Scene deployment method and device for unmanned configuration parameters, equipment and medium

Abstract

Embodiments of the present application disclose a scene deployment method and device for unmanned vehicle configuration parameters, equipment and medium. The method comprises: screening at least one target scene matched with a to-be-deployed scene from a scene library comprising multiple scenes according to a preset scene screening rule, each scene comprising at least one scene tag, and the configuration parameters corresponding to each scene being the configuration parameters of an unmanned vehicle running in the scene; determining an initial configuration parameter set of the unmanned vehicle according to a configuration parameter set corresponding to each target scene; and initializing and deploying the unmanned vehicle based on the initial configuration parameter set. The technical solution of the embodiments of the present application provides a new scene deployment method for unmanned vehicle configuration parameters, and improves the deployment speed and efficiency of unmanned vehicle configuration parameters.

Description

Technical Field

[0001] The embodiments of the present invention relate to computer technology, and more particularly to autonomous driving technology, specifically to a scenario deployment method, apparatus, computer equipment, and storage medium for autonomous driving configuration parameters. Background Technology

[0002] Against the backdrop of the national "new infrastructure" initiative, autonomous driving has garnered significant attention as a typical application scenario. Typically, when deploying an autonomous vehicle in a specific scenario, it's necessary to first initialize and deploy its various autonomous driving configuration parameters (typically, the model parameters of the autonomous driving algorithm and hardware parameters). Following this initialization, extensive testing is conducted on the vehicle in that scenario. This testing includes simulation testing and real-vehicle testing. The testing process involves adjusting and optimizing the algorithm's model and parameters, as well as the hardware parameters. The algorithm and hardware parameters determined after testing constitute the autonomous driving configuration parameters for the vehicle's operation in that scenario.

[0003] Suppose that for the deployment scenario, there is a set of algorithm and hardware parameters that are perfectly matched, called the adaptation values. The initial values ​​of the algorithm, hardware, and other parameters at the start of the initial deployment may be far from the adaptation values ​​of the deployment scenario. As a result, the testing process based on the initial values ​​may exhibit many unexpected behaviors, be very time-consuming, or even fail to converge to the adaptation values ​​no matter how they are adjusted.

[0004] Therefore, how to initialize and deploy the autonomous driving configuration parameters to a suitable initial value before testing has a decisive impact on finding the appropriate value later. However, existing technologies do not have an effective way to deploy the initial value of autonomous driving configuration parameters. Summary of the Invention

[0005] This invention provides a method, apparatus, device, and medium for scenario deployment of autonomous driving configuration parameters, thereby providing a new method for scenario deployment of autonomous driving configuration parameters and improving the deployment efficiency and speed of autonomous driving configuration parameters.

[0006] In a first aspect, embodiments of the present invention provide a scenario deployment method for autonomous driving configuration parameters, the method comprising:

[0007] According to the preset scenario filtering rules, at least one target scenario that matches the scenario to be deployed is selected from the scenario library. The scenario library includes multiple scenarios, each scenario includes at least one scenario tag, and the configuration parameters corresponding to each scenario are the configuration parameters of the unmanned vehicle running in that scenario.

[0008] The initial configuration parameter set of the unmanned vehicle is determined based on the configuration parameter set corresponding to each target scenario.

[0009] The unmanned vehicle is initially deployed based on the set of initial configuration parameters.

[0010] Furthermore, the scene tags include: scene location attribute tags;

[0011] According to preset scenario filtering rules, at least one target scenario matching the scenario to be deployed is selected from the scenario library, including:

[0012] In the scenario library, a set of target scenarios that match the scenario location attribute tags to be deployed are selected, wherein the set of target scenarios includes at least one target scenario.

[0013] Furthermore, the scene tags also include: tags for factors influencing autonomous driving algorithms;

[0014] After filtering the target scene set in the scene library to find scenes whose geographic attribute tags match the scenes to be deployed, it also includes:

[0015] In the set of target scenarios, at least one target scenario is selected that matches the autonomous driving algorithm influencing factor label with the scenario to be deployed, and the selected target scenario is used to update the set of target scenarios.

[0016] Furthermore, the scene tags also include: tags indicating the usage of the autonomous driving function of the unmanned vehicle in the scene;

[0017] After filtering the target scene set in the scene library to find scenes whose geographic attribute tags match the scenes to be deployed, it also includes:

[0018] In the target scenario set, at least one target scenario is selected where the autonomous driving function usage tag of the unmanned vehicle in the scenario matches the scenario to be deployed, and the selected target scenario is used to update the target scenario set.

[0019] Furthermore, the scene tags also include: tags for factors influencing autonomous driving algorithms;

[0020] From the set of target scenarios, at least one target scenario is selected where the autonomous driving function usage tag of the unmanned vehicle in the scenario matches the scenario to be deployed. After updating the set of target scenarios with the selected target scenarios, the process further includes:

[0021] In the set of target scenarios, at least one target scenario is selected that matches the autonomous driving algorithm influencing factor label with the scenario to be deployed, and the selected target scenario is used to update the set of target scenarios.

[0022] Furthermore, the labels for influencing factors of the autonomous driving algorithm include: environmental factor labels;

[0023] From the set of target scenarios, at least one target scenario is selected that matches the autonomous driving algorithm influencing factor tags with the scenario to be deployed, and the set of target scenarios is updated using the selected target scenarios, including:

[0024] In the set of target scenarios, at least one target scenario whose environmental factor label matches the scenario to be deployed is selected, and the set of target scenarios is updated using the selected target scenario.

[0025] Furthermore, the autonomous driving algorithm influencing factor labels also include: vehicle factor labels;

[0026] After selecting at least one target scenario from the target scenario set that matches the environmental factor tag of the scenario to be deployed, and updating the target scenario set with the selected target scenario, the process further includes:

[0027] In the target scenario set, at least one target scenario whose vehicle factor tag matches the scenario to be deployed is selected, and the selected target scenario is used to update the target scenario set.

[0028] Furthermore, the autonomous driving algorithm's influencing factor labels also include: obstacle factor labels;

[0029] After selecting at least one target scenario from the target scenario set that matches the vehicle factor tag with the scenario to be deployed, and updating the target scenario set with the selected target scenario, the process further includes:

[0030] In the set of target scenarios, at least one target scenario whose obstacle factor label matches the scenario to be deployed is selected, and the set of target scenarios is updated using the selected target scenario.

[0031] Furthermore, from the set of target scenarios, at least one target scenario whose environmental factor tags match the scenario to be deployed is selected, including:

[0032] Obtain at least one autonomous driving association algorithm applicable to the scenario to be deployed, and obtain the algorithm impact factor function corresponding to each autonomous driving association algorithm;

[0033] Each algorithmic impact factor function is obtained by weighted summation of at least one algorithmic impact factor and its corresponding weight coefficient; the algorithmic impact factor corresponds to the label factor in the environmental factor label.

[0034] Based on the impact factor values ​​of the scenario to be deployed under each of the algorithm impact factors, calculate the objective function value corresponding to each of the algorithm impact factor functions;

[0035] Based on the safety weights of the influence factor functions of each algorithm, the values ​​of each objective function are weighted and summed to obtain the target algorithm safety degree corresponding to the deployment scenario;

[0036] In the set of target scenarios, obtain the scenario algorithm security degree corresponding to each target scenario;

[0037] The safety of the scenario algorithm is obtained by weighting the scenario function of each algorithm influence factor function and the scenario function value of each target scenario for each algorithm influence factor function. The scenario function value is determined by the tag factor and the corresponding tag factor value in the environmental factor label of each target scenario.

[0038] Calculate the first difference value between the security level of each scenario algorithm and the security level of the target algorithm;

[0039] Obtain the security screening scenarios where the first difference value is less than or equal to the first threshold value, and determine each of the security screening scenarios as the target scenario.

[0040] Furthermore, after determining each of the aforementioned security screening scenarios as the target scenario, the method further includes:

[0041] For each target scenario, perform the following further filtering operation:

[0042] Get the target scene being processed, and get a current scene function value from the target scene being processed;

[0043] In the scenario to be deployed, obtain the current target function value corresponding to the current scenario function value;

[0044] Calculate the second difference value between the current scene function value and the current objective function value;

[0045] If the second difference value is greater than the second threshold value, then the current processing target scenario is filtered out;

[0046] If the second difference value is less than or equal to the second threshold value, then return to the operation of obtaining a current scene function value in the target scene being processed, until all scene function values ​​in the target scene being processed are processed.

[0047] Furthermore, based on the set of configuration parameters corresponding to each target scenario, the initial configuration parameter set of the autonomous vehicle is determined, including:

[0048] If there are multiple target scenarios, then for each configuration parameter item of the unmanned vehicle, the corresponding multiple configuration parameter values ​​are extracted from each configuration parameter set.

