Simulation scene generation method, device and equipment

Abstract

The embodiment of the invention provides a simulation scene generation method, device and equipment, and the method comprises the steps: obtaining scene description information of a target simulation scene, and the scene description information is used for describing the movement information and road condition information of traffic participants; obtaining a global map of the target area, and determining a target local map corresponding to the target simulation scene in the target area based on the scene description information and the global map; and generating a target simulation scene based on the target local map and the scene description information. According to the simulation scene generation method provided by the invention, the construction efficiency and controllability of the simulation scene can be improved.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and specifically to a simulation scene generation method, device and equipment. Background Art

[0002] With the development of autonomous driving technology, autonomous driving testing is required to ensure the safety of the autonomous driving system. In the testing process of autonomous driving equipment, compared with traditional real vehicle road testing, closed-loop simulation testing has the advantages of high scene coverage, safe testing process, and high testing efficiency, so it plays an increasingly important role.

[0003] At present, there are two mainstream ways to build simulation scenes, including: controllable editing of simulation scenes and conversion of real road data. Among them, the construction method of controllable editing of simulation scenes is that users edit the trajectories of traffic participants or choreograph their interactive behaviors with their own vehicles. The construction method of real road test data conversion refers to recording the traffic conditions and observation results on the real road, extracting key element information and converting them into simulation scenes. The above two simulation scene construction methods require users to have professional knowledge of simulators and scene construction, with high construction thresholds and low construction efficiency. Summary of the invention

[0004] In view of this, the present application provides a simulation scene generation method, device and equipment to help solve the problem of low efficiency in simulation scene construction in the prior art.

[0005] In a first aspect, an embodiment of the present application provides a simulation scene generation method, including: obtaining scene description information of a target simulation scene; the scene description information is used to describe the movement information and road condition information of traffic participants; obtaining a global map of the target area, and based on the scene description information and the global map, determining a target local map within the target area corresponding to the target simulation scene; and then generating the target simulation scene based on the target local map and the scene description information.

[0006] In a possible implementation, the movement information of the traffic participants includes the movement information of the vehicle and / or the movement information of other traffic participants; the road condition information includes: static road condition information and / or dynamic road condition information.

[0007] In a possible implementation, obtaining scene description information of a target simulation scene includes: obtaining user description information of the target simulation scene, the user description information being description information of the target simulation scene by a user; and determining the scene description information of the target simulation scene according to the user description information. The type of the user description information includes at least one of a text type, a voice type, and an image type. Exemplarily, the user description information can be used as an input of a scene description model to obtain scene description information output by the scene description model, and the scene description model can be a neural network model.

[0008] In a possible implementation, the above-mentioned determination of the target local map in the target area corresponding to the target simulation scene based on the scene description information and the global map includes: determining the static topology map corresponding to the scene description information based on the scene description information; and then associating the static topology map with the global map to determine the target local map in the target area corresponding to the target simulation scene. Exemplarily, the scene description information can be used as the input of a topology completion model to obtain a static topology map corresponding to the scene description information output by the topology completion model, wherein the topology completion model can be a neural network model, which can complete the input scene description information, thereby generating a static topology map based on the completed and more detailed scene description information.

[0009] In a possible implementation, when the type of user description information includes description information of an image type, a static topology map corresponding to the scene description information can be determined based on the description information of the image type and the scene description information. Exemplarily, the description information of the image type and the scene description information can be used as inputs of a topology completion model to obtain a static topology map corresponding to the scene description information output by the topology completion model.

[0010] In one possible implementation, the static topology map is associated with the global map, and determining the target local map within the target area corresponding to the target simulation scenario includes: based on the static topology map, screening out the target area map with the same road category as that in the static topology map in the global map; then calculating the similarity between the target area map and the static topology map; based on the similarity, determining the target local map within the target area corresponding to the target simulation scenario.

[0011] In a possible implementation, determining the target local map in the target area corresponding to the target simulation scenario based on the similarity includes: determining the target area map having the highest similarity to the static topology map as the target local map in the target area corresponding to the target simulation scenario.

[0012] In one possible implementation, based on the similarity, determining the target local map within the target area corresponding to the target simulation scene includes: outputting to the user the similarity between the filtered target area map and the static topology map or outputting to the user a sorted target area map (which can be sorted from high to low or from low to high based on the similarity); in response to the user's selection operation on the target area map, determining the target area map selected by the user as the target local map within the target area corresponding to the target simulation scene.

[0013] In one possible implementation, generating a target simulation scene based on a target local map and scene description information includes: generating initial state information of the vehicle and movement trajectory information of other traffic participants based on the target local map and scene description information; obtaining preset road boundary rules, and determining target movement trajectory information that complies with the preset road boundary rules from the movement trajectory information of other traffic participants; and then generating a target simulation scene based on the initial state information of the vehicle and the target movement trajectory information of other traffic participants.

[0014] In a possible implementation, the initial state information of the vehicle includes at least one of the position information, speed information and heading angle information of the vehicle at the initial moment in the target simulation scene; the movement trajectory information of other traffic participants includes the driving status at T moments, wherein the driving status at each of the T moments includes at least one of the position information, speed information and heading angle information of other traffic participants, and T is a pre-set integer greater than 0.

[0015] In the second aspect, an embodiment of the present application provides a simulation scene generation device, including: an acquisition unit, used to acquire scene description information of a target simulation scene, the scene description information being used to describe movement information and road condition information of traffic participants; the acquisition unit is also used to acquire a global map of a target area; a processing unit, used to determine a target local map within a target area corresponding to the target simulation scene based on the scene description information and the global map; the processing unit is also used to generate a target simulation scene based on the target local map and the scene description information.

[0016] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the electronic device is triggered to execute any one of the methods described in the first aspect above.

[0017] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any method described in the first aspect.

[0018] In a fifth aspect, an embodiment of the present application provides a computer program product, which includes executable instructions. When the executable instructions are executed on a computer, the computer executes any method described in the first aspect.

[0019] Adopt the scheme provided in the embodiment of the present application, obtain the scene description information of the target simulation scene; obtain the global map of the target area, and determine the target local map in the target area corresponding to the target simulation scene based on the scene description information and the global map of the target area; generate the target simulation scene based on the target local map and the scene description information. In this way, in the embodiment of the present application, after obtaining the scene description information of the target simulation scene, the target simulation scene required by the user can be generated based on the scene description information of the target simulation scene. It can effectively improve the efficiency of scene construction, lower the threshold for constructing simulation scenes, and improve ease of use.

[0020] In addition, in the embodiment of the present application, a target local map that matches the scene description information of the acquired target simulation scene can be determined in the global map, thereby improving the diversity of the target local map and thereby improving the generalization of the generated simulation scene. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments 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 labor.

