Simulated driving scene generation method and system based on multi-level road network structure

Abstract

The invention discloses a simulation driving scene generation method and system based on a multi-level road network structure, and the method comprises the steps: generating a static simulation scene, and integrating the static simulation scene to a driving simulation platform; generating a dynamic object according to the static simulation scene, and integrating the dynamic object to the driving simulation platform; and generating the simulation driving scene in the driving simulation platform. According to the method, diversified road network structures are generated through multi-level hierarchical components, the method is suitable for simulating various traffic and road scenes, and the problems that in the prior art, the road structure of a static scene is single, and a simulation driving scene of a complex road condition cannot be simulated can be solved. According to the simulation driving scene generation method, road networks of different levels can be configured and generated according to the predefined hyper-parameters defined or randomly selected by the user, the continuous driving scene can be generated, and the complex and diversified simulation driving scene requirements can be met.

Description

Technical Field

[0001] The present application relates to the field of quantum waveform control technology, and in particular to a method and system for generating a simulated driving scenario based on a multi-level road network structure. Background Art

[0002] In the training and testing of autonomous driving systems, rich scene data is needed to train each module to improve the robustness of the model and enable a more comprehensive evaluation of the trained model. Collecting driving scene data from the real physical world is both expensive and inefficient. Therefore, automatically generating scene data in driving simulators is the key to accelerating the development of autonomous driving systems. Most existing studies generate simulated driving scenarios by changing the behavior of dynamic objects on simple road networks, without considering the impact of roadside structures and road network structures with high interaction possibilities. In the existing technology, most of them focus on the behavior of dynamic objects in predefined fixed road network structures, while ignoring the important impact of complex road networks on training and simulation scenarios. The generated simulated driving scenarios cannot meet the simulation requirements of complex road networks. Although a small number of studies have considered the collaborative generation of static and dynamic objects, the generated road structures (such as highways) are relatively simple, small in number and fragmented. Therefore, it is impossible to simulate more confrontational and cooperative driving behaviors in the generated scenarios, which occur more frequently at intersections or roundabouts than on highways. In addition, the existing technology is also unable to generate long-term continuous driving scenarios.

[0003] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0004] In view of this, the present application proposes a method and system for generating simulated driving scenes based on a multi-level road network structure. Through multi-level hierarchical components, a diversified road network structure is generated, which is suitable for simulating a variety of traffic and road scenes, and can solve the problem that the road structure of static scenes in the prior art is single and cannot simulate simulated driving scenes with complex road conditions. The simulated driving scene generation method of the present application can configure and generate road networks of different levels according to user-defined or randomly selected predefined hyperparameters, and can generate continuous driving scenes, which can meet the complex and diverse needs of simulated driving scenes.

[0005] The technical solution of this application is as follows:

[0006] The first aspect of the present application is to provide a method for generating a simulated driving scene based on a multi-level road network structure, comprising the following steps:

[0007] Generate a static simulation scenario, and integrate the static simulation scenario into a driving simulation platform;

[0008] generating a dynamic object according to the static simulation scene, and integrating the dynamic object into the driving simulation platform;

[0009] The simulated driving scenario is generated in the driving simulation platform.

[0010] Specifically, the static simulation scene includes a road network, traffic rules and roadside objects.

[0011] Specifically, generating a static simulation scene and integrating the static simulation scene into a driving simulation platform includes:

[0012] Generate a multi-level road network based on hyperparameters;

[0013] Generate traffic rule parameters according to the configured traffic rules;

[0014] Searching for empty locations on the road network and placing roadside objects;

[0015] The static simulation scene including the road network, the traffic rule parameters and the roadside objects is integrated into the driving simulation platform.

[0016] Specifically, the road network includes five levels, namely basic lanes, basic road structures, combined road structures, road segments and road networks, and each level has corresponding hyperparameters.