[0049] The initial set of configuration parameters for the unmanned vehicle is determined based on the multiple configuration parameter values ​​corresponding to each configuration parameter item.

[0050] Furthermore, based on the set of configuration parameters corresponding to each target scenario, the initial configuration parameter set of the autonomous vehicle is determined, including:

[0051] If there are multiple target scenarios, then at least one safe and compatible scenario is selected from all target scenarios based on the numerical value of the scenario algorithm security level of each target scenario.

[0052] The initial configuration parameter set of the unmanned vehicle is determined based on the configuration parameter set corresponding to each safety adaptation scenario.

[0053] Furthermore, after calculating the second difference value between the current scene function value and the current objective function value, the method further includes:

[0054] Based on the second difference value, update the accumulated value of the second difference value corresponding to the target scenario being processed;

[0055] Based on the set of configuration parameters corresponding to each target scenario, the initial set of configuration parameters for the autonomous vehicle is determined, including:

[0056] If there are multiple target scenarios, then among the target scenarios, the similar adaptation scenario with the smallest cumulative second difference value is selected.

[0057] The set of configuration parameters corresponding to similar adaptation scenarios is determined as the initial configuration parameter set of the unmanned vehicle.

[0058] Furthermore, after initializing and deploying the unmanned vehicle based on the initial configuration parameter set, the process also includes:

[0059] The unmanned vehicle is controlled to conduct operational tests in the scenario to be deployed.

[0060] Furthermore, controlling the unmanned vehicle to perform operational tests in the scenario to be deployed includes:

[0061] If it is determined that only the scene location attribute tag is used, and at least one target scene matching the scene to be deployed is selected from the scene library, then the unmanned vehicle is controlled to perform software-in-the-loop test, hardware-in-the-loop test, vehicle-in-the-loop test and vehicle-on-the-road test in the scene to be deployed.

[0062] If the combined use scenario regional attribute label and autonomous driving algorithm influencing factor label are determined, or the combined use scenario regional attribute label and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, at least one target scenario matching the scenario to be deployed is selected in the scenario library, and the unmanned vehicle is controlled to conduct hardware-in-the-loop testing and real-vehicle road testing in the scenario to be deployed.

[0063] If the combined usage scenario's regional attribute label, autonomous driving algorithm influencing factor label, and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, and at least one target scenario matching the scenario to be deployed is selected from the scenario library, then the unmanned vehicle is controlled to conduct real-vehicle road testing in the scenario to be deployed.

[0064] Secondly, embodiments of the present invention also provide a scenario deployment device for unmanned driving configuration parameters, the device comprising:

[0065] The target scene filtering module is used to filter at least one target scene that matches the scene to be deployed in the scene library according to the preset scene filtering rules. The scene library includes multiple scenes, each scene includes at least one scene tag, and the configuration parameters corresponding to each scene are the configuration parameters of the unmanned vehicle running in that scene.

[0066] The initial configuration parameter set determination module is used to determine the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each target scenario.

[0067] The unmanned vehicle initialization and deployment module is used to initialize and deploy the unmanned vehicle based on the initial configuration parameter set.

[0068] Thirdly, embodiments of the present invention also provide an electronic device, the electronic device comprising:

[0069] One or more processors;

[0070] Storage device for storing one or more programs;

[0071] When the one or more programs are executed by the one or more processors, the one or more processors implement the scenario deployment method for autonomous driving configuration parameters as described in any embodiment of the present invention.

[0072] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements a scenario deployment method for autonomous driving configuration parameters as described in any embodiment of the present invention.

[0073] This invention addresses the problem of slow and inefficient parameter deployment caused by improper initialization when deploying autonomous driving configuration parameters in new scenarios. It employs a pre-defined scenario filtering rule to select at least one target scenario from a scenario library containing multiple scenarios; determines the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each target scenario; and performs initial deployment of the unmanned vehicle based on the initial configuration parameter set. This method provides a new scenario deployment method for unmanned driving configuration parameters, improving the deployment speed and efficiency of unmanned vehicle parameters. Attached Figure Description

[0074] Figure 1 A flowchart illustrating a scenario deployment method for autonomous driving configuration parameters provided in Embodiment 1 of the present invention;

[0075] Figure 2a A flowchart illustrating another scenario deployment method for unmanned driving configuration parameters provided in Embodiment 2 of the present invention;

[0076] Figure 2b This is a schematic diagram of a scene labeling system applicable to Embodiment 1 of the present invention;

[0077] Figure 3a A flowchart illustrating another scenario deployment method for autonomous driving configuration parameters provided in Embodiment 3 of the present invention;

[0078] Figure 3b This is a schematic diagram illustrating a method for selecting at least one target scenario whose environmental factor label matches the scenario to be deployed from the target scenario set, as provided in Embodiment 3 of the present invention.

[0079] Figure 3c This is a flowchart of the process in Embodiment 3 of the present invention, which describes the further screening of each target scenario after determining each security screening scenario as a target scenario.

[0080] Figure 4 This is a schematic diagram of the structure of a scene deployment device for unmanned driving configuration parameters provided in Embodiment 4 of the present invention;

[0081] Figure 5 This is a schematic diagram of the structure of an electronic device provided in Embodiment 5 of the present invention. Detailed Implementation

[0082] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0083] Example 1

[0084] Figure 1 This is a flowchart of a scenario deployment method for autonomous driving configuration parameters provided in Embodiment 1 of the present invention. This embodiment is applicable to the initial deployment of autonomous driving configuration parameters of an autonomous vehicle according to an actual driving scenario. The method can be executed by a scenario deployment device for autonomous driving configuration parameters. This device can be implemented by software and / or hardware, and can be integrated into a terminal or server. The method specifically includes the following steps:

[0085] S110. According to the preset scene filtering rules, select at least one target scene from the scene library that matches the scene to be deployed.

[0086] The scenario library can include multiple scenarios, each scenario can include at least one scenario tag, and the configuration parameters corresponding to each scenario are the configuration parameters of the autonomous vehicle running in that scenario.

[0087] Scene tags refer to information describing a scene or an autonomous vehicle operating within a scene, under a defined descriptive dimension. A scene tag can include a tag name and a tag value corresponding to that name. The tag name can be understood as the descriptive dimension, and the tag value can be the descriptive information under that dimension. For example, a scene tag in the form of "Scene Location Attribute: Airport" describes the scene's descriptive information under the scene location attribute dimension as: Airport.

[0088] Optionally, the descriptive information (or tag value) in a scene tag can be a single-level descriptive information or multiple-level descriptive information. For example, a scene tag in the form of "Scene Location Attribute: City Road - Downtown Area" is used to describe a scene as a downtown area in a city road under the description dimension of the scene location attribute.

[0089] In this embodiment, each scenario may include one or more scenario tags, and each scenario tag may have one or more levels of description dimensions. That is, it may have one or more tag names at different levels. For example, a scenario tag in the form of "Autonomous Driving Algorithm Influence Factors - Environmental Factors: Visibility = 50" is used to describe the description information under the description dimension of environmental factors in an autonomous driving algorithm influence factor as visibility = 50.

[0090] Different scene tags can have different scene levels, and different scene levels can correspond to different scene filtering orders. The above will be explained in detail later.

[0091] In an optional implementation of this embodiment, the scene label may include one or more of the following: scene location attribute label, autonomous driving algorithm influencing factor label, and autonomous driving function usage label of the unmanned vehicle in the scene.

[0092] As an example rather than a limitation, the scenario location attribute label can refer to a scenario label named "Scene Location Attribute," the autonomous driving algorithm influencing factor label can refer to a scenario label named "Autonomous Driving Algorithm Influencing Factor," and the autonomous driving function usage label of an unmanned vehicle in a scenario can refer to a scenario label named "Autonomous Driving Function Usage of an Unmanned Vehicle in a Scene."

[0093] The scene location attribute tag refers to a tag that describes the scene under the descriptive dimension of location attributes (such as the geographical area where the scene is located or the road attributes in the scene). Accordingly, the tag values ​​in the scene location attribute tag can be: highway, park, urban road - downtown area or urban road - urban-rural fringe area, etc.

[0094] The labels for influencing factors of autonomous driving algorithms refer to tags that describe a scene within the descriptive dimension of various influencing factors corresponding to the model and parameters of the autonomous driving algorithm. Accordingly, the label values ​​in these labels can be the factor values ​​of one or more specific influencing factors, such as light intensity = 100, visibility = 50, etc. Furthermore, the label values ​​in these labels can be further divided into label factors and label factor values. One label factor corresponds to one influencing factor, such as "light intensity" and "visibility" mentioned above; one label factor value corresponds to the factor value of one influencing factor, such as "100" and "50" mentioned above.