[0022] Figure 1 A flowchart of a simulation scene generation method provided in an embodiment of the present application;

[0023] Figure 2a An example schematic diagram of user description information provided in an embodiment of the present application;

[0024] Figure 2b An exemplary schematic diagram of another user description information provided in an embodiment of the present application;

[0025] Figure 2c An example schematic diagram of obtaining scene description information provided in an embodiment of the present application;

[0026] Figure 3a A schematic diagram of an example process for generating a simulation scenario provided in an embodiment of the present application;

[0027] Figure 3b An example schematic diagram of a static topology diagram provided in an embodiment of the present application;

[0028] Figure 3c An example schematic diagram of a target local map provided in an embodiment of the present application;

[0029] Figure 4 An example schematic diagram of generating a static topology diagram provided in an embodiment of the present application;

[0030] Figure 5a An example schematic diagram of a target simulation scenario provided in an embodiment of the present application;

[0031] Figure 5b Another example schematic diagram of generating a target simulation scenario provided in an embodiment of the present application;

[0032] Figure 6a A schematic diagram of an application of a simulation scenario provided in an embodiment of the present application;

[0033] Figure 6b A schematic diagram of another application of a simulation scenario provided in an embodiment of the present application;

[0034] Figure 7 A schematic diagram of the structure of a simulation scene generation device provided in an embodiment of the present application;

[0035] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0036] In order to better understand the technical solution of the present application, the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0037] It should be clear that the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0038] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0039] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0040] In the related technology, there are two mainstream construction methods of simulation scenes, including: controllable editing of simulation scenes and conversion of real road data. Among them, the construction method of controllable editing of simulation scenes is that users edit the trajectories of traffic participants or choreograph their interactive behaviors with their own vehicles. The construction method of real road test data conversion refers to recording the traffic conditions and observation results on the real road, extracting key element information and converting them into simulation scenes. The above two construction methods of simulation scenes require users to have professional knowledge of simulators and scene construction, with high construction thresholds and low construction efficiency.

[0041] In response to the above problems, an embodiment of the present application provides a simulation scene generation method, device and equipment, the method specifically comprising: obtaining scene description information of a target simulation scene; obtaining a global map of the target area, and determining a target local map in the target area corresponding to the target simulation scene based on the scene description information and the global map of the target area; generating a target simulation scene based on the target local map and the scene description information. In this way, in an embodiment of the present application, after obtaining the scene description information of the target simulation scene, the target simulation scene required by the user can be generated based on the scene description information of the target simulation scene. The efficiency and controllability of scene construction can be effectively improved, and professional domain knowledge of scene simulators and scene construction is no longer required, which lowers the threshold for use and improves ease of use.

[0042] The details are described below.

[0043] See also Figure 1 , is a flow chart of a simulation scene generation method provided in an embodiment of the present application. Figure 1 As shown, the method includes:

[0044] Step S101: Obtain scene description information of a target simulation scene.

[0045] The scene description information of the target simulation scene is used to describe the movement information and road condition information of traffic participants.

[0046] In an embodiment of the present application, in order to ensure the safety of the autonomous driving system, an autonomous driving test needs to be performed. Therefore, it is necessary to pre-generate a simulation scene for testing the autonomous driving system. That is, the simulation scene is a file that describes the vehicle's driving environment during the autonomous driving test. Normally, the vehicle's driving environment usually includes the road condition information of the vehicle and the movement information of traffic participants. Therefore, in order to generate the target simulation scene required by the user, it is necessary to first obtain the scene description information of the target simulation scene. In some embodiments, the simulation scene generation device can obtain the scene description information of the target simulation scene from other devices. Alternatively, the user can directly input the scene description information of the target simulation scene, and at this time, the simulation scene generation device uses the description information input by the user as the scene description information of the target simulation scene. Since the scene description information of the target simulation scene is used to describe the movement information and road condition information of the traffic participants, if the user description information input by the user is directly used as the scene description information of the target simulation scene, when the description information input by the user is not detailed enough for the description of the target simulation scene, for example, it only contains one of the movement information and road condition information of the traffic participants, at this time, directly using the description information input by the user as the scene description information of the target simulation scene may cause the target simulation scene generated subsequently to be inaccurate and not the simulation scene required by the user. In addition, some types of user description information input by the user are difficult to understand, for example, the sketch of the target simulation scene drawn by the user, that is, the user description information of the image type is difficult to identify and understand. If the user description information input by the user is directly used as the scene description information of the target simulation scene, it will also cause the target simulation scene generated subsequently to be not the simulation scene required by the user.

[0047] In order to improve the accuracy of generating the target simulation scene, the scene description information of the target simulation scene can be obtained in the following manner when receiving the description information input by the user. That is, obtaining the scene description information of the target simulation scene includes: obtaining the user description information of the target simulation scene; using the user description information as the input of the scene description model to obtain the scene description information of the target simulation scene output by the scene description model.

[0048] The user description information refers to the description information of the target simulation scene by the user; the scene description model is pre-trained and used to extract and supplement the information of the simulation scene from the input user description information and output the scene description information of the target simulation scene.

[0049] In an embodiment of the present application, a user can input his description information of the target simulation scene to the simulation scene generation device, that is, input the user description information of the target simulation scene to the simulation scene generation device. The user description information input by the user may include the movement information and / or road condition information of the traffic participants. In other words, when the user inputs the user description information of the target simulation scene to the simulation scene generation device, the user description information may not describe the target simulation scene in detail enough. The simulation scene generation device can use the acquired user description information of the target simulation scene as the input of the preset scene description model, and input it into the scene description model. At this time, after receiving the user description information, the scene description model can understand the simulation scene information of the user description information and perform reasonable reasoning of the simulation scene to realize information extraction of the simulation scene, perform reasonable reasoning and supplement the description of the target simulation scene missing in the user description information, obtain the scene description information of the target simulation scene, and output it. At this time, the simulation scene generation device can obtain the scene description information of the target simulation scene output by the scene description model. In some embodiments, the target content that needs to be included in the scene description information of the target simulation scene can be pre-set. At this time, when the scene description model is training the model, the scene description model can be continuously trained to output the scene description information of the target simulation scene containing the target content. In this way, the scene description model extracts the information of the simulation scene from the user description information, and supplements the missing description information in the user description information based on the extracted information of the target simulation scene and the target content that needs to be included in the scene description information, thereby obtaining the final output result.

[0050] It should be noted that the user description information of the target simulation scene may be a simple description of the target simulation scene by the user, or may be a detailed description of the target simulation scene by the user, and this application does not impose any restrictions on this.

[0051] In other embodiments, the target content required to be included in the scene description information of the target simulation scene can be preset as simulation scene prompt information, so that the scene description model can output the scene description information of the target simulation scene according to the preset simulation scene prompt information. At this time, taking the user description information of the target simulation scene as the input of the scene description model, and obtaining the scene description information of the target simulation scene output by the scene description model includes: taking the user description information of the target simulation scene as the input of the scene description model, and obtaining the scene description information of the target simulation scene output by the scene description model according to the preset simulation scene prompt information. Among them, the preset simulation scene prompt information is used to prompt the scene description model that the scene description information of the target simulation scene output must include the movement information and road condition information of the traffic participants.