[0017] Specifically, generating a multi-level road network according to hyperparameters includes:

[0018] Configuring the first hyperparameter to generate a plurality of the basic lanes, wherein the types of the basic lanes include a straight lane, a curved lane, a spiral lane, and a polynomial lane;

[0019] configuring a second hyperparameter, combining one or more of the basic lanes to form a plurality of the basic road structures;

[0020] configuring a third hyperparameter, combining one or more of the basic path structures to form a plurality of the combined path structures;

[0021] configuring a fourth hyperparameter, combining one or more of the combined road structures to form a plurality of the road segments;

[0022] The fifth hyperparameter is configured to combine one or more of the combined road structures to form a plurality of the road networks.

[0023] Specifically, generating a dynamic object according to the static simulation scene and integrating the dynamic object into the driving simulation platform includes:

[0024] In the static simulation scene, one or more intelligent agents are configured, and the intelligent agents travel along a predetermined path;

[0025] Control one or more of the intelligent agents to form a traffic flow;

[0026] The traffic flow is integrated into the driving simulation platform.

[0027] The second aspect of the present application is to provide a simulation driving scene generation system based on a multi-level road network structure, comprising:

[0028] A static object generator, used to generate a static simulation scene, wherein the static simulation scene includes a road network, traffic rules and roadside objects;

[0029] A dynamic object generator, used to generate dynamic objects, wherein the dynamic objects include intelligent agents and traffic flows;

[0030] Driving simulation platform, used to integrate dynamic simulation scenes and dynamic objects, generate simulated driving scenes and output them;

[0031] The driving scene output device is used to connect to a display device to display the simulated driving scene.

[0032] Specifically, the static object generator includes a road network generator, a traffic rule generator and a roadside object generator.

[0033] Specifically, the road network generator includes a basic lane component, a basic road structure component, a combined road structure component, a road segment component and a road network component.

[0034] Specifically, the third aspect of the present application is to provide a non-transitory computer-readable storage medium, on which is stored a program for a method for generating a simulated driving scene based on a multi-level road network structure, and when the program is executed by a processor, the following steps are implemented:

[0035] Generate a static simulation scenario, and integrate the static simulation scenario into a driving simulation platform;

[0036] generating a dynamic object according to the static simulation scene, and integrating the dynamic object into the driving simulation platform;

[0037] The simulated driving scenario is generated in the driving simulation platform.

[0038] Compared with the prior art, the present invention has the following beneficial effects: the present invention generates a diversified road network structure through multi-level hierarchical components, which is suitable for simulating a variety of traffic and road scenes, and can solve the problem that the road structure of static scenes in the prior art is single and cannot simulate simulated driving scenes with complex road conditions. The simulated driving scene generation method of the present invention can configure and generate road networks of different levels according to user-defined or randomly selected predefined hyperparameters, and can generate continuous driving scenes, which can meet the needs of complex and diverse simulated driving scenes. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0040] Figure 1 It is a flowchart of a method for generating a simulated driving scene based on a multi-level road network structure according to an embodiment of the present application.

[0041] Figure 2 It is a framework diagram of the hierarchical collaborative simulation driving scenario generation method of an embodiment of the present application.

[0042] Figure 3 It is a schematic diagram of hyperparameters at various levels of the parameterized road network structure of an embodiment of the present application.

[0043] Figure 4 Schematic diagram of a method for generating a diversified road network according to an embodiment of the present application

[0044] Figure 5 It is a schematic diagram of a partial combination circuit structure of an embodiment of the present application.

[0045] Figure 6 It is a schematic diagram of a diversified road network of an embodiment of the present application.

[0046] Figure 7 It is a schematic diagram of a road network in an embodiment of the present application.

[0047] Figure 8 It is a simulated driving scene diagram generated by a simulated driving scene generation method based on a multi-level road network structure of the present application.

[0048] Fig. 9 It is a simulated driving scene diagram generated by a simulated driving scene generation method based on a multi-level road network structure of the present application.