[0095] The "purpose label" for autonomous driving functions in a scenario refers to a label describing the scenario under the descriptive dimension of the autonomous driving function or purpose of the autonomous vehicle. Accordingly, the label values ​​in the autonomous driving function purpose label can be: precise parking or precise charging docking, etc.

[0096] In this embodiment, each scenario in the scenario library corresponds to a matching set of configuration parameters. The set of configuration parameters includes one or more configuration parameters, which refer to the configuration parameters of the autonomous vehicle operating in that scenario. Optionally, the autonomous vehicle can be pre-deployed and tested in some typical scenarios to obtain various configuration parameters of the autonomous vehicle operating in those typical scenarios, i.e., the autonomous driving algorithm parameters and the autonomous vehicle's hardware parameters, etc. The configuration parameters can be expressed as parameter items and matching parameter values. For example, a configuration parameter could be: camera scanning frequency = 500Hz.

[0097] Scene filtering rules can refer to the rules for selecting target scenes from the scene library that meet the matching degree requirements of the scene description information of the scene to be deployed and the scene tags of the scenes already stored in the scene library.

[0098] The deployment scenario refers to a scenario where driving parameters for autonomous vehicles need to be deployed. The deployment scenario also includes the characteristics of the autonomous vehicle that needs to operate in that scenario.

[0099] The target scene can be a scene selected from multiple scenes already stored in the scene library according to scene filtering rules. The number of target scenes can be one or more.

[0100] In an optional implementation of this embodiment, natural language description information corresponding to the scene to be deployed can be obtained, such as regional introduction information or map information. Then, one or more scene description fields for describing the scene can be filtered out from the natural language description information by preset filtering rules. Then, the scene description fields can be matched with the tag values ​​in the scene tags of each scene in the scene library.

[0101] In another optional implementation of this embodiment, the location information of the scene to be deployed can be sent to the tag platform first. The tag platform can then establish one or more scene tags for the scene to be deployed according to the scene tag establishment rules of each scene in the scene library. After that, the scene tag of the scene to be deployed can be matched with the corresponding type of scene tags of multiple existing scenes in the scene library. According to the preset scene filtering rules, at least one target scene that matches the scene tag of the scene to be deployed can be filtered in the scene library, and the configuration parameters corresponding to at least one target scene can be obtained.

[0102] In one optional implementation of this embodiment, the scene to be deployed can be filtered using only one scene tag of each scene in the scene library. For example, the target scene can be obtained by matching only one or more scene description fields of the scene to be deployed with the scene location attribute tags of each scene in the scene library; or, the target scene can be obtained by matching the scene location attribute tags of the scene to be deployed with the scene location attribute tags of each scene in the scene library.

[0103] In another optional implementation of this embodiment, multiple scene tags of each scene in the scene library can be used to jointly filter the scene to be deployed. For example, firstly, multiple primary screening scenes matching the scene to be deployed can be filtered out based on the scene location attribute tags of each scene in the scene library. Then, multiple target scenes matching the scene to be deployed can be filtered out again based on the autonomous driving algorithm influencing factor tags of each primary screening scene, or the autonomous driving function usage tags of the unmanned vehicle in the scene.

[0104] Correspondingly, if it is necessary to use multiple scene tags from various scenes in the scene library to finally filter and obtain the target scene, the different scene tags can be sorted by their different descriptive dimensions (for example, scene location attribute tags > autonomous driving algorithm influencing factor tags > autonomous driving function usage tags of driverless cars in the scene). Then, the final target scene can be obtained by sorting by the filtering priority and filtering multiple times.

[0105] Specifically, the method of using multiple scene tags from various scenes in the scene library to ultimately filter and obtain the target scene can include:

[0106] Based on the scene location attribute tags, the system filters out target scenes that match the scene to be deployed from the scene library. Then, based on the autonomous driving algorithm influencing factor tags, the system filters out target scenes that match the scene to be deployed from the first set of target scenes.

[0107] Alternatively, based on the scene location attribute tags, select target scenes from the scene library that match the scene to be deployed; then, based on the autonomous driving function usage tags of the autonomous vehicle in the scene, select target scenes from the first set of target scenes that match the scene to be deployed.

[0108] Alternatively, based on the scene location attribute tags, select target scenes from the scene library that match the scene to be deployed; then, based on the autonomous driving algorithm influencing factor tags, select target scenes from the target scenes that match the scene to be deployed; and finally, based on the autonomous driving function usage tags of the driverless vehicle in the scene, select target scenes from the target scenes that match the scene to be deployed.

[0109] Alternatively, based on the scene location attribute tags, select target scenes from the scene library that match the scene to be deployed; then, based on the autonomous driving function usage tags of the autonomous vehicle in the scene, select target scenes from the first set of target scenes that match the scene to be deployed; and finally, based on the autonomous driving algorithm influencing factor tags, select target scenes from the second set of target scenes that match the scene to be deployed.

[0110] It is understood that those skilled in the art can use appropriate scenario filtering rules to filter out at least one target scenario that matches the scenario to be deployed from the scenario library according to actual needs. This embodiment does not limit the method of finally filtering out the target scenario.

[0111] For example, natural language processing can be used to first identify the scene description field of the scene to be deployed as "airport". Then, one or more scenes in the scene library with the tag value of "airport" can be identified as the target scene. As another example, scene location attribute tags in binary representation can be pre-established for the scene to be deployed. For example, "00" represents airport, "01" represents highway, "10" represents parking lot, and "11" represents park. Similarly, the scene location attribute tags of each scene are stored in the scene library in the same binary form. Then, one or more scenes with the same binary value of the scene location attribute tag in the scene library as the binary value of the scene location attribute tag of the scene to be deployed can be identified as the target scene.

[0112] S120. Determine the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each target scenario.

[0113] The configuration parameter set can refer to the collection of all configuration parameters of the autonomous vehicle corresponding to the target scenario, including algorithm parameters and hardware parameters. The initial configuration parameter set can refer to the collection of all configuration parameters directly used to initialize the deployment of the autonomous vehicle in the scenario to be deployed.

[0114] Optionally, the initial configuration parameter set of the autonomous vehicle can be determined based on the configuration parameter set corresponding to each target scenario, and may include:

[0115] If there are multiple target scenarios, then for each configuration parameter item of the unmanned vehicle, multiple corresponding configuration parameter values ​​are extracted from each configuration parameter set; and the initial configuration parameter set of the unmanned vehicle is determined based on the multiple configuration parameter values ​​corresponding to each configuration parameter item.

[0116] Among them, the configuration parameter item can be, for example, the camera scanning frequency. The configuration parameter value can be a value that corresponds one-to-one with the configuration parameter item. For example, if the value of "camera scanning frequency" is "500Hz", then "500Hz" is the configuration parameter value.

[0117] Specifically, if there are multiple target scenarios and each target scenario corresponds to a set of configuration parameters, multiple configuration parameter values ​​for each configuration parameter item of the autonomous vehicle can be extracted from the configuration parameter sets corresponding to each target scenario. Then, the average of the multiple configuration parameter values ​​corresponding to each configuration parameter item is taken to obtain the initial configuration parameter value of each configuration parameter item, thus determining the initial configuration parameter set of the autonomous vehicle.

[0118] Correspondingly, if there is only one target scenario, the set of configuration parameters (algorithm parameters and hardware parameters of the vehicle running in the target scenario) corresponding to that target scenario will be directly used as the initial set of configuration parameters for the autonomous vehicle in the scenario to be deployed.

[0119] S130. Initialize and deploy the unmanned vehicle based on the initial configuration parameter set.

[0120] In this embodiment, the configuration parameter items of the unmanned vehicle can be assigned values ​​according to the correspondence between each configuration parameter item and configuration parameter value in the initial configuration parameter set, thereby realizing the initial deployment of the unmanned vehicle in the deployment scenario.

[0121] The technical solution of this invention addresses the problem of slow and inefficient parameter deployment caused by improper initialization when deploying autonomous driving configuration parameters in new scenarios. This is achieved by selecting at least one target scenario from a scenario library containing multiple scenarios according to preset scenario filtering rules; determining the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each target scenario; and performing initial deployment of the unmanned vehicle based on the initial configuration parameter set. This invention provides a new scenario deployment method for unmanned driving configuration parameters, improving the deployment speed and efficiency of unmanned vehicle parameters.

[0122] Based on the above technical solution, after initializing and deploying the unmanned vehicle based on the initial configuration parameter set, it is preferable to further include: controlling the unmanned vehicle to run and test in the deployment scenario.

[0123] Among them, operational testing can refer to the process of controlling the unmanned vehicle to actually run in the scenario to be deployed, in order to adjust and optimize the set of initial configuration parameters of the unmanned vehicle.