[0052] That is, the simulation scene generation device can use the acquired user description information of the target simulation scene as the input of the preset scene description model and input it into the scene description model. After receiving the user description information, the scene description model can extract the information of the simulation scene from the user description information, and under the prompt of the preset simulation scene prompt information, supplement the description of the target simulation scene missing in the user description information, obtain the scene description information of the target simulation scene, and output it. At this time, the simulation scene generation device can obtain the scene description information of the target simulation scene output by the scene description model.

[0053] It should be understood that the preset simulation scene prompt information is used to prompt the content that the scene description model needs to output. At this time, the scene description model can output the inference result of the required target simulation scene description instance according to the preset simulation scene prompt information, that is, the scene description information of the target simulation scene. The specific prompt content of the preset simulation scene prompt information is pre-set by the user according to actual needs. For example, the preset simulation scene prompt information includes descriptions of the movement information of traffic participants and descriptions of road conditions.

[0054] As a possible implementation manner, the movement information of the traffic participants includes the movement information of the vehicle and / or the movement information of other traffic participants; the road condition information includes: static road condition information and / or dynamic road condition information.

[0055] The traffic participants refer to moving objects participating in the traffic, including but not limited to at least one of the vehicle, other vehicles, and pedestrians. When constructing the target simulation scene, the movement information of the traffic participants in the target simulation scene needs to be described.

[0056] The vehicle movement information includes: straight driving, U-turn, reverse driving, turning, cruise following, overtaking, meeting, obstacle avoidance, merging in, merging out, changing lanes, continuous turning, braking, accelerating, deviating from the center line of the lane, stationary, rushing, giving way, etc. Of course, the vehicle movement information may also include other information, such as location, driving speed, heading angle, acceleration, etc., which is not limited in this application.

[0057] Other traffic participants may be pedestrians, cyclists, cars, trucks, buses, other vehicles, other objects, etc. At this time, the movement information of other traffic participants includes: cutting in, cutting out, borrowing lanes, emergency braking, crossing, blocking, overtaking, etc. Of course, the movement information of other participants may also include other information, such as the position, moving speed, heading angle, acceleration, etc. of other traffic participants, which is not limited in this application.

[0058] Road condition information refers to information about road conditions in the target simulation scene. Road condition information can be divided into static road condition information and dynamic road condition information. Static road condition information includes at least one of road type, road category, and lane information. Of course, static road condition information can also include other road information, such as static obstacles in the road, such as buildings, curbs, traffic barriers, low non-traffic barriers, etc., which are not limited in this application.

[0059] The road type includes at least one of: expressway type, national road type, provincial road type, urban express road type, ordinary road type, main road type, secondary branch road type, non-navigation road type, etc. Of course, the road type may also include other road types, which is not limited in this application.

[0060] Road categories include: intersections, viaducts, gaps, ramps, steep slopes, straight roads, curves, bridges, tunnels, hump bridges, etc. Of course, road categories may also include other road categories, which are not limited in this application.

[0061] In some embodiments, the intersection can be divided into a crossroad, a T-shaped intersection, an X-shaped intersection, a uturn intersection, a roundabout, a NN intersection, a complex intersection, etc. Of course, the intersection can also be other types of intersections, and this application does not limit this. In some embodiments, the gap can be divided into a left gap, a right gap, and a double-sided gap. Wherein, the left gap refers to the gap on the left side of the vehicle along the direction of travel of the vehicle. The right gap refers to the gap on the right side of the vehicle along the direction of travel of the vehicle. In some embodiments, the ramp can be divided into a ramp entrance and a ramp exit. In some embodiments, a steep slope can be divided into a slope, an uphill slope, a downhill slope, etc. Of course, a steep slope can also be divided into other slope types, and this application does not limit this. In some embodiments, a curve can be divided into an s-bend, a right-angle curve, a curve of different curvatures, etc. Of course, a curve can also be divided into other types of curves, and this application does not limit this.

[0062] Lane information: quantity, direction, type, characteristics, etc. Of course, the lane information also includes other information, which is not limited in this application.

[0063] In some embodiments, the number of lanes can be divided into one-way lanes, two-way lanes, three-way lanes, multiple lanes, etc. The lane direction can be divided into one-way, two-way, etc. The lane type can be divided into ordinary lanes, right-turn lanes, U-turn lanes, waiting lanes, tidal lanes, variable lanes, shoulder lanes, bus lanes, non-motorized vehicle lanes, etc. The lane characteristics can be divided into one to two, two to one, width to two, lane number reduction / increase, unchanged lanes, etc. Of course, the above-mentioned number of lanes can also be divided into other numbers, lane directions can also be divided into other directions, lane types can also be divided into other types, and lane characteristics can also be divided into other characteristics, and this application does not limit this.

[0064] Dynamic traffic information refers to traffic information that changes within a certain period of time, and may include information such as road construction, accident scenes, and traffic control. Of course, it may also include other information, such as traffic light information, etc., which is not limited in this application. Since dynamic traffic information is traffic information that changes within a certain period of time, it can also be called temporary operation information.

[0065] In some embodiments, in order to improve the accuracy of the target simulation scene construction, the preset simulation scene prompt information can usually be prompted in multiple dimensions, and the following is explained by taking prompts in four dimensions as an example. The preset simulation scene prompt information includes: describing the static road condition information in the target simulation scene, describing the vehicle movement information in the target simulation scene, describing the movement information of other traffic participants in the target simulation scene, and describing the temporary operation information in the target simulation scene. At this time, after the simulation scene generation device obtains the user description information of the target simulation scene, in order to improve the accuracy of the target simulation scene description information, the user description information of the target simulation scene can be used as the input of the scene description model and input into the scene description model. At this time, after the scene description model receives the user description information, it can extract the description information of the target simulation scene in the user description information, and under the prompt of the preset simulation scene prompt information, the extracted description information is supplemented in the above four dimensions to obtain the final scene description information, wherein the final scene description information is described from the above four dimensions. The simulation scene generation device can obtain the scene description information of the target simulation scene output by the scene description model.

[0066] In some embodiments, if the user description information input by the user has already described the target simulation scene from the four dimensional directions prompted by the preset simulation scene prompt information, the scene description model can directly obtain the scene description information in the four dimensional directions prompted by the preset simulation scene prompt information from the user description information, and thus can directly obtain the scene description information of the target simulation scene from the user description information.

[0067] If the user description information of the target simulation scene input by the user does not describe the target simulation scene in the four dimensional directions prompted by the preset simulation scene prompt information, the scene description model needs to extract the description of the target simulation scene based on the user description information, and supplement the extracted description of the target simulation scene in the four dimensional directions prompted by the preset simulation scene prompt information to obtain the final scene description information that describes the target simulation scene in the four dimensional directions prompted by the preset simulation scene prompt information.

