Sensor type selection method and device based on scene, electronic equipment and storage medium

Abstract

The invention discloses a scene-based sensor type selection method and device, electronic equipment and a storage medium. The method comprises the following steps: determining a verification risk level corresponding to a target scene feature of a vehicle and obtaining characteristic parameters of each type of candidate sensors; selecting at least one group of candidate sensor combination in each candidate sensor based on a preset redundancy range and each characteristic parameter; determining a comprehensive utility index corresponding to each candidate sensor combination, and determining the candidate sensor combination with the maximum comprehensive utility index as a candidate sensor combination to be detected; and verifying the to-be-tested candidate sensor combination according to a verification configuration set corresponding to the verification risk level, and determining the verified to-be-tested candidate sensor combination as a target sensor combination of the vehicle. According to the invention, the problem of cost waste caused by sensor redundancy in the traditional scheme is solved, and the scene adaptability and reliability of the target sensor combination in practical application are enhanced.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a scenario-based sensor selection method, device, electronic device, and storage medium. Background Art

[0002] The reliability of sensor selection is directly related to the safety of autonomous driving. However, traditional sensor selection methods have obvious defects such as not fully considering deployment costs and ignoring the diverse needs of actual application scenarios. As a result, the selection reliability is poor. Therefore, providing a sensor selection method that accurately adapts to the scenario, takes into account cost-effectiveness and high reliability has become a difficult problem that needs to be overcome in the field of commercial vehicle autonomous driving technology. Summary of the Invention

[0003] The present invention provides a scenario-based sensor selection method, device, electronic device and storage medium, which solves the problem of cost waste caused by sensor redundancy in traditional solutions and enhances the scenario adaptability and reliability of the target sensor combination in practical applications.

[0004] One aspect of an embodiment of the present invention provides a scenario-based sensor selection method, including:

[0005] Determine the verification risk level corresponding to the target scenario characteristics of the vehicle and obtain the characteristic parameters of each type of candidate sensor;

[0006] selecting at least one set of candidate sensor combinations from among the candidate sensors based on a preset redundancy range and each of the characteristic parameters;

[0007] Determining the comprehensive utility index corresponding to each candidate sensor combination, and determining the candidate sensor combination having the largest comprehensive utility index as the candidate sensor combination to be tested;

[0008] The candidate sensor combination to be tested is verified according to the verification configuration set corresponding to the verification risk level, and the candidate sensor combination to be tested that passes the verification is determined as the target sensor combination of the vehicle.

[0009] One aspect of an embodiment of the present invention provides a scenario-based sensor selection device, comprising:

[0010] A data acquisition module is used to determine the verification risk level corresponding to the target scene characteristics of the vehicle and obtain the characteristic parameters of each type of candidate sensor;

[0011] a redundancy screening module, configured to select at least one set of candidate sensor combinations from among the candidate sensors based on a preset redundancy range and each of the characteristic parameters;

[0012] The utility selection module is configured to determine a comprehensive utility index corresponding to each of the candidate sensor combinations, and determine the candidate sensor combination with the maximum comprehensive utility index as the candidate sensor combination to be tested.

[0013] The test verification module is configured to verify the candidate sensor combination to be tested according to the verification configuration set corresponding to the verification risk level, and determine the candidate sensor combination that passes the verification as the target sensor combination of the vehicle.

[0014] In another aspect of the embodiments of the present application, an electronic device is provided, which includes:

[0015] at least one processor; and

[0016] a memory in communication with the at least one processor;

[0017] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the scene-based sensor selection method of any of the embodiments of the present application.

[0018] In another aspect of the embodiments of the present application, a computer readable storage medium is provided, which includes computer instructions for enabling a processor to perform the scene-based sensor selection method of any of the embodiments of the present application when executed by the processor.