[0124] The advantage of this setup is that it allows the configuration parameters of the autonomous vehicle to be better adapted to the deployment scenario, thereby improving the driving accuracy of the autonomous vehicle in the deployment scenario.

[0125] Optionally, controlling the unmanned vehicle to perform operational tests in the scenario to be deployed may include:

[0126] If it is determined that only the scene location attribute tag is used, and at least one target scene matching the scene to be deployed is selected from the scene library, then the unmanned vehicle is controlled to perform software-in-the-loop test, hardware-in-the-loop test, vehicle-in-the-loop test and vehicle-on-the-road test in the scene to be deployed.

[0127] If the combined use scenario regional attribute label and autonomous driving algorithm influencing factor label are determined, or the combined use scenario regional attribute label and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, at least one target scenario matching the scenario to be deployed is selected in the scenario library, and the unmanned vehicle is controlled to conduct hardware-in-the-loop testing and real-vehicle road testing in the scenario to be deployed.

[0128] If the combined usage scenario's regional attribute label, autonomous driving algorithm influencing factor label, and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, and at least one target scenario matching the scenario to be deployed is selected from the scenario library, then the unmanned vehicle is controlled to conduct real-vehicle road testing in the scenario to be deployed.

[0129] Understandably, the more scenario tags used in the screening process, the closer the final target scenario will be to the deployment scenario. Consequently, the efficiency of initializing and deploying the autonomous vehicle using the configuration parameter set of the target scenario will be higher, and the various configuration parameters of the autonomous vehicle will be easier to optimize to obtain actual operational adaptation values. As a result, the types of tests used in actual operation testing, namely software-in-the-loop testing, hardware-in-the-loop testing, vehicle-in-the-loop testing, and vehicle-road testing mentioned above, can be simplified.

[0130] The advantage of this setup is that it allows for the implementation of preset scene filtering rules based on different combinations of scene tags, enabling the selective omission of some testing steps to reduce the pressure on the testing process.

[0131] Example 2

[0132] Figure 2aThis is a flowchart of another scenario deployment method for autonomous driving configuration parameters provided in Embodiment 2 of the present invention. Based on the above embodiments, this embodiment refines the operation of selecting at least one target scenario matching the scenario to be deployed from the scenario library according to preset scenario filtering rules. When the scenario tag simultaneously includes scenario geographic attribute tags, autonomous driving algorithm influencing factor tags, and autonomous driving function purpose tags, the method specifically includes the following steps:

[0133] S210. In the scene library, filter out the set of target scenes that match the scene location attribute tags to be deployed, wherein the set of target scenes includes at least one target scene.

[0134] The scenario library includes multiple scenarios, each scenario includes at least one scenario tag, and the configuration parameters corresponding to each scenario are the configuration parameters of the autonomous vehicle running in that scenario.

[0135] Based on the above embodiments, the method may further include: using a pre-established scene tagging system to create matching scene tags for each scene in the scene library, or for the scene to be deployed. For example, Figure 2b This is a schematic diagram of a scene labeling system applicable to Embodiment 1 of the present invention. In this system, each label in the first-level scene corresponds to a scene location attribute label, each label in the second-level scene corresponds to an autonomous driving algorithm influencing factor label, and each label in the third-level scene corresponds to an autonomous driving function usage label for the unmanned vehicle in the scene.

[0136] Specifically, (1) Level 1 description: based on geographical regions and road characteristics, such as airports, industrial parks, and urban roads.

[0137] (2) Secondary Description: Algorithm influencing factors refer to factors that affect the model and parameter selection of the algorithm. Algorithm influencing factors can be categorized into three types based on the object: environmental factors, vehicle-specific factors, and obstacle factors. ① For example, influencing factors for localization algorithms include the specific environment (localization performance is better in open airports than in occluded areas); influencing factors for perception algorithms include 3D environmental information, such as illumination and the reflectivity of the illuminated object. ② For example, influencing factors for target tracking algorithms include the point cloud density and velocity of obstacles; influencing factors for perception algorithms include the velocity of obstacles. ③ For example, the accuracy of localization and perception algorithms is related to the vehicle's construction and operating parameters (sensor calibration and initial brake values, etc.). Construction parameters include vehicle width and height, while operating parameters include speed. In some embodiments, the construction parameters of the same vehicle model are identical. For example, autonomous driving taxis (Robotaxis) have the same vehicle height and width, and the same sensor calibration; however, their actual operating parameters (such as vehicle speed) are different. In summary, the factors influencing an algorithm are multifaceted and work together. The parameters of an algorithm (such as a perception algorithm) are related to multiple object categories (such as ambient lighting, obstacle speed, vehicle structure, and operating parameters).

[0138] (3) Level 3 description: The specific function or purpose of the autonomous vehicle in the scenario. For example... Figure 2b As shown, precise parking and precise docking charging are examples. Different functions or applications require different algorithm models and parameters. For instance, in docking multi-trailer lifts and precisely reversing to dock with express delivery lockers, although both employ target recognition and target tracking algorithms, the specific functions and accuracy requirements differ, thus the choice of algorithm parameter settings and models will also differ.

[0139] Based on the three-level scene tag descriptions described above, a scene can be accurately represented. Correspondingly, based on the above tag system, corresponding scene tags can be created for each scene in the scene library.

[0140] It needs to be pointed out again. Figure 2b The labeling system shown has different scene levels for different scene labels. In reality, different scene labels may not have the above-mentioned level differences. Those skilled in the art can create scene labels that are suitable for use according to the actual situation.

[0141] S220. In the target scenario set, at least one target scenario is selected where the autonomous driving function usage tag of the unmanned vehicle in the scenario matches the scenario to be deployed, and the selected target scenario is used to update the target scenario set.

[0142] S230. In the target scenario set, at least one target scenario that matches the autonomous driving algorithm influencing factor label with the scenario to be deployed is selected, and the selected target scenario is used to update the target scenario set.

[0143] In one optional embodiment of this example, the labels for influencing factors of the autonomous driving algorithm may include one or more of the following: environmental factor labels, vehicle factor labels, and obstacle factor labels.

[0144] As an example, and not a limitation, environmental factor labels can be labels for environmental factors that affect the accuracy of the algorithm, such as weather, illumination, and the presence of shading in the surrounding environment. Vehicle factor labels can be labels for factors of the vehicle itself that affect the accuracy of the algorithm, such as vehicle height, width, and speed. Obstacle factor labels can be labels for obstacle factors that affect the accuracy of the algorithm, such as the point cloud density and speed of obstacles.

[0145] In an optional implementation of this embodiment, the target scenario set can be filtered using only one factor label from the autonomous driving algorithm's influencing factor labels. For example, target scenarios can be obtained by matching only one or more environmental factor description fields of the scenario to be deployed with the environmental factor labels of each target scenario in the target scenario set.

[0146] In another optional implementation of this embodiment, multiple factor tags from the autonomous driving algorithm's influencing factor tags can be used to jointly filter the target scene set. For example, firstly, the environmental factor tags of each scene in the target scene set are used to filter multiple primary screening scenarios. Then, the vehicle factor tag or obstacle factor tag is used for secondary screening to obtain at least one target scene.

[0147] Specifically, the method of ultimately selecting target scenarios by using multiple autonomous driving algorithm influencing factor labels for each scenario in the target scenario set can include:

[0148] Based on the environmental factor labels, at least one target scenario that matches the scenario to be deployed is selected from the set of target scenarios.

[0149] Alternatively, based on environmental factor tags, select target scenarios from the target scenario set that match the environmental factor tags with the scenario to be deployed; then, based on the vehicle factor tags, select target scenarios from the first set of target scenarios that match the vehicle factor tags with the scenario to be deployed.

[0150] Alternatively, based on environmental factor tags, select target scenarios from the target scenario set that match the environmental factor tags with the scenario to be deployed; then, based on obstacle factor tags, select target scenarios from the first set of target scenarios that match the obstacle factor tags with the scenario to be deployed.

[0151] Alternatively, based on environmental factor tags, select target scenarios from the target scenario set that match the environmental factor tags with the scenario to be deployed; then, based on the vehicle factor tags, select target scenarios from the target scenarios that match the vehicle factor tags with the scenario to be deployed; and then, based on obstacle factor tags, select target scenarios from the target scenarios that match the obstacle factor tags with the scenario to be deployed.

[0152] S240. Determine if there are multiple target scenarios: if yes, execute S250; otherwise, execute S260.

[0153] S250. For each configuration parameter item of the unmanned vehicle, extract the corresponding multiple configuration parameter values ​​from each configuration parameter set, and execute S270.