[0068] It should be noted that when the scene description model supplements the description of the extracted target simulation scene, there may be multiple possibilities for the content to be supplemented. The scene description model can select one possibility from the multiple possibilities of the content to be supplemented as the supplementary content. At this time, the scene description model outputs scene description information of a target simulation scene. Of course, the scene description model can also supplement each possible possibility of the content to be supplemented or at least two possibilities among the multiple possibilities of the content to be supplemented as supplementary content. At this time, the scene description model outputs scene description information of a target simulation scene for each supplementary content, so at least two scene description information can be output, and at least two scene description information are different. This application does not limit this.

[0069] In this way, in the embodiment of the present application, by utilizing the scene description model, it is possible to accurately identify and understand the user's needs for the target simulation scene, and by pre-setting the simulation scene prompt information, it is possible to infer an accurate and specific scene instance description, and guide the scene recognition model to output the required scene description information of the target simulation scene, with a high degree of intelligence and good scene controllability.

[0070] As a possible implementation, the type of the user description information of the target simulation scene includes at least one of a text type, a voice type, and an image type. That is, the type of the user description information of the target simulation scene input by the user can be one or more types, such as Figure 2a As shown. For example, the acquired user description information may include fuzzy text type description information and image type description information, and then both types of description information are input into the simulation scene generation device. In this way, the simulation scene generation device can acquire multimodal user description information, improve the accuracy of the target simulation scene, and reduce the requirements for user input, thereby improving the construction efficiency.

[0071] As a possible implementation, the user description information of the target simulation scene at least includes user description information of the image type, which can improve the accuracy of generating subsequent static topology graphs in the target simulation scene, thereby improving the accuracy of constructing the target simulation scene.

[0072] It should be noted that the type of user description information of the target simulation scenario may also be other types, and this application does not impose any limitation on this.

[0073] For example, if the user description information of the target simulation scene input by the user is a text description, and the content is: intersection, the car goes straight, and the private car merges slowly, and the user also inputs the user description information of the image type, such as Figure 2b and Figure 3aAs shown. In this example, the image-type user description information input by the user can be called a sketch. At this time, the simulation scene generation device can obtain two types of user description information. The simulation scene generation device inputs both types of user description information it obtains into the scene description model. The scene description model receives the text-type user description information and the image-type user description information, and can extract the description of the target simulation scene from the text-type user description information and the image-type user description information, and supplement the extracted description of the target simulation scene in the four dimensional directions prompted by the preset simulation scene prompt information, so that the scene description information of the final target simulation scene can be output. Assume that the scene description information of the target simulation scene output by the scene description model is as follows: the static road condition information in the target simulation scene is described as: an irregular two-lane intersection; the movement information of the vehicle in the target simulation scene is described as: the vehicle is in the first lane on the right, driving at a speed of 27km / h, and going straight through the intersection; the movement information of other traffic participants in the target simulation scene is described as: under smooth traffic conditions, a car turns right into the vehicle lane at a speed of 20km / h from the first lane on the right side of the intersection relative to the right side of the vehicle; the temporary operation information in the target simulation scene is described as: the vehicle passes through the initial lane with a green light, such as Figure 2c and Figure 3a As shown. Among them, through the text type user description information and the image type user description information, it can be known that the scene description model supplements the vehicle speed of the vehicle in the target simulation scene, supplements the speed of other traffic participants in the movement information of other traffic participants in the target simulation scene, and supplements the temporary operation information.

[0074] In some embodiments, after receiving user description information of the target simulation scene input by the user, the simulation scene generation device can first feature encode the user description information of the target simulation scene input by the user, obtain the feature information in the user description information, and input the feature information into the scene description model.

[0075] It should be noted that the above-mentioned scene description model is a pre-trained network model, and its training data may be user description information of a simulation scene input by a pre-set user, and description information of the simulation scene.

[0076] Step S102: Acquire a global map of the target area, and determine a target local map within the target area corresponding to the target simulation scene based on the scene description information and the global map of the target area.

[0077] The global map of the target area refers to a map containing all locations in the target area. The local target map is a partial map of the global map of the target area. The target area is the area corresponding to the pre-set target simulation scenario.

[0078] In an embodiment of the present application, after the simulation scene generation device obtains the scene description information of the target simulation scene, it is necessary to associate it with the target area. At this time, the simulation scene generation device can obtain a global map of the target area, for example, it is necessary to conduct an automatic driving test of a vehicle in area A. At this time, the simulation scene generation device needs to generate a simulation scene corresponding to area A. At this time, it is necessary to associate the scene description information of the acquired target simulation scene with the map of area A. Therefore, it is necessary to first obtain a global map of the target area. In some embodiments, the global map of the target area can be preset by the user, and the map can be a high-precision map, which can improve the accuracy of the target simulation scene. After obtaining the global map of the target area, the simulation scene generation device can determine the regional map that matches the scene description information in the global map of the target area according to the scene description information of the target simulation scene, and use the determined regional map as the target local map in the target area corresponding to the target simulation scene.

[0079] As a possible implementation method, in order to reduce the workload of the simulation scene generation device and improve the accuracy of constructing the target simulation scene, the corresponding static topology map can be first determined based on the scene description information of the target simulation scene, so as to determine the target local map corresponding to the target simulation scene in the global map of the target area according to the static topology map. At this time, based on the scene description information of the target simulation scene and the global map of the target area, determining the target local map in the target area corresponding to the target simulation scene includes: based on the scene description information of the target simulation scene, obtaining the static topology map corresponding to the scene description information; associating the static topology map with the global map, and determining the target local map in the target area corresponding to the target simulation scene.

[0080] The static topology map refers to a static road condition map generated according to the scene description information of the target simulation scene.

[0081] In an embodiment of the present application, the simulation scene generation device can generate a static topology map that is consistent with the scene description information according to the scene description information of the target simulation scene. At this time, the static topology map contains the road condition information in the scene description information of the target simulation scene information. That is, the simulation scene generation device can generate a road condition map described by the path information according to the road condition information in the scene description information of the target simulation scene, that is, generate a static topology map. After obtaining the static topology map, the static topology map is associated with the global map of the target area, and the target local map that meets the requirements of the target simulation scene is determined in the global map. That is, the target local map in the target area corresponding to the target simulation scene is determined. In an embodiment of the present application, the target local map that is consistent with the scene description information can be determined in the global map according to the scene description information of the acquired target simulation scene, thereby improving the diversity of the target local map and thereby improving the generalization of the generated simulation scene.

[0082] As a possible implementation, in order to improve the accuracy of determining the target local map and reduce the processing complexity of the simulation scene generation device, the scene description information based on the target simulation scene and obtaining the static topology map corresponding to the scene description information include:

[0083] The scene description information of the target simulation scene is used as the input of the topology completion model, and a static topology map corresponding to the scene description information is obtained as the output of the topology completion model.

[0084] Among them, the topology completion model is pre-trained and is used to complete the road condition map according to the input scene description information, and output a static topology map corresponding to the scene description information.