[0049] Fig.10It is a schematic diagram of a simulated driving scene generation system based on a multi-level road network structure of the present application. DETAILED DESCRIPTION

[0050] The following will be combined with the implementation methods of this application to clearly and completely describe the technical solutions in the implementation methods of this application. Obviously, the described implementation methods are only part of the implementation methods of this application, not all of the implementation methods. Based on the implementation methods in this application, all other implementation methods obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0051] It should be understood that the terms used in this specification of the embodiments of the present application are only for the purpose of describing specific embodiments and are not intended to limit the embodiments of the present application. As used in the specification of the embodiments of the present application and the appended claims, unless the context clearly indicates otherwise, the singular forms of "a", "an" and "the" are intended to include plural forms.

[0052] In the training and testing of autonomous driving systems, rich scene data is needed to train each module to improve the robustness of the model and enable a more comprehensive evaluation of the trained model. Collecting driving scene data from the real physical world is both expensive and inefficient. Therefore, automatically generating scene data in driving simulators is the key to accelerating the development of autonomous driving systems. Most existing studies generate simulated driving scenarios by changing the behavior of dynamic objects on simple road networks, without considering the impact of roadside structures and road network structures with high interaction possibilities. In the existing technology, most of them focus on the behavior of dynamic objects in predefined fixed road network structures, while ignoring the important impact of complex road networks on training and simulation scenarios. The generated simulated driving scenarios cannot meet the simulation requirements of complex road networks. Although a small number of studies have considered the collaborative generation of static and dynamic objects, the generated road structures (such as highways) are relatively simple, small in number and fragmented. Therefore, it is impossible to simulate more confrontational and cooperative driving behaviors in the generated scenarios, which occur more frequently at intersections or roundabouts than on highways. In addition, the existing technology is also unable to generate long-term continuous driving scenarios.

[0053] In view of this, the present application proposes a method and system for generating simulated driving scenes based on a multi-level road network structure. Through multi-level hierarchical road network components, a diversified road network structure is generated, which is suitable for simulating a variety of traffic and road scenes, and can solve the problem that the road structure of static scenes in the prior art is single and cannot simulate simulated driving scenes with complex road conditions. The simulated driving scene generation method of the present application can configure and generate road networks of different levels according to user-defined or randomly selected predefined hyperparameters, and can generate continuous driving scenes, which can meet the complex and diverse needs of simulated driving scenes.

[0054] like Figure 1 and Figure 2 As shown, the first aspect of the present application is to provide a method for generating a simulated driving scene based on a multi-level road network structure, comprising the following steps:

[0055] S1. Generate a static simulation scene, and integrate the static simulation scene into a driving simulation platform;

[0056] S2. generating a dynamic object according to the static simulation scene, and integrating the dynamic object into the driving simulation platform;

[0057] S3. Generate the simulated driving scenario in the driving simulation platform.

[0058] Specifically, in the embodiments of the present application, the static simulation scene includes a road network, traffic rules, and roadside objects, and they are configured in sequence, that is, the road network is configured first, and then road markings, speed limits, traffic signs, etc. are configured on the basis of the road network according to the traffic rules, and then roadside objects, such as buildings, green belts, etc., are configured in the blank positions of the road network.

[0059] Specifically, step S1, generating a static simulation scene, and integrating the static simulation scene into a driving simulation platform, includes:

[0060] S11, generating a multi-level road network according to the hyper parameters;

[0061] S12, generating traffic rule parameters according to the configured traffic rules;

[0062] S13, searching for empty locations in the road network and placing roadside objects;

[0063] S14, integrating the static simulation scene including the road network, the traffic rule parameters and the roadside objects into the driving simulation platform.

[0064] Specifically, the road network includes five levels of road network components, namely basic lanes, basic road structures, combined road structures, road segments and road networks, and each level has corresponding hyperparameters. The multi-level road network of the present application is divided into five levels, the first level of basic lanes, the second level of basic road structures, the third level of combined road structures, the fourth level of road segments, and the fifth level of road networks. From the first level to the fifth level, the complexity increases successively, and the high-level road network components are composed of components of the lower level. Therefore, the high-level road network components are generated by combining the low-level road network components, and can be flexibly generated according to user-defined or randomly selected hyperparameters. For example Figure 3, which is a schematic diagram of hyperparameters at various levels of the parameterized road network structure of an embodiment of the present application, summarizes some of the hyperparameters of the present application for flexibly generating diversified road network components at various levels.

