, and temperature, while light is represented by information including light intensity, angle, and light source. Information on third-party traffic participants includes expected traffic participants and an unexpected traffic participant; Unexpected traffic participants include those who suddenly enter the subject vehicle's angle or whose main vehicle is suddenly affected by certain factors that affect normal driving. Based on the description of the meta-scenario, the meta-scenario is defined through formal modeling.
5. The twin scene generation and verification method according to claim 4, characterized in that, In step two, the formal modeling is performed as follows: definition It is an n-ary predicate, and (t1, t2, t3, t4) is a term; Using atomic formulas The term represents the meta-scenario; in the atomic formula, the term represents the element constituting the meta-scenario, where t1 represents queue information, t2 represents road information, t3 represents environmental information, and t4 represents traffic participant information.
6. The twin scene generation and verification method according to claim 1, characterized in that, In step two, according to the definition of a meta-scenario, each meta-scenario is represented as an atomic formula; extracting meta-scenarios from meta-scenarios is equivalent to extracting atomic formulas from pre-constructed formulas; in the formal model, a meta-scenario is divided into i time points, and the formula of a meta-scenario contains i atomic formulas: In this formula, is a constraint variable, where the value of j ranges from [0, m], and m is the upper limit value set in the constraint variable. j is a free variable, where the value range is (a, b], b>a, and the values of a and b are determined according to the actual meaning of the variable and the corresponding requirements.
7. The twin scene generation and verification method according to claim 1, characterized in that, In step three, a twin scenario is constructed based on the term substitution theorem. The meta-scenario L is a first-order language, I is the implementation of L, and A(x) i )∈F(L) is a formula, x i It is A(x) i Let t be a free variable of x; i Let v be a free variable, v' be the task of L in I, v' be the equivalent task of v, and v'(x) i ) = v(t).
8. The twin scene generation and verification method according to claim 1, characterized in that, In step four, let k be the transition in the timed automaton TA. If the time constraint of k can never be valid, then the time constraint of k is said to be unsatisfied; otherwise, the time constraint of k is satisfied. The generated twin scenario is valid only when the time constraint of k is satisfied. The twin scenario is constructed as a transition sequence of a timed automaton, that is, assuming that the path starts from the initial time k0 and passes through k i ,…,k j Migrate to k, <k0,k i ,…,k j ,k> is a migration sequence, using <k0,k i ,…,k j ,k> represents a twin scenario; assuming k i and k j It is a transition sequence <k0,k1,…,k n The two transitions in >, if k i and k j Is it adjacent or transition k? i or k j If a migration from k0 to k is a time-dependent migration where the clock value in the time constraint is set to 0 at the end of the migration sequence, then k0 to k is called a time-dependent migration; otherwise, k0 to k is called a non-time-dependent migration.
9. The twin scene generation and verification method according to claim 1, characterized in that, In step four, a complete scene consists of timed automata. The network of timed automata on a common clock and action set consists of n timed automata. The transition relationships between states constitute the process of traffic changes in the scene, including changes in the positions of traffic participants, changes in weather, and changes in traffic rules.
10. A system for implementing the method as described in any one of claims 1-9, the system comprising a meta-scene conversion module, a meta-scene extraction module, a twin scene generation module, and a twin scene verification module.