[0085] That is, the network model can be pre-trained to obtain a topology completion model. The topology completion model can generate a static topology map corresponding to the scene description information when receiving the scene description information of the simulation scene. At this time, the simulation scene generation device can input the scene description information of the target simulation scene into the topology completion model. After the topology completion model receives the scene description information, it can draw a road condition map according to the scene description information, thereby obtaining a static topology map. In some embodiments, the topology completion model can infer and supplement the road condition map only based on the road condition information in the scene description information of the target simulation scene to obtain a static topology map.

[0086] In order to improve the accuracy of the static topology map output by the topology completion model, the topology completion model can refer to the user description information of the image type in addition to the scene description information of the target simulation scene when generating the static topology map. At this time, the scene description information of the target simulation scene is used as the input of the topology completion model, and the static topology map corresponding to the scene description information output by the topology completion model includes:

[0087] When the type of user description information of the target simulation scene includes an image type, the image type user description information and the scene description information of the target simulation scene are used as inputs of the topology completion model to obtain a static topology map corresponding to the scene description information output by the topology completion model.

[0088] In an embodiment of the present application, when the user description information of the target simulation scene includes an image type, the user description information of the target simulation scene of the image type and the scene description information of the target simulation scene can be used as inputs of a topology completion model. In this way, the topology completion model can use the scene description information and the user description information of the image type to supplement the road condition map in the target simulation scene. In some embodiments, the topology completion model can automatically supplement some details in the road conditions, such as lane information, zebra crossings, road markings, etc., thereby outputting a corresponding static topology map.

[0089] As described in the above example, the user description information of the target simulation scene input by the user includes image type and text type. Figure 2b and Figure 3a As shown. Figure 3a , Figure 3b and Figure 4 As shown, after the simulation scene generation device obtains the scene description information of the target simulation scene, it can Figure 2b The user description information of the target simulation scene of the image type shown and the scene description information of the target simulation scene are input into the topology completion model. The topology completion model draws the road condition map in the target simulation scene according to the scene description information of the target simulation scene and the user description information of the target simulation scene of the image type, and supplements some details of the road condition, and outputs a static topology map corresponding to the description information, such as Figure 3b and 4 After obtaining the static topology map, the static topology map can be associated with the global map to determine the target local map in the target area corresponding to the target simulation scenario. Figure 3a , Figure 3b and Figure 3c shown.

[0090] In some embodiments, in order to reduce the complexity of the topology completion model, the scene description information of the target simulation scene and the user description information of the target simulation scene of the image type can be first converted into information that the topology completion model can recognize. That is, the simulation scene generation device can first encode the scene description information of the target simulation scene and the user description information of the target simulation scene of the image type to obtain a feature vector, wherein the feature vector is information that the topology completion model can recognize. The simulation scene generation device performs feature splicing on the feature vector of the scene description information of the target simulation scene and the feature vector of the user description information, and uses the spliced ​​feature vector as the input of the topology completion model. Alternatively, the simulation scene generation device can also directly use the feature vector of the scene description information of the target simulation scene and the feature vector of the user description information as the input of the topology completion model, respectively, and input them into the topology completion model.

[0091] It should be noted that the above topology completion model can be a pre-trained neural network generation model, such as a GAN (Generative Adversarial Nets, Generative Adversarial Network) model, a VAE (Variational Auto-Encoder) model or a DM (Diffusion Model) model, etc. The training data can be the description information of the simulation scene of the image type or the description information of the simulation scene and the user description information, as the input data of the network model, and the static topology map corresponding to the pre-set description information of the simulation scene is used as the output data of the network model.

[0092] After obtaining the static topology map, the simulation scene generation device can associate the static topology map with the global map of the target area, thereby determining the road condition map of the target simulation scene in the global map of the target area. At this time, the above-mentioned associating the static topology map corresponding to the description information with the global map of the target area to determine the target local map in the target area corresponding to the target simulation scene includes:

[0093] According to the static topology map corresponding to the description information, a target area map with the same road category as that in the static topology map corresponding to the scene description information is screened out in the global map of the target area; the similarity between the screened target area map and the static topology map is calculated; based on the similarity between the target area map and the static topology map, a target local map within the target area corresponding to the target simulation scene is determined.

[0094] In an embodiment of the present application, after the simulation scene generation device determines the static topology map corresponding to the scene description information, the part with the same road category as the static topology map can be found in the global map of the target area according to the road category in the static topology map, which is the available target area map. After the target area map is screened out, it is necessary to find a map similar to the static topology map in the screened target area map, which can be achieved by calculating the similarity between the screened target area map and the static topology map. In some embodiments, the simulation scene generation device can calculate the cosine similarity between the screened target area map and the static topology map to determine the similarity between the two. In some embodiments, the simulation scene generation device can also calculate the similarity between the screened target area map and the static topology map by a hash algorithm. After calculating the similarity between the screened target area map and the static topology map, the target local map in the target area corresponding to the target simulation scene can be determined according to the similarity between the screened target area map and the static topology map. In some embodiments, the target area map whose similarity with the static topology map exceeds a preset similarity threshold can be determined as the target local map in the target area corresponding to the target simulation scene.

[0095] It should be understood that in the embodiment of the present application, the similarity between the target area map and the static topology map may also be calculated by other algorithms, and the present application does not limit this.

[0096] In some embodiments, in order to improve the accuracy of the target local map in the target area corresponding to the determined target simulation scenario, the above-mentioned determination of the target local map in the target area corresponding to the target simulation scenario based on the similarity between the screened target area map and the static topology map includes:

[0097] Based on the similarity between the filtered target area map and the static topology map corresponding to the scene description information, the target area map with the highest similarity to the static topology map is determined as the target local map in the target area corresponding to the target simulation scene.

[0098] That is, the simulation scene generation device determines the similarity between the target area map and the static topology map. The higher the similarity, the more similar the target area map is to the static topology map. At this time, the possibility that the target area map is the target local map of the target simulation scene required by the user is greater. Based on this, in order to improve the accuracy of the target simulation scene construction, the target area map with the highest similarity to the static topology map can be determined as the target local map in the target area corresponding to the target simulation scene. Among them, when there are multiple target area maps with the highest similarity to the static topology map, multiple target area maps with the highest similarity can be determined as the target local map in the target area corresponding to the target simulation scene. It is also possible to sort multiple target area maps according to the similarity between the target area map and the static topology map, and determine the target area map with a similarity greater than a preset threshold as the target local map.

[0099] As a possible implementation method, the target local map in the target area corresponding to the target simulation scene can be determined according to the result selected by the user from the filtered target area map. At this time, the above-mentioned determination of the target local map in the target area corresponding to the target simulation scene based on the similarity between the filtered target area map and the static topology map includes: outputting to the user the similarity between the filtered target area map and the static topology map corresponding to the description information; in response to the user's selection operation on the target area map, determining the target area map selected by the user as the target local map in the target area corresponding to the target simulation scene.