[0065] Specifically, the types of basic lanes at the first level include straight lanes, curved lanes, spiral lanes, and polynomial lanes. The hyperparameters corresponding to the basic lanes are the first hyperparameters, and the first hyperparameters corresponding to the straight lanes include length, starting orientation, and lane width. The first hyperparameters corresponding to the curved lanes include length, starting orientation, lane width, and starting curvature. The first hyperparameters corresponding to the spiral lanes include length, starting orientation, lane width, and starting curvature and end curvature. The first hyperparameters corresponding to the polynomial lanes include length, starting orientation, lane width, and starting curvature, end curvature, and polynomial parameters.

[0066] Specifically, a polynomial lane refers to a set of multiple lanes that have a spatial relationship with each other. The spatial relationship can be a two-dimensional or three-dimensional spatial function. For example, a polynomial lane can be represented by y=a+bx+c, where a, b, c are coefficients, and x and y represent the location of the center line of the lane.

[0067] Specifically, the types of the second-level basic road structure include straight roads, curved roads, spiral roads, and polynomial roads. The parameters corresponding to the second-level basic road structure are second hyperparameters, and any type of second hyperparameters include the number of left lanes, the number of right lanes, and the width of each lane. The second-level basic road structure is composed of the first-level basic lanes. For example, a straight road can be composed of one or more straight roads. A polynomial road can be composed of one or more straight roads, one or more curved roads, and one or more spiral roads.

[0068] Specifically, the types of combined road structures at the third level include intersections, roundabouts, forks and changes in the number of lanes. The parameters corresponding to the combined road structures at the third level are third hyperparameters. Intersections may include T-intersections, cross intersections, 5-way intersections, 6-way intersections, 8-way intersections, etc. The third hyperparameters corresponding to the intersections include the number of intersections, intersection radius, turning mode, connection mode and road type. Preferably, the number of intersections ranges from 3 to 10, preferably 3, 4, 5, 6, and 8. The third hyperparameters corresponding to the roundabouts include the number of intersections, intersection radius, roundabout radius, turning mode, connection mode and road type. The change in the number of lanes refers to the increase or decrease in the number of lanes in a certain section of the path of a combined road structure. In this way, more complex combined road structures can be controlled. The third hyperparameters corresponding to the change in the number of lanes include the connection radius and the road type. The combined road structure at the third level is composed of one or more basic road structures at the second level. For example, a crossroad consists of four straight roads and a circular spiral road, and any two adjacent straight roads are connected by an arc road. Figure 5 The diagram shown is a schematic diagram of some combined road structures of the embodiments of the present application, including T-shaped structures with different hyperparameters, cross-shaped structures without hyperparameters, intersections with different numbers of intersections, ring structures with different hyperparameters, and forked roads and variable road structures with different hyperparameters. These embodiments are only some of the embodiments of the present application, and the present application can set various combined road structures.

[0069] Specifically, the parameters corresponding to the road segments (or road sections) of the fourth level are fourth hyperparameters, including the number of basic road structures and the type of technical road structures. A road segment is composed of one or more combined road structures.

[0070] Specifically, the parameters corresponding to the fifth level of the road network are the fifth hyperparameters, including the road network width and the road network length. The road network is composed of one or more road segments. For example, Figure 6 (a), (b), (c) and (d) are different road network structures formed by different hyperparameter configurations. Figure 7 yes Figure 6 The specific presentation of (a) in .