[0154] S260. Determine the unique set of configuration parameters for the target scenario as the initial set of configuration parameters for the unmanned vehicle, and execute S270.

[0155] S270. Initialize and deploy the unmanned vehicle based on the initial configuration parameter set.

[0156] S280. Control the unmanned vehicle to conduct operational tests in the scenario to be deployed.

[0157] The technical solution of this invention addresses the problem that existing autonomous vehicles may experience slow and inefficient parameter deployment due to improper initialization when deploying autonomous driving configuration parameters in new scenarios. This invention provides a new method for deploying autonomous driving configuration parameters in a scenario, improving the speed and efficiency of parameter deployment for autonomous vehicles.

[0158] Example 3

[0159] Figure 3a This is a flowchart of another scenario deployment method for autonomous driving configuration parameters provided in Embodiment 3 of the present invention. Based on the above embodiments, in the process of selecting at least one target scenario that matches the autonomous driving algorithm influencing factor labels with the scenario to be deployed, the autonomous driving algorithm influencing factor labels used simultaneously include: environmental factor labels, vehicle factor labels, and obstacle factor labels. The method specifically includes the following steps:

[0160] S310. In the scene library, filter out the set of target scenes that match the scene location attribute tags to be deployed, wherein the set of target scenes includes at least one target scene.

[0161] The scenario library includes multiple scenarios, each scenario includes at least one scenario tag, and the configuration parameters corresponding to each scenario are the configuration parameters of the autonomous vehicle running in that scenario.

[0162] S320. In the target scenario set, at least one target scenario is selected where the autonomous driving function usage tag of the unmanned vehicle in the scenario matches the scenario to be deployed, and the selected target scenario is used to update the target scenario set.

[0163] S330. In the target scenario set, at least one target scenario whose environmental factor label matches the scenario to be deployed is selected, and the selected target scenario is used to update the target scenario set.

[0164] Among them, Figure 3b The diagram illustrates a method for filtering at least one target scenario from the set of target scenarios whose environmental factor labels match the scenario to be deployed. Figure 3b The method includes:

[0165] S3301. Obtain at least one autonomous driving association algorithm applicable to the scenario to be deployed, and obtain the algorithm influence factor function corresponding to each autonomous driving association algorithm.

[0166] Among them, autonomous driving association algorithms can refer to the algorithms required for autonomous driving of unmanned vehicles in the scenario to be deployed. For example, the autonomous driving association algorithms in scenario A may only include localization algorithms, typically VSLAM (Visual Simultaneous Localization and Mapping) or LSLAM (Lidar Simultaneous Localization and Mapping) algorithms, or fusion localization algorithms, etc.; the autonomous driving association algorithms in scenario B may include localization algorithms (e.g., the aforementioned VSLAM algorithm, LSLAM algorithm, or fusion localization algorithm), reversing algorithms with trailers, precise docking algorithms, as well as dangerous goods recognition algorithms and pedestrian crossing guardrail recognition algorithms, etc.

[0167] The algorithm influence factor function has a one-to-one correspondence with the autonomous driving association algorithm. Each algorithm influence factor function can be obtained by weighted summation of at least one algorithm influence factor and its corresponding weight coefficient. The algorithm influence factors included in the function represent the factors that affect the accuracy of the matched autonomous driving association algorithm. These algorithm influence factors correspond to the label factors in the environmental factor label of the autonomous driving algorithm influence factor tag.

[0168] In a specific example, the accuracy of an autonomous driving positioning algorithm is affected by two algorithmic factors: antenna obstruction rate and visibility (corresponding to the label factors in the environmental factors label of the autonomous driving algorithm's influencing factors). Therefore, an algorithmic influencing factor function of the form "f(positioning algorithm) = a * GPS antenna obstruction rate + b * visibility" can be constructed, corresponding to this autonomous driving positioning algorithm. The weighting coefficients a and b can be determined experimentally or selected from preset empirical values.

[0169] S3302. Based on the influence factor values ​​of the scenario to be deployed under each of the algorithm influence factors, calculate the objective function value corresponding to each of the algorithm influence factor functions.

[0170] The impact factor value refers to the corresponding numerical value of the algorithm's impact factor in the scenario to be deployed. For example, visibility (algorithm impact factor) = 50 (impact factor value). The objective function value can be the function value calculated by substituting each impact factor value into the algorithm's impact factor function. The objective function value can be expressed in numerical form. In a specific example, assuming GPS antenna obstruction rate = 1.2, visibility = 50, a = 0.5, and b = 0.5, then for the algorithm's impact factor function f (positioning algorithm) = a * GPS antenna obstruction rate + b * visibility, the calculated objective function value is 25.6. The number of objective function values ​​corresponds to the number of autonomous driving association algorithms applicable to the scenario to be deployed.

[0171] S3303. Based on the safety weights of the influence factor functions of each algorithm, the values ​​of each objective function are weighted and summed to obtain the target algorithm security degree corresponding to the deployment scenario.

[0172] The safety weight refers to the safety level weight of each autonomous driving association algorithm required for autonomous driving in the deployment scenario. Different autonomous driving association algorithms can have different safety weights. Generally, the higher the impact of an autonomous driving association algorithm on the safety of autonomous driving, the higher its safety weight. The target algorithm safety level can be the weighted sum of the objective function values ​​corresponding to the influence factor functions of each algorithm in the deployment scenario and their respective safety weights. For example: the target algorithm safety level S for scenario 1 is S = m*f (localization algorithm) + n*f (perception algorithm) + ... where m and n can represent the safety weights of the localization algorithm and the perception algorithm, respectively.

[0173] As mentioned earlier, after calculating the target function values ​​corresponding to f (positioning algorithm) and f (perception algorithm) respectively, and combining the preset safety weights m and n, the safety degree S of the target algorithm can be calculated by using the formula S = m * f (positioning algorithm) + n * f (perception algorithm).

[0174] For example, in a scenario where the failure of the fusion localization algorithm is more dangerous than the failure of the perception algorithm, the safety weight of the fusion localization algorithm can be set higher. This is because when the perception algorithm cannot identify the type of target or obstacle in the image, it can be defined as a general obstacle to be avoided, or even allowed to pull over; however, the failure of the fusion localization algorithm will lead to inaccurate vehicle positioning, potentially resulting in collisions. As another example, if the failure of both the fusion localization and perception algorithms is more dangerous than the failure of the planning and control algorithm, then the safety weight of the fusion localization and perception algorithms can be set higher than that of the planning and control algorithm. This is because the fusion localization and perception algorithms are the inputs to the planning and control algorithm; without localization and perception capabilities, the vehicle cannot perform planning and control operations. Specifically, the values ​​of the safety weights can be determined empirically or through testing and verification.

[0175] S3304. In the set of target scenes, obtain the scene algorithm security degree corresponding to each target scene.

[0176] The safety degree of the scenario algorithm is obtained by weighting the safety weight of each algorithm influence factor function and the scenario function value of each target scenario for each algorithm influence factor function. The scenario function value is determined by the label factor and the corresponding label factor value in the environmental factor label of each target scenario.

[0177] As mentioned earlier, the label factor and label factor value can be obtained by further subdividing the label value in the environmental label. For example, in a scene label such as "Autonomous Driving Algorithm Influence Factors - Environmental Factors: Illuminance = 30 lux", illuminance can represent the label factor in the environmental factor label of the target scene, and 30 lux can represent the label factor value corresponding to the label factor.

[0178] Since the algorithm impact factors correspond to the label factors of the environmental factor labels in the autonomous driving algorithm impact factor tags, for each algorithm impact factor function in the deployment scenario, the label factor values ​​of the label factors in the environmental factor labels of the target scenario that match the algorithm impact factors can be substituted into each algorithm impact factor function to obtain the scenario function value of the target scenario for each algorithm impact factor function. Then, based on the safety weight of each algorithm impact factor function and the scenario function value of each target scenario for each algorithm impact factor function, the scenario algorithm safety level corresponding to each target scenario is calculated.

[0179] S3305. Calculate the first difference value between the security level of each scenario algorithm and the security level of the target algorithm.

[0180] The first difference value can be the absolute value of the difference between the security level of the scene algorithm in each target scene and the security level of the target algorithm in the scene to be deployed.

[0181] S3306. Obtain the security screening scenarios where the first difference value is less than or equal to the first threshold value, and determine each of the security screening scenarios as the target scenario.

[0182] The first threshold value can be a pre-set upper limit value of the first difference value.

[0183] S340. In the target scenario set, at least one target scenario whose vehicle factor tag matches the scenario to be deployed is selected, and the selected target scenario is used to update the target scenario set.

[0184] S350. In the target scene set, at least one target scene whose obstacle factor label matches the scene to be deployed is selected, and the selected target scene is used to update the target scene set.