[0100] That is, after determining the similarity between the filtered target area map and the static topology map corresponding to the description information, the simulation scene generation device outputs the similarity between the filtered target area map and the static topology map corresponding to the description information to the user. In some embodiments, the similarity between the filtered target area map and the static topology map corresponding to the description information can be output to the user in the order of similarity from high to low (or from low to high), or the sorted target area map can be output to the user. The user can determine the target area map that is closest to the road condition map of the target simulation scene that the user needs to build in the filtered target area map based on the similarity between the filtered target area map and the static topology map corresponding to the description information or the sorted target area map. At this time, the simulation scene generation device responds to the user's selection operation on the target area map and determines the target area map selected by the user as the target local map in the target area corresponding to the target simulation scene. In some embodiments, the user can select multiple target area maps, and at this time the simulation scene generation device can determine the multiple target area maps selected by the user as the target local map in the target area corresponding to the target simulation scene.

[0101] In this way, the simulation scene generation device can first automatically complete and generate a static topology map corresponding to the scene description information based on the scene description information of the target simulation scene, and find the target local map in the target area corresponding to at least one target simulation scene with the highest similarity in the global map of the target area, thereby realizing scene generation under diverse topologies and further improving the efficiency of target simulation scene construction.

[0102] Step S103: Generate a target simulation scene based on the target local map and scene description information.

[0103] In an embodiment of the present application, after obtaining the target local map, the simulation scene generation device can determine the movement trajectory of the traffic participants according to the scene description information, and then combine the movement trajectory of the traffic participants with the target local map to generate a target simulation scene.

[0104] In a possible implementation, generating a target simulation scene based on the target local map and scene description information includes:

[0105] Based on the target local map and scene description information, the initial driving state of the vehicle and the moving trajectory information of other traffic participants in the target simulation scene are generated; the preset road boundary rules are obtained, and the target moving trajectory information that meets the preset road boundary rules is screened out from the moving trajectory information of other traffic participants in the target simulation scene; based on the initial driving state of the vehicle and the target moving trajectory information of other traffic participants in the target simulation scene, the target simulation scene is generated.

[0106] Among them, other traffic participants refer to other traffic participants in the target simulation scene except the vehicle. The initial driving state information of the vehicle refers to the state information of the vehicle at the initial moment in the target simulation scene. The moving trajectory information of other traffic participants refers to the driving state of other traffic participants at each moment in the T moments when they move in the target simulation scene. T can be a pre-set length of time that other traffic participants exist in the target simulation scene. T is an integer greater than 0.

[0107] In the embodiment of the present application, after the simulation scene generation device determines the target local map in the target area corresponding to the target simulation scene, since the target local map in the target area corresponding to the target simulation scene only contains road condition information and does not contain the movement information of the vehicle and other traffic participants, it is necessary to combine the movement information of the vehicle and other traffic participants with the target local map in the target area corresponding to the target simulation scene. At this time, the simulation scene generation device can generate the initial state information of the vehicle and the movement trajectory information of other traffic participants in the target simulation scene according to the target local map in the target area corresponding to the target simulation scene and the scene description information of the target simulation scene. For example, according to the target local map in the target area corresponding to the target simulation scene and the scene description information of the target simulation scene, the driving position, driving speed, driving direction, acceleration and other state information of the vehicle at the initial moment of the target simulation scene are generated; and the driving state of other traffic participants at T moments in the target simulation scene, wherein the driving state at each moment may include the driving position, driving speed, driving direction, acceleration and other information of the participant. In some embodiments, the simulation scene generation device can generate the initial state information of the vehicle and the movement trajectory information of other traffic participants in the target simulation scene based on the movement information of the traffic participants in the scene description information of the target simulation scene. For example, the simulation scene generation device can estimate the movement trajectory information of other traffic participants in the target simulation scene from the initial moment to the T moment, as well as the initial state information of the vehicle at the initial moment, based on the movement information of the traffic participants in the scene description information of the target simulation scene. At this time, the simulation scene generation device can determine the position, speed, movement direction, acceleration, etc. of other traffic participants based on the movement information of the traffic participants in the scene description information of the target simulation scene, and then estimate the movement trajectory information of other traffic participants in the target simulation scene from the initial moment to the T moment. Wherein, T is an integer greater than 0.

[0108] As a possible implementation method, in order to reduce the working difficulty of the simulation scene generation device, the target local map within the target area corresponding to the target simulation scene and the scene description information of the target simulation scene can be used as the input of the interactive behavior generation model to obtain the initial state information of the vehicle and the movement trajectory information of other traffic participants in the target simulation scene output by the interactive behavior generation model.

[0109] Among them, the interactive behavior generation model is pre-trained and is used to process the target local map in the target area and the scene description information of the target simulation scene corresponding to the input target simulation scene, and obtain the initial state information of the vehicle and the movement trajectory information of other traffic participants in the target simulation scene.

[0110] It should be noted that the interactive behavior generation model is a pre-trained neural network generation model, such as a GAN model, a VAE model or a DM model, and its training data can be the movement trajectory information of the vehicle and other traffic participants, as well as the target local map corresponding to the simulation scene in the target area and the scene description information of the simulation scene. For example, when the interactive behavior generation model is a DM model, its input training data is the scene description information of the simulation scene, the target local map corresponding to the simulation scene in the target area, etc., and the output is the trajectory sequence information of the vehicle and traffic participants. The movement trajectory information of the vehicle and other traffic participants can be the driving state sequence of the vehicle from the initial to time T, the state sequence of other traffic participants, and the target local map of the trajectory embedded in the simulation scene in the target area, refer to Figure 5a As shown. That is, Figure 5a An example diagram showing the target local map combined with the initial state information of the vehicle and the moving trajectory information of other traffic participants in the target simulation scene. The driving state mainly includes information such as position, speed, and heading angle. For example, the state of the vehicle at time t is S e,t ={x et ,y et ,v et ,θ et}, and the state of the i-th other traffic participant is S oi,t ={x oi,t ,y oi,t ,v oi,t ,θ oi,t}.x et ,y et represents the position of the vehicle at time t, v et represents the speed of the vehicle at time t, θ et Represents the heading angle of the vehicle at the tth moment. oi,t ,y oi,t represents the position of the i-th other traffic participant at time t, v oi,t , represents the speed of the i-th other traffic participant at time t, θ oi,t represents the heading angle of the i-th other traffic participant at the t-th time. i is an integer greater than 0. After the interactive behavior generation model is trained, the movement trajectory information of other traffic participants can be output according to the target local map in the target area corresponding to the input target simulation scene and the scene description information of the target simulation scene. Figure 5b In some embodiments, the interactive behavior generation model is guided by the mechanism to denoise, and the movement trajectory information of other traffic participants is sampled and restored.