[0071] like Figure 7The figure is a schematic diagram of a road network of an embodiment, which includes a straight road (Z1), a plurality of curved roads (H1, H2, H3, H4), a roundabout (HD1) and a three-way road (CL1), wherein the straight road and the curved road are both bidirectional six-lane roads; the roundabout (HD1) has 4 intersections, including four cross-shaped straight roads (Z2, Z3, Z4, Z5) and a circular spiral road (D1), and the circular spiral road (D1) has 7 lanes; the three-way road (CL1) includes three straight roads (Z6, Z7, Z8) and a plurality of curved roads (H5) connecting the straight roads (Z6, Z7, Z8). When generating this road network, you can first configure the first-level lane parameters, such as configuring the starting curvature and end curvature of each curved road, the length and width of each straight road, and then configure the second-level left lane number and right lane number to 3, thereby forming a two-way 6-lane straight road or curved road, and configure the number of lanes of the spiral road to 7; further configure a roundabout and 3 forks, where the roundabout is configured with 4 intersections and sets a specific intersection radius; further, configure the hyperparameters of the road section to determine the number and type of basic road structures; finally, the final road network output is achieved by determining the road network width and road network length.

[0072] Specifically, before generating high-level road network components, the hyperparameters of low-level road network components should be set first. By combining road network components of different levels or at different levels, a diverse road network suitable for different regions (e.g., rural areas, cities, and highways) can be constructed. For example, if a user wants to experience driving scenes in rural areas, a single-lane straight road can be selected as the road network. If a user wants to experience driving scenes in first-tier cities, a combination of roundabouts + two-way 6-lane roads + curved roads can be selected as the road network, and the parameters of each component are configured in sequence from low to high levels, and finally a complex road network is output.

[0073] like Figure 6 The figure shows a schematic diagram of a complex road network, which includes multiple basic components. Users can configure hyperparameters at different levels to generate the required diversified road network as needed, or directly select a designed diversified road network from the preset database.

[0074] Specifically, this application uses FlexiRoad, a flexibly configurable road network generator. Figure 4 As shown, it is a generation framework diagram of the parameterized hierarchical road network of the present application. Therefore, step S11, generating a multi-level road network according to the hyper parameters, includes:

[0075] S111, configuring a first hyperparameter to generate a plurality of basic lanes, wherein the types of the basic lanes include a straight lane, a curved lane, a spiral lane, and a polynomial lane;

[0076] S112, configuring a second hyper parameter, combining one or more basic lanes to form a plurality of basic road structures;

[0077] S113, configuring a third hyperparameter, combining one or more of the basic path structures to form a plurality of combined path structures;

[0078] S114, configuring a fourth hyperparameter, combining one or more of the combined road structures to form a plurality of the road segments;

[0079] S115. Configure the fifth hyperparameter, and combine one or more of the combined road structures to form a plurality of the road networks.

[0080] Specifically, for example, in the process of generating the fourth-level road segment, it can be performed in a plane coordinate system or a three-dimensional coordinate system, and the following steps can be followed:

[0081] (1) First, the parameters are given: the number n (required) and type (optional) of basic road structures and the arrangement method (such as horizontal arrangement, vertical arrangement, center-divergent arrangement);

[0082] (2) Then determine the original coordinates (x, y) and randomly select any basic road structure at the origin, represented by parameter α;

[0083] (3) Let Last = α. For α = 2, ..., n, search for blank positions above, below, left, and right of Last to form a set C.

[0084] (4) Randomly select a blank position in C to place another basic road structure, which is β;

[0085] (5) Use any spiral road γ to connect α and β to generate a road segment containing α, β, and γ;

[0086] (6) Output the generated road segment.

[0087] Specifically, step S12 generates traffic rule parameters according to the configured traffic rules, including:

[0088] After the road network is generated, it is necessary to set traffic rules (such as road markings, speed limits, and traffic signs) based on the road network according to relevant standards (GB-5768). Specifically, road markings include informative markings (such as dotted lines, solid lines), prohibitory markings (such as stop lines, yield lines), and warning markings (such as deceleration lines). For each lane in the previously generated road network, when setting road markings, first find all road markings that meet the rules in the input setting rule set (specified in GB-5768) in the input road marking set, and then find these road markings on each side of the lane. Similarly, when setting the speed limit, first determine the area where the vehicle is located, and then randomly select the speed limit related to the area from the input speed limit set (specified in GB-5768). The pseudo code of the rule-based traffic rule generator GTR is shown in Table 1.