[0019] In the embodiments of the present application, the characteristic parameters of each type of candidate sensor are obtained, at least one candidate sensor combination is selected in each candidate sensor based on the obtained characteristic parameters and a preset redundancy range, a comprehensive utility index corresponding to each candidate sensor combination is determined, the candidate sensor combination with the maximum comprehensive utility index is determined as the candidate sensor combination to be tested, a verification risk level corresponding to the target scene feature of the vehicle is determined, the candidate sensor combination to be tested is verified according to the verification configuration set corresponding to the verification risk level, and the candidate sensor combination that passes the verification is determined as the target sensor combination of the vehicle. The embodiments of the present application filter the candidate sensor combination by setting the preset redundancy range, solve the problem of cost waste caused by sensor redundancy in the traditional scheme, determine the verification risk level of the target scene in which the vehicle is located based on the target scene feature, and use the differentiated verification configuration set matched with the verification risk level to verify and select the sensor, so that the actual application scene demand is fully considered in the sensor selection process, and the scene adaptability and reliability of the target sensor combination in the actual application are enhanced.

[0020] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 This is a flow chart of a scenario-based sensor selection method provided according to the first embodiment of the present invention;

[0023] Figure 2 is a flow chart of another scenario-based sensor selection method provided according to the second embodiment of the present invention;

[0024] Figure 3 is a flow chart of another scenario-based sensor selection method provided according to the third embodiment of the present invention;

[0025] Figure 4 This is a flowchart of a sensor selection method for a commercial vehicle autonomous driving scenario provided by the fourth embodiment of the present invention;

[0026] Figure 5 A redundancy calculation flow chart provided in the fourth embodiment of the present invention;

[0027] Figure 6 A flow chart of sensor interface deployment provided by the fourth embodiment of the present invention;

[0028] Figure 7 A flowchart of a three-level verification candidate sensor combination provided by the fourth embodiment of the present invention;

[0029] Figure 8 A flow chart of a non-dominated sorting genetic algorithm provided in Example 4 of the present invention;

[0030] Figure 9 A schematic structural diagram of a scenario-based sensor selection device provided in Example 5 of the present invention;

[0031] Figure 10 This is a block diagram of an electronic device for a sensor selection method for an execution scenario provided by Example 6 of the present invention. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] Example 1

[0035] Figure 1 A flowchart of a scenario-based sensor selection method is provided for the first embodiment of the present invention. The embodiment of the present invention is applicable to the scenario of configuring sensors for autonomous commercial vehicles. The method can be executed by a scenario-based sensor selection device. The scenario-based sensor selection device can be implemented in the form of hardware and / or software. The scenario-based sensor selection device can be configured in an electronic device. Figure 1 As shown, the method includes:

[0036] S110: Determine the verification risk level corresponding to the target scene characteristics of the vehicle and obtain characteristic parameters of each type of candidate sensor.

[0037] Among them, the target scene characteristics refer to a series of quantitative parameters that describe the vehicle when it is driving in the target scene, which are used to quantify the various environmental conditions that the vehicle may encounter during driving. For example, the target scene characteristics may include: visibility, obstacle density, road curvature, operating speed and other quantitative parameters.

[0038] The verification risk level refers to the risk level of the target scene based on the characteristics of the target scene. For example, the verification risk level can be divided into multiple levels, such as low risk to high risk. The higher the verification risk level, the more complex the target scene, the greater the safety challenges that the vehicle may encounter when driving in the target scene, and the higher the performance requirements for each type of candidate sensor.

[0039] Candidate sensors refer to sensors that have been pre-screened for the target scene, and may include types such as lidar, millimeter-wave radar, and cameras. Sensors of the same type may include multiple models. For example, lidar sensors may include models such as 64-line lidar and 128-line lidar, millimeter-wave radar sensors may include models such as 77GHz long-range radar and 24GHz short-range radar, and camera sensors may include models such as 2.3-megapixel vehicle-mounted cameras, 8.3-megapixel vehicle-mounted cameras, and three-way forward-facing cameras.

[0040] Characteristic parameters refer to a series of indicators that describe sensor performance. For example, characteristic parameters include but are not limited to reliability, functional weight, and measurement range. Different types and models of sensors have different characteristic parameters.