[0111] Since the movement information of the traffic participants in the scene description information of the target simulation scene does not describe the movement trajectory of other traffic participants at each moment, in order to improve the accuracy of the target simulation model, corresponding movement trajectory information can be generated for each possible movement trajectory of other traffic participants, so that the simulation scene generation device can obtain at least one movement trajectory information of other traffic participants in the target simulation scene. After the simulation scene generation device obtains at least one movement trajectory information of other traffic participants in the target simulation scene, since the movement trajectory of other traffic participants in the target simulation scene may not meet the road boundary constraint condition, for example, the movement trajectory of other traffic participants in the target simulation scene is outside the road boundary, that is, the vehicle has driven out of the road boundary, then this movement trajectory information is obviously not in line with the norm. Therefore, after obtaining at least one movement trajectory information of other traffic participants in the target simulation scene, it is necessary to determine the movement trajectory information that meets the norm in at least one movement trajectory information of other traffic participants in the target simulation scene. At this time, the simulation scene generation device obtains the preset road boundary rule, and selects the movement trajectory information that meets the preset road boundary rule from the movement trajectory information of other traffic participants in the target simulation scene as the target movement trajectory information of other traffic participants in the target simulation scene. The simulation scene generation device can detect whether the movement trajectory of other traffic participants at each moment recorded in each movement trajectory information of the target simulation scene meets the preset road boundary rules according to the preset road boundary rules. Thus, the movement trajectory information that meets the preset road boundary rules can be screened out as the target movement trajectory information of other traffic participants in the target simulation scene.

[0112] In some embodiments, in order to quickly determine the target moving trajectory information of other traffic participants, the simulation scene generation device may calculate the moving trajectory information of each other traffic participant according to the formula: The metric score that meets the road boundary constraints is calculated, and then the target movement trajectory information of other traffic participants is determined based on the metric score. i,t ,y i,t is the trajectory coordinate of the i-th other traffic participant at time t, is the boundary coordinate of the road / lane, ε is the allowable error, J off-road Score the metric for compliance with the road boundary constraints for the i-th other traffic participant.

[0113] In this way, the simulation scene generation device can calculate the metric score of compliance with the road boundary constraints corresponding to each mobile trajectory information of each other traffic participant, so that for each other traffic participant, from the metric score of compliance with the road boundary constraints corresponding to each mobile trajectory information of the traffic participant, the mobile trajectory information whose corresponding metric score of compliance with the road boundary constraints is less than a preset score threshold can be selected as the target mobile trajectory information of the other traffic participant.

[0114] After determining the initial state information of the vehicle and the target movement trajectory information of other traffic participants in the target simulation scene, the simulation scene generation device can combine the initial state information of the vehicle and the target movement trajectory information of other traffic participants in the target simulation scene with the target local map in the target area corresponding to the target simulation scene to generate the target simulation scene.

[0115] As a possible implementation method, the initial state information of the vehicle includes at least one of the position information, speed information and heading angle information of the vehicle at the initial moment in the target simulation scene; the movement trajectory information includes the driving state at T moments, wherein the driving state at each moment includes at least one of the position information, speed information and heading angle information, and T is a preset integer greater than 0.

[0116] When testing autonomous driving, the acquired simulation scenario can be directly used as a test case and loaded into the simulator, such as Figure 6a As shown. At this point, in this step, the target simulation scene can be directly generated. The behavior of the self-driving car is controlled by the autonomous driving system to be tested. In the target simulation scene, the surrounding environment and dynamic obstacles interact with the self-driving car, thereby testing the use effect of the autonomous driving control system algorithm, performing version quality care and capability climbing, and further optimizing the capability boundary of the autonomous driving control system algorithm according to the use effect.

[0117] Alternatively, when testing autonomous driving, you can also use simulation scenario data for testing, such as Figure 6b As shown. At this point, in this step, you only need to determine the data information in the target simulation scene and save the data information directly. When testing, you can select the scenes with good performance of the regulatory control system as positive examples and the scenes with poor performance of the regulatory control system as negative examples after automatic evaluation item screening, and supplement the screened simulation data as the training data set of the regulatory control AI model to perform tuning training on the AI ​​regulatory control model, so as to obtain the tuned AI model parameters and use them for the control of autonomous driving vehicles.

[0118] As a possible implementation method, when generating the target simulation scene, the various data in the target simulation scene can also be determined. At this time, the initial state information of the vehicle and the target movement trajectory information of other traffic participants in the target simulation scene and the target local map in the target area corresponding to the target simulation scene can be directly stored as the data of the target simulation scene for subsequent use. At this time, the initial state information of the vehicle, the target movement trajectory information of other traffic participants in the target simulation scene, the target local map in the target area corresponding to the target simulation scene and other data can be stored as the data of the target simulation scene.

[0119] Thus, in the embodiment of the present application, after obtaining the scene description information of the target simulation scene, the target simulation scene required by the user can be generated based on the scene description information of the target simulation scene. This can effectively improve the efficiency of scene construction, eliminate the need for scene simulators and professional field knowledge of scene construction, lower the threshold for use, and improve ease of use.

[0120] See also Figure 7 A schematic diagram of the structure of a simulation scene generation device provided in an embodiment of the present application. Figure 7 As shown, the device comprises:

[0121] The acquisition unit 701 is used to acquire scene description information of a target simulation scene.

[0122] The scene description information of the target simulation scene is used to describe the movement information and road condition information of traffic participants.

[0123] The acquisition unit 701 is further used to acquire a global map of the target area.

[0124] The processing unit 702 is used to determine a target local map in a target area corresponding to the target simulation scene based on the scene description information and the global map.

[0125] The processing unit 702 is further configured to generate a target simulation scene based on the target local map and the scene description information.

[0126] As a possible implementation, the acquisition unit 701 is specifically used to acquire user description information of the target simulation scene and determine the scene description information according to the user description information. For example, the user description information of the target simulation scene can be used as the input of the scene description model to obtain the scene description information output by the scene description model.

[0127] The user description information refers to the description information of the target simulation scene by the user. The scene description model is pre-trained and is used to extract and supplement the information of the simulation scene from the input user description information and output the scene description information of the target simulation scene.

[0128] As a possible implementation manner, the movement information of the traffic participants includes the movement information of the vehicle and / or the movement information of other traffic participants; the road condition information includes: static road condition information and / or dynamic road condition information.

[0129] As a possible implementation manner, the type of the user description information of the target simulation scene includes at least one of a text type, a voice type, and an image type.

[0130] As a possible implementation method, the processing unit 702 is specifically used to obtain a static topology map corresponding to the scene description information based on the scene description information; associate the static topology map with the global map to determine the target local map corresponding to the target simulation scene in the target area.

[0131] As a possible implementation manner, the processing unit 702 is specifically configured to determine a static topology map corresponding to the scene description information according to the description information of the image type and the scene description information when the type of the user description information includes an image type.

[0132] For example, the scene description information can be used as the input of the topology completion model to obtain a static topology map corresponding to the scene description information output by the topology completion model. The topology completion model is pre-trained and is used to complete the road condition map according to the input scene description information and output a static topology map corresponding to the scene description information.

[0133] As a possible implementation method, the processing unit 702 is specifically used to filter out a target area map with the same road category as in the static topology map in the global map based on the static topology map; calculate the similarity between the filtered target area map and the static topology map; and based on the similarity, determine the target local map in the target area corresponding to the target simulation scenario.

[0134] As a possible implementation manner, the processing unit 702 is specifically configured to determine the target area map having the highest similarity to the static topology map as the target local map in the target area corresponding to the target simulation scenario.