[0089] Table 1 Algorithm pseudo code

[0090]

[0091] Table 1 shows the input parameters of the algorithm pseudo code:

[0092] (1) The road network N is a set of several road networks generated by steps S111 to S115;

[0093] (2) The road marking setting rule set R is extracted from the standard GB-5768;

[0094] (3) The road marking set Y is also extracted from standard GB-5768.3, including: white dashed line, white solid line, yellow dashed line, yellow solid line, double white dashed line, double white solid line, white dashed and solid line, double yellow solid line, double yellow dashed line, yellow dashed and solid line;

[0095] (4) Speed ​​limit set SC uses common data from real life: for example, the speed limits on a two-way three-lane expressway are 120-110, 110-90, and 90-60.

[0096] The algorithm flow of the algorithm pseudocode in Table 1 is as follows: First, for each generated lane, test whether each type of marking in the road marking set meets the rules in R, and put the marking sets that meet the requirements into a set; then randomly select a type of marking for the left and right sides of each road; for the speed limit, randomly select a speed limit in the speed limit set according to the regional location of the road.

[0097] In this application, a traffic rule generator is configured to implement the setting of traffic rules in the road network to simulate real driving scenarios in accordance with relevant standards.

[0098] Specifically, step S13, searching for empty positions of the road network and placing roadside objects includes searching for empty spaces of a map containing the road network. Then, the roadside objects are randomly placed at feasible positions on both sides of the road network, and the number of floors of the buildings can be set arbitrarily. Specifically, any roadside structure can be imported into the system of the present application as long as it has a corresponding 3D model.

[0099] In summary, the static object generator of the present application sequentially implements the settings of road network, traffic rules and roadside objects, thereby generating a complete static scene containing a diverse road network. Afterwards, the static scene can be integrated into the driving simulation platform. At the same time, dynamic objects are generated based on the static scene.

[0100] Specifically, the present application configures a dynamic object generator, which is responsible for controlling dynamic objects in the driving scene, such as vehicles, bicycles, and pedestrians. The dynamic object generator includes an intelligent agent manager and a traffic flow manager.

[0101] Specifically, the rule-based agents include two agents: Random Lane Keeping and Waypoint Tracking. Taking the Random Lane Keeping agent as an example, it first randomly selects a starting point from the road network, and then randomly selects an action from the discrete action space "{keep lane, decelerate, accelerate, turn left, turn right}", while ensuring that no collision occurs at each step until the agent reaches the preset goal or exits the driving scene. Specifically, lane keeping, deceleration, and acceleration are actions for longitudinal control, while left turn and right turn are used for lateral control. Similarly, the waypoint tracking agent moves along a given waypoint path or a waypoint path randomly selected by a depth-first search algorithm. Specifically, the waypoint path is smoothed to ensure that the vehicle can accurately follow the predetermined path. These vehicles are controlled by adjusting their steering angles and accelerations.

[0102] Specifically, the traffic flow manager creates traffic flow by controlling multiple agents simultaneously on a previously generated scenario. First, several vehicles controlled by randomly selected agents are placed in the driving scenario to ensure that there are no position conflicts. Then, the movement trajectory of each vehicle is generated by its corresponding agent. In addition, the movement of dynamic objects can also be controlled using the built-in functions of driving simulators, such as the CARLA simulator and the traffic simulation function in the SUMO simulator.

[0103] In summary, S2, generating a dynamic object according to the static simulation scene, and integrating the dynamic object into the driving simulation platform, includes:

[0104] S21. In the static simulation scene, configure one or more intelligent agents, and the intelligent agents travel along a predetermined path;

[0105] S22. Control one or more of the intelligent agents to form a traffic flow.