[0041] Specifically, based on the vehicle's historical driving data and / or industry standards, and in combination with the vehicle's target scene characteristics, the target scenes in which the vehicle is traveling are manually graded to determine the verification risk level corresponding to the vehicle's target scene characteristics, or a risk assessment model is established, and the vehicle's target scene characteristics are input into the risk assessment model. The risk assessment model outputs the verification risk level corresponding to the vehicle's target scene characteristics after calculation. The risk assessment model can be a machine learning model or an expert system model, and the target scene characteristics can be input into the risk assessment model in the form of a matrix or structured data. The characteristic parameters of each type of candidate sensor are obtained by referring to the sensor product manual provided by the sensor supplier, or the sensor configuration file is parsed by an analysis tool to extract the characteristic parameters of each type of candidate sensor from the sensor configuration file.

[0042] S120 : Select at least one candidate sensor combination from among the candidate sensors based on a preset redundancy range and various characteristic parameters.

[0043] The preset redundancy range refers to a pre-set sensor redundancy threshold range, used to guide sensor combination selection. The preset redundancy range may include a maximum redundancy threshold and a minimum redundancy threshold. The range between the maximum redundancy threshold and the minimum redundancy threshold is considered the preset redundancy range. This range is determined based on vehicle safety requirements, cost budget, and / or sensor characteristics.

[0044] A candidate sensor combination is a set of sensors selected from all candidate sensors that need to work together. The redundancy of each candidate sensor in the candidate sensor combination is within a preset redundancy range. Furthermore, the functions of each candidate sensor in the candidate sensor combination can be complementary. For example, a candidate sensor combination may include a camera sensor and a radar sensor, where the camera sensor functions for image recognition and the radar sensor functions for ranging. The coordinated operation of the camera and radar sensors can realize specific services in specific scenarios.

[0045] Specifically, a preset redundancy range is pre-set based on information such as the vehicle's safety requirements, cost budget, and / or sensor characteristics to guide the selection of sensor combinations. The pre-configured preset redundancy range is pre-stored in a buffer of the vehicle or a configuration file of the sensor. By scanning the buffer or the configuration file, the pre-configured preset redundancy range can be obtained. Based on the preset redundancy range, all possible sensor combinations are listed by using an exhaustive method, and then screened among all possible sensor combinations according to characteristic parameters. During the screening process, sensor combinations with excessively high costs and / or excessive power consumption can be eliminated, and the remaining sensor combinations among all possible sensor combinations are recorded as candidate sensor combinations. Alternatively, an optimization algorithm, such as a genetic algorithm and a particle swarm algorithm, is used to search for the optimal sensor combination among the candidate sensors. The objective function of the optimization algorithm can be that the characteristic parameters of each candidate sensor satisfy the preset redundancy range.

[0046] S130: Determine the comprehensive utility index corresponding to each candidate sensor combination, and determine the candidate sensor combination with the largest comprehensive utility index as the candidate sensor combination to be tested.

[0047] The comprehensive utility index can be understood as an indicator used to quantitatively evaluate the overall performance of a candidate sensor combination. The comprehensive utility index can be derived by summing or weighted summing multiple sub-indicators, including perception coverage and deployment cost. A larger value for the comprehensive utility index indicates a better overall performance of the candidate sensor combination.

[0048] The candidate sensor combination to be tested refers to an optimal candidate sensor combination selected from the candidate sensor combination. The candidate sensor combination to be tested has the maximum comprehensive utility index. The candidate sensor combination to be tested needs to be further verified to confirm its applicability, reliability and other performance.

[0049] Specifically, for each screened candidate sensor combination, data is collected for sub-indicators of each sensor in each candidate sensor combination. The sub-indicators may include indicators such as perception coverage and deployment cost. The comprehensive utility index of each candidate sensor combination is determined by summing or weighted summing the obtained sub-indicators. The hierarchical analysis method or the grey relational analysis method is used to score each comprehensive utility index. The comprehensive utility index with the highest score is marked as the one with the maximum comprehensive utility index. The candidate sensor combination corresponding to the maximum comprehensive utility index is determined as the candidate sensor combination to be tested. Alternatively, an algorithm such as a genetic algorithm or a particle swarm optimization algorithm is used with the maximization of the comprehensive utility index as the objective function. The candidate sensor combination with the maximum comprehensive utility index is searched for in the solution space formed by the candidate sensor combinations. The candidate sensor combination corresponding to the maximum comprehensive utility index is determined as the candidate sensor combination to be tested.