[0185] S360. Determine the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each target scenario.

[0186] In an optional implementation of this embodiment, determining the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each target scenario may include:

[0187] If there are multiple target scenarios, at least one safe and compatible scenario is selected from all target scenarios based on the numerical value of the scenario algorithm safety level of each target scenario; and the initial configuration parameter set of the unmanned vehicle is determined based on the configuration parameter set corresponding to each safe and compatible scenario.

[0188] Among them, the security adaptation scenario can be the target scenario with the highest matching degree to the scenario to be deployed, selected based on the algorithm's security level.

[0189] Specifically, when there are multiple target scenarios, and each target scenario corresponds to a scenario algorithm safety level, at least one safe and adaptable scenario can be selected from all target scenarios based on the magnitude of the scenario algorithm safety level. If there is only one safe and adaptable scenario, the set of configuration parameters for this safe and adaptable scenario can be used to determine the initial configuration parameter set for the autonomous vehicle. If there are more than one safe and adaptable scenario, the average value of the configuration parameters can be taken to determine the initial configuration parameter set for the autonomous vehicle. It should be noted that the number of safe and adaptable scenarios will not be too large, for example, it can be 2-3.

[0190] S370. Initialize and deploy the unmanned vehicle based on the initial configuration parameter set.

[0191] The technical solution of this invention addresses the problem that existing autonomous vehicles may experience slow and inefficient parameter deployment due to improper initialization when deploying autonomous driving configuration parameters in new scenarios. This invention provides a new method for deploying autonomous driving configuration parameters in a scenario, improving the speed and efficiency of parameter deployment for autonomous vehicles.

[0192] Based on the above embodiments, after determining each security screening scenario as a target scenario in step S330, further screening operations can be performed on each target scenario. The above further screening operations can occur before or after each step between S330 and S370, and this embodiment does not limit this.

[0193] Correspondingly, such as Figure 3c As shown, the above-mentioned further screening methods may include:

[0194] S3100. Among the selected target scenarios, obtain the target scenario currently being processed;

[0195] S3200: Obtain a current scene function value from the target scene being processed.

[0196] S3300: In the scenario to be deployed, obtain the current target function value corresponding to the current scenario function value.

[0197] S3400, Calculate the second difference value between the current scene function value and the current target function value.

[0198] S3500: Determine whether the second difference value is greater than the second threshold value. If yes, execute S3600; otherwise, execute S3700.

[0199] S3600, Filter out the current processing target scenario and execute S3800.

[0200] S3700: Determine whether the processing of all scene function values ​​in the current target scene has been completed: if yes, execute S3800; otherwise, return to execute S3200.

[0201] S3800: Determine whether the processing of all target scenes has been completed. If yes, end the filtering process; otherwise, return to execute S3100.

[0202] In S3100-S3800, the current scene function value can be the value of a scene function corresponding to the target scene being processed, such as the value of "f(positioning algorithm)" in "f(positioning algorithm) = a * GPS antenna obstruction rate + b * visibility". The current target function value can be the value of the algorithm influence factor function corresponding to the current scene function. Corresponding to the previous example, the current target function value is also f(positioning algorithm) = a * GPS antenna obstruction rate + b * visibility. The second difference value can be the absolute value of the difference between the current scene function value and the current target scene function value. The second threshold value can be the upper limit of the second difference value.

[0203] In this embodiment, for each target scenario, a scenario function is sequentially selected from all scenario functions corresponding to the currently processed target scenario as the current scenario function, and its scenario function value is obtained. Furthermore, the current target function value corresponding to the current scenario function value is obtained in the scenario to be deployed. A second difference value is calculated between the current scenario function value and the current target function value, and it is then determined whether this second difference value exceeds a second threshold. If it does, there is no need to continue judging the second difference values ​​between other scenario function values ​​of the currently processed target scenario and their corresponding target function values; the currently processed target scenario is directly filtered out from the target scenario set. If it does not exceed the threshold, the next scenario function and its value for the currently processed target scenario can be obtained as the new current scenario function value, and the aforementioned judgment operation is performed until all scenario function values ​​in the currently processed target scenario are processed. Clearly, when the second difference values ​​between all scenario function values ​​of the currently processed target scenario and their corresponding current target function values ​​do not exceed the second threshold, the currently processed target scenario can be retained.

[0204] Furthermore, if the above-described further filtering process is adopted, after calculating the second difference value between the current scene function value and the current target function value, the process may further include:

[0205] Based on the second difference value, update the accumulated value of the second difference value corresponding to the target scenario being processed.

[0206] That is, for example, the target scenario being processed currently includes scenario function value 1, scenario function value 2, and scenario function value 3. In the scenario to be deployed, the target function value 1 corresponding to scenario function value 1 is obtained, and the second difference value 1 between scenario function value 1 and target function value 1 is calculated. It is determined whether the second difference value 1 is greater than the second threshold value. If it is greater, the target scenario being processed is filtered out. If it is not greater, in the scenario to be deployed, the target function value 2 corresponding to scenario function value 2 is obtained, and the second difference value 2 between scenario function value 2 and target function value 2 is calculated. It is determined whether the second difference value 2 is greater than the second threshold value. If it is greater, the target scenario being processed is filtered out. If it is not greater, in the scenario to be deployed, the target function value 3 corresponding to scenario function value 3 is obtained, and the second difference value 3 is calculated. It is determined whether the second difference value 3 is greater than the second threshold value. If it is greater, the target scenario being processed is filtered out. If it is not greater, the second difference value 1, the second difference value 2, and the second difference value 3 are summed as the accumulated second difference value corresponding to the target scenario being processed.

[0207] Accordingly, based on the set of configuration parameters corresponding to each target scenario, the initial configuration parameter set of the autonomous vehicle is determined, which may include:

[0208] If there are multiple target scenarios, then among the target scenarios, the similar adaptation scenario with the smallest cumulative second difference value is selected; the set of configuration parameters corresponding to the similar adaptation scenario is determined as the initial configuration parameter set of the unmanned vehicle.

[0209] Example 4

[0210] Figure 4 This is a schematic diagram of a scene deployment device for autonomous driving configuration parameters provided in Embodiment 4 of the present invention. This device can execute the scene deployment methods for autonomous driving configuration parameters involved in the above embodiments. (See reference...) Figure 4 The device includes: a target scene selection module 410, an initial configuration parameter set determination module 420, and an unmanned vehicle initialization deployment module 430. Among them:

[0211] The target scene filtering module 410 is used to filter out at least one target scene that matches the scene to be deployed in the scene library according to the preset scene filtering rules. The scene library includes multiple scenes, each scene includes at least one scene tag, and the configuration parameters corresponding to each scene are the configuration parameters of the unmanned vehicle running in that scene.

[0212] The initial configuration parameter set determination module 420 is used to determine the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each target scenario.

[0213] The unmanned vehicle initialization deployment module 430 is used to initialize and deploy the unmanned vehicle based on the initial configuration parameter set.

[0214] The technical solution of this invention addresses the problem of slow and inefficient parameter deployment caused by improper initialization when deploying autonomous driving configuration parameters in new scenarios. This is achieved by selecting at least one target scenario from a scenario library containing multiple scenarios according to preset scenario filtering rules; determining the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each target scenario; and performing initial deployment of the unmanned vehicle based on the initial configuration parameter set. This invention provides a new scenario deployment method for unmanned driving configuration parameters, improving the deployment speed and efficiency of unmanned vehicle parameters.

[0215] Optionally, in the above-described device, the scene tag may include: a scene location attribute tag;

[0216] The target scene filtering module 410 can be used for:

[0217] In the scenario library, a set of target scenarios that match the scenario location attribute tags to be deployed are selected, wherein the set of target scenarios includes at least one target scenario.

[0218] Optionally, in the above-mentioned device, the scene label may further include: labels for influencing factors of autonomous driving algorithms;

[0219] The device preferably includes a first target scene set update module, used to filter out a set of target scenes in the scene library whose scene location attribute tags match the scene to be deployed:

[0220] In the set of target scenarios, at least one target scenario is selected that matches the autonomous driving algorithm influencing factor label with the scenario to be deployed, and the selected target scenario is used to update the set of target scenarios.

[0221] Optionally, in the above-mentioned device, the scene label may also include: a label indicating the autonomous driving function of the unmanned vehicle in the scene;

[0222] The device further includes a second target scene set update module, used to filter out a set of target scenes in the scene library whose scene location attribute tags match the scene to be deployed:

[0223] In the target scenario set, at least one target scenario is selected where the autonomous driving function usage tag of the unmanned vehicle in the scenario matches the scenario to be deployed, and the selected target scenario is used to update the target scenario set.