[0106] After step S2, generating a dynamic object according to the static simulation scene and integrating the dynamic object into the driving simulation platform, the method further includes S3, generating the simulated driving scene in the driving simulation platform.

[0107] Specifically, the driving simulation platform may use existing popular driving simulators (eg, CARLA, SMARTS, ESMINI) to render the generated driving scenarios.

[0108] Specifically, the present application is also configured with a display that can display the generated simulated driving scene.

[0109] like Figure 8 As shown, a simulated driving scene diagram is generated by a simulated driving scene generation method based on a multi-level road network structure of the present application. The figure shows a simulated road network structure of simulated urban roads. Four simulated scenes are selected and enlarged for display. It can be seen that the figure not only has different structures and types of roads, but also buildings and intelligent entities, and different states of buildings and intelligent entities are displayed in different sections of the road. Therefore, the present application can realize the simulation of urban road traffic and enhance the experience and realism of the simulated scene.

[0110] like Fig. 9 As shown, a simulated driving scene diagram is generated by a simulated driving scene generation method based on a multi-level road network structure of the present application. The diagram shows a road network structure including a roundabout, and different types of simulated traffic flows of cars are added to the static simulation scene. The specific road conditions of a certain section can be arbitrarily enlarged through the display.

[0111] In summary, this application sets five levels of road network components, sets the hyperparameters of each level in order from low to high, combines components of different levels, generates a variety of road networks, and is suitable for a variety of driving scenarios. This improves the complexity and authenticity of the simulated road network, and can combine different road sections or road structures to generate a continuous road network that is not limited by time to adapt to continuous driving scenarios, which can provide detailed and diverse scenarios for adversarial and cooperative simulations in the technology of autonomous driving systems. In addition, the roadside object generator of this application can input any 3D model, and the parameters or data of the entire scene can be integrated into the existing driving simulator, which improves practicality and scalability.

[0112] like Fig.10 As shown, the second aspect of the present application is to provide a simulation driving scene generation system 10 based on a multi-level road network structure, comprising:

[0113] A static object generator 20, used to generate a static simulation scene, wherein the static simulation scene includes a road network, traffic rules and roadside objects;

[0114] A dynamic object generator 30, used to generate a dynamic object, wherein the dynamic object includes an intelligent agent 31 and a traffic flow manager 32;

[0115] The driving simulation platform 40 is used to integrate dynamic simulation scenes and dynamic objects, generate simulated driving scenes and output them.

[0116] The driving scene outputter 50 is used to connect to a display device to display the simulated driving scene.

[0117] Specifically, the static object generator 20 includes a road network generator 21 , a traffic rule generator 22 and a roadside object generator 23 .

[0118] Specifically, the road network generator 21 includes a basic lane component 211 , a basic road structure component 212 , a combined road structure component 213 , a road segment component 214 and a road network component 215 .

[0119] Specifically, the driving simulation platform 40 includes a static object manager 41 and a dynamic object manager 42 .

[0120] Specifically, the driving scene outputter 50 includes a customization manager 51 and a scene selection manager 52 .

[0121] Specifically, the third aspect of the present application is to provide a non-transitory computer-readable storage medium, on which is stored a program for a method for generating a simulated driving scene based on a multi-level road network structure, and when the program is executed by a processor, the following steps are implemented:

[0122] S1. Generate a static simulation scene, and integrate the static simulation scene into a driving simulation platform;

[0123] S2. generating a dynamic object according to the static simulation scene, and integrating the dynamic object into the driving simulation platform;

[0124] S3. Generate the simulated driving scenario in the driving simulation platform.

[0125] In summary, the present application proposes a method and system for generating simulated driving scenes based on a multi-level road network structure, the method comprising: generating a static simulation scene, and integrating the static simulation scene into a driving simulation platform; generating a dynamic object according to the static simulation scene, and integrating the dynamic object into the driving simulation platform; generating the simulated driving scene in the driving simulation platform. The present application generates a variety of road network structures through multi-level hierarchical components, which is suitable for simulating a variety of traffic and road scenes, and can solve the problem in the prior art that the road structure of static scenes is single and cannot simulate simulated driving scenes with complex road conditions. The simulated driving scene generation method of the present application can configure and generate road networks of different levels according to user-defined or randomly selected predefined hyperparameters, and can generate continuous driving scenes, which can meet the needs of complex and diverse simulated driving scenes.