[0135] As a possible implementation method, the processing unit 702 is specifically used to output to the user the similarity between the filtered target area map and the static topology map; in response to the user's selection operation on the target area map, the target area map selected by the user is determined as the target local map within the target area corresponding to the target simulation scene.

[0136] As a possible implementation method, the processing unit 702 is specifically used to generate the initial state information of the vehicle and the moving trajectory information of other traffic participants in the target simulation scene based on the target local map and the target scene description information; obtain the preset road boundary rules, and determine the target moving trajectory information that complies with the preset road boundary rules from the moving trajectory information of other traffic participants in the target simulation scene; generate the target simulation scene based on the initial state information of the vehicle and the target moving trajectory information of other traffic participants.

[0137] As a possible implementation, the initial state information of the vehicle includes at least one of the position information, speed information and heading angle information of the vehicle at the initial moment in the target simulation scene. The movement trajectory information includes the driving state at T moments, wherein the driving state at each of the T moments includes at least one of the position information, speed information and heading angle information. T is a preset integer greater than 0.

[0138] Corresponding to the above embodiments, the present application also provides an electronic device. Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, the electronic device 800 may include: a processor 801, a memory 802 and a communication unit 803. These components communicate through one or more buses, and those skilled in the art can understand that the structure of the electronic device shown in the figure does not constitute a limitation on the embodiment of the present invention, it can be a bus structure or a star structure, and can also include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0139] The communication unit 803 is used to establish a communication channel so that the electronic device can communicate with other devices, receive user data sent by other devices or send user data to other devices.

[0140] The processor 801 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. It runs or executes software programs, instructions, and / or modules stored in the memory 802, and calls data stored in the memory to perform various functions of the electronic device and / or process data. The processor can be composed of an integrated circuit (IC), for example, it can be composed of a single packaged IC, or it can be composed of multiple packaged ICs with the same or different functions. For example, the processor 801 can only include a central processing unit (CPU). In an embodiment of the present invention, the CPU can be a single computing core or multiple computing cores.

[0141] The memory 802 is used to store the execution instructions of the processor 801. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0142] When the execution instruction in the memory 802 is executed by the processor 801, the electronic device 800 can execute Figure 1 Some or all of the steps in the illustrated embodiments.

[0143] In a specific implementation, the present invention further provides a computer storage medium, wherein the computer storage medium may store a program, and when the program is executed, the program may include some or all of the steps in each embodiment of the simulation scene generation method provided by the present invention. The storage medium may be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0144] In a specific implementation, the present invention further provides a computer program product, wherein the computer program product includes executable instructions, and when the executable instructions are executed on a computer, the computer executes part or all of the steps in each embodiment of the simulation scene generation method provided by the present invention.

[0145] Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution in the embodiments of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a disk, an optical disk, etc., and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention or some parts of the embodiments.

[0146] In this specification, the same or similar parts between the various embodiments can be referred to each other. In particular, for the device embodiment and the terminal embodiment, since they are basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description in the method embodiment.

Claims

1. A simulation scene generation method, characterized in that: include: Acquire scene description information of a target simulation scene, wherein the scene description information is used to describe movement information and road condition information of traffic participants; Acquire a global map of the target area, and determine a target local map within the target area corresponding to the target simulation scene based on the scene description information and the global map; The target simulation scene is generated based on the target local map and the scene description information.

2. The method according to claim 1, characterized in that The movement information of the traffic participants includes the movement information of the vehicle itself and / or the movement information of other traffic participants; the road condition information includes: static road condition information and / or dynamic road condition information.

3. The method according to claim 1 or 2, characterized in that: The step of acquiring the scene description information of the target simulation scene includes: Acquire user description information of the target simulation scenario, where the user description information is description information of the target simulation scenario by the user; The scenario description information of the target simulation scenario is determined according to the user description information.

4. The method according to claim 3, characterized in that The type of the user description information includes at least one of a text type, a voice type and an image type.

5. The method according to any one of claims 1 to 4, characterized in that: The determining, based on the scene description information and the global map, a target local map within the target area corresponding to the target simulation scene comprises: Based on the scene description information, determining a static topology map corresponding to the scene description information; The static topology map is associated with the global map to determine a target local map within the target area corresponding to the target simulation scenario.

6. The method according to claim 5, characterized in that The determining, based on the scene description information, a static topology map corresponding to the scene description information includes: When the type of the user description information includes description information of an image type, a static topology map corresponding to the scene description information is determined according to the description information of the image type and the scene description information.

7. The method according to claim 5 or 6, characterized in that: The step of associating the static topology map with the global map to determine a target local map within the target area corresponding to the target simulation scenario includes: According to the static topological map, a target area map having the same road category as that in the static topological map is screened out in the global map; Calculating the similarity between the target area map and the static topology map; Based on the similarity, a target local map within the target area corresponding to the target simulation scene is determined.

8. The method according to claim 7, characterized in that The determining, based on the similarity, a target local map within the target area corresponding to the target simulation scene comprises: The target area map having the highest similarity to the static topology map is determined as the target local map in the target area corresponding to the target simulation scenario.

9. The method according to claim 7, characterized in that: The determining, based on the similarity, a target local map within the target area corresponding to the target simulation scene comprises: Outputting to the user the similarity between the filtered target area map and the static topology map; In response to a user's selection operation on a target area map, the target area map selected by the user is determined as a target local map within the target area corresponding to the target simulation scene.

10. The method according to any one of claims 1 to 9, characterized in that: The generating the target simulation scene based on the target local map and the scene description information comprises: Based on the target local map and the scene description information, generate initial state information of the vehicle and movement trajectory information of other traffic participants; Acquire a preset road boundary rule, and determine target moving trajectory information that meets the preset road boundary rule from the moving trajectory information of the other traffic participants; The target simulation scene is generated based on the initial state information of the vehicle and the target movement trajectory information of the other traffic participants.

11. The method according to claim 10, characterized in that The initial state information of the vehicle includes at least one of position information, speed information and heading angle information of the vehicle at an initial moment in the target simulation scene; The movement trajectory information of the other traffic participants includes the driving status at T moments, wherein the driving status at each of the T moments includes at least one of the position information, speed information and heading angle information of the other traffic participants, and T is a pre-set integer greater than 0.

12. A simulation scene generation device, characterized in that: include: An acquisition unit, used to acquire scene description information of a target simulation scene, wherein the scene description information is used to describe movement information and road condition information of traffic participants; The acquisition unit is further used to acquire a global map of the target area; A processing unit, configured to determine a target local map in the target area corresponding to the target simulation scene based on the scene description information and the global map; The processing unit is further used to generate the target simulation scene based on the target local map and the scene description information.

13. An electronic device, characterized in that: The electronic device comprises a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the electronic device executes the method according to any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 11.

15. A computer program product, characterized in that The computer program product comprises executable instructions, and when the executable instructions are executed on a computer, the computer is caused to perform the method according to any one of claims 1 to 11.

Images