[0224] Optionally, in the above-mentioned device, the scene label may further include: labels for influencing factors of autonomous driving algorithms;

[0225] The device further includes a third target scene set update module, used to filter out at least one target scene from the target scene set whose autonomous driving function usage tag matches the scene to be deployed, and then update the target scene set with the filtered target scene:

[0226] In the set of target scenarios, at least one target scenario is selected that matches the autonomous driving algorithm influencing factor label with the scenario to be deployed, and the selected target scenario is used to update the set of target scenarios.

[0227] Optionally, in the above-mentioned device, the labels for influencing factors of the autonomous driving algorithm may include: environmental factor labels;

[0228] The first target scene set update module or the third target scene set update module includes:

[0229] The target scenario set update unit is used to filter out at least one target scenario in the target scenario set whose environmental factor label matches the scenario to be deployed, and update the target scenario set with the filtered target scenario.

[0230] Optionally, in the above-mentioned device, the autonomous driving algorithm influencing factor label may further include: vehicle factor label;

[0231] The device further includes a fourth target scene set update module, used to filter out at least one target scene whose environmental factor label matches the scene to be deployed from the target scene set, and then update the target scene set with the filtered target scene:

[0232] In the target scenario set, at least one target scenario whose vehicle factor tag matches the scenario to be deployed is selected, and the selected target scenario is used to update the target scenario set.

[0233] Optionally, in the above-mentioned device, the autonomous driving algorithm influencing factor label may further include: obstacle factor label;

[0234] The device further includes a fifth target scene set update module, used to filter out at least one target scene from the target scene set that matches the vehicle factor tag with the scene to be deployed, and then update the target scene set with the filtered target scene:

[0235] In the set of target scenarios, at least one target scenario whose obstacle factor label matches the scenario to be deployed is selected, and the set of target scenarios is updated using the selected target scenario.

[0236] Optionally, the target scene set update unit in the above-mentioned device can be used for:

[0237] Obtain at least one autonomous driving association algorithm applicable to the scenario to be deployed, and obtain the algorithm impact factor function corresponding to each autonomous driving association algorithm;

[0238] Each algorithmic impact factor function is obtained by weighted summation of at least one algorithmic impact factor and its corresponding weight coefficient; the algorithmic impact factor corresponds to the label factor in the environmental factor label.

[0239] Based on the impact factor values ​​of the scenario to be deployed under each of the algorithm impact factors, calculate the objective function value corresponding to each of the algorithm impact factor functions;

[0240] Based on the safety weights of the influence factor functions of each algorithm, the values ​​of each objective function are weighted and summed to obtain the target algorithm safety degree corresponding to the deployment scenario;

[0241] In the set of target scenarios, obtain the scenario algorithm security degree corresponding to each target scenario;

[0242] The safety of the scenario algorithm is obtained by weighting the scenario function of each algorithm influence factor function and the scenario function value of each target scenario for each algorithm influence factor function. The scenario function value is determined by the tag factor and the corresponding tag factor value in the environmental factor label of each target scenario.

[0243] Calculate the first difference value between the security level of each scenario algorithm and the security level of the target algorithm;

[0244] Obtain the security screening scenarios where the first difference value is less than or equal to the first threshold value, and determine each of the security screening scenarios as the target scenario.

[0245] Optionally, the above-described apparatus further includes a current processing target scene filtering module, used after determining each of the security screening scenarios as the target scene:

[0246] For each target scenario, perform the following further filtering operation:

[0247] Get the target scene being processed, and get a current scene function value from the target scene being processed;

[0248] In the scenario to be deployed, obtain the current target function value corresponding to the current scenario function value;

[0249] Calculate the second difference value between the current scene function value and the current objective function value;

[0250] If the second difference value is greater than the second threshold value, then the current processing target scenario is filtered out;

[0251] If the second difference value is less than or equal to the second threshold value, then return to the operation of obtaining a current scene function value in the target scene being processed, until all scene function values ​​in the target scene being processed are processed.

[0252] Optionally, in the above-described apparatus, the initial configuration parameter set determination module 420 includes:

[0253] The configuration parameter value extraction unit is used to extract multiple corresponding configuration parameter values ​​from each configuration parameter set for each configuration parameter item of the unmanned vehicle if there are multiple target scenarios.

[0254] The first initial configuration parameter set unit is used to determine the initial configuration parameter set of the unmanned vehicle based on multiple configuration parameter values ​​corresponding to each configuration parameter item.

[0255] Optionally, in the above-described apparatus, the initial configuration parameter set determination module 420 includes:

[0256] The security adaptation scenario filtering unit is used to filter at least one security adaptation scenario from all target scenarios if there are multiple target scenarios, based on the numerical value of the scenario algorithm security degree of each target scenario.

[0257] The second initial configuration parameter set unit is used to determine the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each safety adaptation scenario.

[0258] Optionally, the above-described apparatus further includes a second difference value accumulation update module, used after calculating the second difference value between the current scene function value and the current target function value:

[0259] Based on the second difference value, update the accumulated value of the second difference value corresponding to the target scenario being processed;

[0260] The initial configuration parameter set determination module 420 can be used for:

[0261] If there are multiple target scenarios, then among the target scenarios, the similar adaptation scenario with the smallest cumulative second difference value is selected.

[0262] The set of configuration parameters corresponding to similar adaptation scenarios is determined as the initial configuration parameter set of the unmanned vehicle.

[0263] Optionally, the above-mentioned device further includes an unmanned vehicle operation test control module, which, after initializing and deploying the unmanned vehicle based on the initial configuration parameter set, further includes:

[0264] The unmanned vehicle is controlled to conduct operational tests in the scenario to be deployed.

[0265] Optionally, the unmanned vehicle operation test control module in the above-mentioned device can be used for:

[0266] If it is determined that only the scene location attribute tag is used, and at least one target scene matching the scene to be deployed is selected from the scene library, then the unmanned vehicle is controlled to perform software-in-the-loop test, hardware-in-the-loop test, vehicle-in-the-loop test and vehicle-on-the-road test in the scene to be deployed.

[0267] If the combined use scenario regional attribute label and autonomous driving algorithm influencing factor label are determined, or the combined use scenario regional attribute label and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, at least one target scenario matching the scenario to be deployed is selected in the scenario library, and the unmanned vehicle is controlled to conduct hardware-in-the-loop testing and real-vehicle road testing in the scenario to be deployed.

[0268] If the combined usage scenario's regional attribute label, autonomous driving algorithm influencing factor label, and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, and at least one target scenario matching the scenario to be deployed is selected from the scenario library, then the unmanned vehicle is controlled to conduct real-vehicle road testing in the scenario to be deployed.

[0269] The scene deployment device for autonomous driving configuration parameters provided in this embodiment of the invention can execute the scene deployment method for autonomous driving configuration parameters provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0270] Example 5

[0271] Figure 5This is a schematic diagram of the structure of an electronic device provided in Embodiment 5 of the present invention, as shown below. Figure 5 As shown, the device includes a processor 510, a storage device 520, an input device 530, and an output device 540; the number of processors 510 in the device can be one or more. Figure 5 Taking a processor 510 as an example; the processor 510, storage device 520, input device 530 and output device 540 in the device can be connected via a bus or other means. Figure 5 Taking the example of a connection between China and Israel via a bus.

[0272] Storage device 520, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the scene deployment method for autonomous driving configuration parameters in this embodiment of the invention (e.g., the target scene filtering module 410, the initial configuration parameter set determination module 420, and the autonomous vehicle initialization deployment module 430 in the scene deployment device for autonomous driving configuration parameters). Processor 510 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in storage device 520, thereby implementing the aforementioned scene deployment method for autonomous driving configuration parameters, which includes:

[0273] According to the preset scenario filtering rules, at least one target scenario that matches the scenario to be deployed is selected from the scenario library. The scenario library includes multiple scenarios, each scenario includes at least one scenario tag, and the configuration parameters corresponding to each scenario are the configuration parameters of the unmanned vehicle running in that scenario.

[0274] The initial configuration parameter set of the unmanned vehicle is determined based on the configuration parameter set corresponding to each target scenario.

[0275] The unmanned vehicle is initially deployed based on the set of initial configuration parameters.

[0276] Storage device 520 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function; the data storage area may store data created based on terminal usage. Furthermore, storage device 520 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory, or other non-volatile solid-state storage device. In some instances, storage device 520 may further include memory remotely located relative to processor 510, which can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0277] Input device 530 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the device. Output device 540 may include display devices such as a display screen.

[0278] Example 6

[0279] Embodiment 6 of the present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is used to perform a scenario deployment method for autonomous driving configuration parameters, the method comprising:

[0280] According to the preset scenario filtering rules, at least one target scenario that matches the scenario to be deployed is selected from the scenario library. The scenario library includes multiple scenarios, each scenario includes at least one scenario tag, and the configuration parameters corresponding to each scenario are the configuration parameters of the unmanned vehicle running in that scenario.