[0126] The above description is only a preferred implementation mode of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for generating a simulated driving scene based on a multi-level road network structure, characterized in that: The following steps are involved: Generate a static simulation scenario, and integrate the static simulation scenario into a driving simulation platform; generating a dynamic object according to the static simulation scene, and integrating the dynamic object into the driving simulation platform; The simulated driving scenario is generated in the driving simulation platform.

2. The method for generating a simulated driving scene based on a multi-level road network structure according to claim 1, characterized in that: The static simulation scene includes a road network, traffic rules and roadside objects.

3. The method for generating a simulated driving scene based on a multi-level road network structure according to claim 2, characterized in that: The generating of a static simulation scene and integrating the static simulation scene into a driving simulation platform includes: Generate a multi-level road network based on hyperparameters; Generate traffic rule parameters according to the configured traffic rules; Searching for empty locations on the road network and placing roadside objects; The static simulation scene including the road network, the traffic rule parameters and the roadside objects is integrated into the driving simulation platform.

4. The method for generating a simulated driving scene based on a multi-level road network structure according to claim 3, characterized in that: The road network includes five levels, namely basic lanes, basic road structures, combined road structures, road segments and road networks, and each level has corresponding hyperparameters.

5. The method for generating a simulated driving scene based on a multi-level road network structure according to claim 4, characterized in that: The generating of a multi-level road network according to the hyper parameters comprises: Configuring the first hyperparameter to generate a plurality of the basic lanes, wherein the types of the basic lanes include a straight lane, a curved lane, a spiral lane, and a polynomial lane; configuring a second hyperparameter, combining one or more of the basic lanes to form a plurality of the basic road structures; configuring a third hyperparameter, combining one or more of the basic path structures to form a plurality of the combined path structures; configuring a fourth hyperparameter, combining one or more of the combined road structures to form a plurality of the road segments; The fifth hyperparameter is configured to combine one or more of the combined road structures to form a plurality of the road networks.

6. The method for generating a simulated driving scene based on a multi-level road network structure according to claim 1, characterized in that: The generating of a dynamic object according to the static simulation scene and integrating the dynamic object into the driving simulation platform comprises: In the static simulation scene, one or more intelligent agents are configured, and the intelligent agents travel along a predetermined path; Control one or more of the intelligent agents to form a traffic flow; The traffic flow is integrated into the driving simulation platform.

7. A simulation driving scene generation system based on a multi-level road network structure, characterized in that: include: A static object generator, used to generate a static simulation scene, wherein the static simulation scene includes a road network, traffic rules and roadside objects; A dynamic object generator, used to generate dynamic objects, wherein the dynamic objects include intelligent agents and traffic flows; Driving simulation platform, used to integrate dynamic simulation scenes and dynamic objects, generate simulated driving scenes and output them; The driving scene output device is used to connect to a display device to display the simulated driving scene.

8. The system for generating a simulated driving scene based on a multi-level road network structure according to claim 7, characterized in that: The static object generator includes a road network generator, a traffic rule generator and a roadside object generator.

9. The system for generating a simulated driving scene based on a multi-level road network structure according to claim 8, characterized in that: The road network generator includes a basic lane component, a basic road structure component, a combined road structure component, a road segment component and a road network component.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores a program for a method for generating a simulated driving scene based on a multi-level road network structure, and when the program is executed by a processor, the following steps are implemented: Generate a static simulation scenario, and integrate the static simulation scenario into a driving simulation platform; generating a dynamic object according to the static simulation scene, and integrating the dynamic object into the driving simulation platform; The simulated driving scenario is generated in the driving simulation platform.

Images