[0281] The initial configuration parameter set of the unmanned vehicle is determined based on the configuration parameter set corresponding to each target scenario.

[0282] The unmanned vehicle is initially deployed based on the set of initial configuration parameters.

[0283] Of course, the computer-readable storage medium provided in the embodiments of the present invention stores a computer program thereon. The program is not limited to the method operation described above, but can also execute related operations in the scenario deployment method of autonomous driving configuration parameters provided in any embodiment of the present invention.

[0284] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0285] It is worth noting that in the above-mentioned embodiments of the scenario deployment device for autonomous driving configuration parameters, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.

[0286] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. A method for deploying a scenario of unmanned configuration parameters, characterized in that, include: In the scenario library, a set of target scenarios matching the scenario location attribute tags are selected. The target scenario set includes at least one target scenario, and the scenario library includes multiple scenarios. Each scenario includes at least one scenario tag, and the configuration parameters corresponding to each scenario are the configuration parameters of the autonomous vehicle running in that scenario. The scenario tag includes one or more of the following: scenario location attribute tag, autonomous driving algorithm influencing factor tag, and autonomous driving function usage tag of the autonomous vehicle in the scenario. The autonomous driving algorithm influencing factor tag includes environmental factor tag. Obtain at least one autonomous driving association algorithm applicable to the deployment scenario, and obtain the algorithm influence factor function corresponding to each autonomous driving association algorithm; wherein, the autonomous driving association algorithm refers to the algorithm required for autonomous driving of unmanned vehicles in the deployment scenario; Based on the impact factor values ​​of the scenario to be deployed under each of the algorithm impact factors, calculate the objective function value corresponding to each of the algorithm impact factor functions; Based on the safety weights of the influence factor functions of each algorithm, the values ​​of each objective function are weighted and summed to obtain the target algorithm safety degree corresponding to the deployment scenario; In the set of target scenarios, obtain the scenario algorithm security degree corresponding to each target scenario; Calculate the first difference value between the security level of each scenario algorithm and the security level of the target algorithm; Obtain the security screening scenarios with a first difference value less than or equal to a first threshold value, determine each of the security screening scenarios as the target scenarios, and update the target scenario set using the selected target scenarios; Alternatively, in the target scenario set, at least one target scenario whose autonomous driving function usage tag of the unmanned vehicle in the scenario matches the scenario to be deployed is selected, and the target scenario set is updated using the selected target scenario. Alternatively, in the target scenario set, at least one target scenario whose autonomous driving function usage tag of the unmanned vehicle in the scenario matches the scenario to be deployed is selected, and the target scenario set is updated using the selected target scenario; at least one target scenario whose autonomous driving algorithm influencing factor tag matches the scenario to be deployed is selected, and the target scenario set is updated using the selected target scenario. The initial configuration parameter set of the unmanned vehicle is determined based on the configuration parameter set corresponding to each target scenario. The unmanned vehicle is initially deployed based on the initial configuration parameter set; The unmanned vehicle is controlled to run tests in the scenario to be deployed; wherein, the running test is the process of controlling the unmanned vehicle to actually run in the scenario to be deployed, so as to adjust and optimize the set of initial configuration parameters of the unmanned vehicle. The operational test includes: if it is determined that only the scene location attribute label is used, at least one target scene matching the scene to be deployed is selected in the scene library, and then the unmanned vehicle is controlled to perform software-in-the-loop test, hardware-in-the-loop test, vehicle-in-the-loop test and vehicle-on-the-road test in the scene to be deployed; If the combined use scenario location attribute label and autonomous driving algorithm influencing factor label are determined, or the combined use scenario location attribute label and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, at least one target scenario matching the scenario to be deployed is selected in the scenario library, and the unmanned vehicle is controlled to conduct hardware-in-the-loop testing and real-vehicle road testing in the scenario to be deployed. If the combined usage scenario's regional attribute label, autonomous driving algorithm influencing factor label, and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, and at least one target scenario matching the scenario to be deployed is selected from the scenario library, then the unmanned vehicle is controlled to conduct real-vehicle road testing in the scenario to be deployed.

2. The method of claim 1, wherein, The autonomous driving algorithm's influencing factor labels also include: vehicle-specific factor labels; After selecting at least one target scenario from the target scenario set that matches the environmental factor tag of the scenario to be deployed, and updating the target scenario set with the selected target scenario, the process further includes: In the target scenario set, at least one target scenario whose vehicle factor tag matches the scenario to be deployed is selected, and the selected target scenario is used to update the target scenario set.

3. An apparatus for deploying a scenario of unmanned configuration parameters, characterized in that, include: The target scene filtering module is used to filter out a set of target scenes from the scene library that match the scene location attribute tags with the scene to be deployed. The target scene set includes at least one target scene, the scene library includes multiple scenes, each scene includes at least one scene tag, and the configuration parameters corresponding to each scene are the configuration parameters of the unmanned vehicle running in that scene. The scene tag includes one or more of the following: scene location attribute tag, autonomous driving algorithm influencing factor tag, and autonomous driving function usage tag of the unmanned vehicle in the scene. The autonomous driving algorithm influencing factor tag includes environmental factor tag. The target scenario screening module is also used to obtain at least one autonomous driving association algorithm applicable to the scenario to be deployed, and to obtain the algorithm influence factor function corresponding to each autonomous driving association algorithm; wherein, the autonomous driving association algorithm refers to the algorithm required for autonomous driving of unmanned vehicles in the scenario to be deployed. Based on the impact factor values ​​of the scenario to be deployed under each of the algorithm impact factors, calculate the objective function value corresponding to each of the algorithm impact factor functions; Based on the safety weights of the influence factor functions of each algorithm, the values ​​of each objective function are weighted and summed to obtain the target algorithm safety degree corresponding to the deployment scenario; In the set of target scenarios, obtain the scenario algorithm security degree corresponding to each target scenario; Calculate the first difference value between the security level of each scenario algorithm and the security level of the target algorithm; Obtain the security screening scenarios with a first difference value less than or equal to a first threshold value, determine each of the security screening scenarios as the target scenarios, and update the target scenario set using the selected target scenarios; Alternatively, it can be used to filter out at least one target scenario in the target scenario set that matches the autonomous driving function usage tag of the unmanned vehicle in the scenario with the scenario to be deployed, and update the target scenario set with the filtered target scenario; Alternatively, it is also used to filter out at least one target scenario in the target scenario set that matches the autonomous driving function usage tag of the unmanned vehicle in the scenario with the scenario to be deployed, and to update the target scenario set with the filtered target scenario; and to filter out at least one target scenario that matches the autonomous driving algorithm influencing factor tag with the scenario to be deployed, and to update the target scenario set with the filtered target scenario. The initial configuration parameter set determination module is used to determine the initial configuration parameter set of the unmanned vehicle based on the configuration parameter set corresponding to each target scenario. The unmanned vehicle initialization and deployment module is used to initialize and deploy the unmanned vehicle based on the initial configuration parameter set; The unmanned vehicle operation test control module is used to control the unmanned vehicle to conduct operation tests in the deployment scenario; wherein, the operation test is the process of controlling the unmanned vehicle to actually run in the deployment scenario in order to adjust and optimize the unmanned vehicle's initial configuration parameter set; The operational test includes: if it is determined that only the scene location attribute label is used, at least one target scene matching the scene to be deployed is selected in the scene library, and then the unmanned vehicle is controlled to perform software-in-the-loop test, hardware-in-the-loop test, vehicle-in-the-loop test and vehicle-on-the-road test in the scene to be deployed; If the combined use scenario location attribute label and autonomous driving algorithm influencing factor label are determined, or the combined use scenario location attribute label and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, at least one target scenario matching the scenario to be deployed is selected in the scenario library, and the unmanned vehicle is controlled to conduct hardware-in-the-loop testing and real-vehicle road testing in the scenario to be deployed. If the combined usage scenario's regional attribute label, autonomous driving algorithm influencing factor label, and autonomous driving function usage label of the unmanned vehicle in the scenario are determined, and at least one target scenario matching the scenario to be deployed is selected from the scenario library, then the unmanned vehicle is controlled to conduct real-vehicle road testing in the scenario to be deployed.

4. An electronic device, comprising: The electronic device includes: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the scenario deployment method of the autonomous driving configuration parameters as described in any one of claims 1-2.

5. A computer-readable storage medium having stored thereon a computer program, characterized in that, When executed by the processor, the program implements the scenario deployment method for the autonomous driving configuration parameters as described in any one of claims 1-2.

Images