[0050] S140 : Verify the candidate sensor combination to be tested according to the verification configuration set corresponding to the verification risk level, and determine the candidate sensor combination to be tested that passes the verification as the target sensor combination of the vehicle.

[0051] The verification configuration set refers to a series of verification standards set for different verification risk levels, used to verify the performance of the candidate sensor combination under test. The verification configuration set may include standards such as interface verification standards and data verification standards. Interface verification standards can be understood as a series of rules for configuring the interface between the sensor combination under test and the external system. Data verification standards can be understood as a series of quantitative indicators used to measure whether the performance of the candidate sensor combination under test meets the standards. For example, interface verification standards may include conditions such as mechanical interface configuration, electrical interface configuration, and communication interface configuration. Data verification standards may include missed detection rate thresholds, delay thresholds, and compensation accuracy thresholds.

[0052] The target sensor combination is the candidate sensor combination to be tested that has been verified using the validation configuration set. The target sensor combination is the sensor combination that will ultimately be used in the vehicle. This target sensor combination will be used in the actual deployment of the vehicle's autonomous driving system to achieve optimal perception and safety.

[0053] Specifically, a set of verification configuration sets for different verification risk levels is pre-configured for verifying the performance of the candidate sensor combination to be tested. The pre-configured verification configuration set is pre-stored in the vehicle's cache or sensor configuration file. By scanning the cache or configuration file, the pre-configured verification configuration set is obtained based on the one-to-one correspondence between the verification risk level and the verification configuration set. Based on the verification configuration set, the candidate sensor combination to be tested can be verified in different scenarios from dimensions such as interface configuration completion and / or data value compliance. The verification scenarios may include extreme weather scenarios and / or sensor failure scenarios. If, in verification scenarios such as extreme weather scenarios and / or sensor failures, the interface configuration completion of the candidate sensor combination to be tested meets the requirements and / or the data values ​​are compliant, that is, the data values ​​are within a preset threshold range, the candidate sensor combination to be tested is determined to have passed verification and is determined as the target sensor combination for the vehicle to be used in actual vehicle operation.

[0054] In an embodiment of the present invention, characteristic parameters of each type of candidate sensor are obtained, at least one candidate sensor combination is selected from each candidate sensor based on the obtained characteristic parameters and a preset redundancy range, a comprehensive utility index corresponding to each candidate sensor combination is determined, the candidate sensor combination with the largest comprehensive utility index is determined as the candidate sensor combination to be tested, the verification risk level corresponding to the target scene characteristics of the vehicle is determined, the candidate sensor combination to be tested is verified according to the verification configuration set corresponding to the verification risk level, and the candidate sensor combination to be tested that passes the verification is determined as the target sensor combination of the vehicle. In an embodiment of the present invention, the problem of cost waste caused by sensor redundancy in traditional solutions is solved by setting a preset redundancy range to screen candidate sensor combinations; by determining the verification risk level of the target scene in which the vehicle is located based on the target scene characteristics, and using a differentiated verification configuration set that matches the verification risk level to verify and select sensors, the actual application scenario requirements are fully considered in the sensor selection process, thereby enhancing the scenario adaptability and reliability of the target sensor combination in actual applications.

[0055] Example 2

[0056] Figure 2 A flowchart of another scenario-based sensor selection method is provided for the second embodiment of the present invention. This embodiment of the present invention refines the above embodiment. Specifically, it refines the specific steps of how to determine the verification risk level, how to obtain sensor characteristic parameters, and how to select candidate sensor combinations.

[0057] like Figure 2 As shown, another scenario-based sensor selection method may include the following specific steps:

[0058] S210: Acquire a target scene corresponding to the vehicle, and extract target scene features of the target scene.

[0059] The target scenario refers to the specific driving scenario of the vehicle. The target scenario can be described from the dimensions of environmental characteristics and / or task requirements. For example, under the dimension of environmental characteristics, the target scenario can be scenes such as highways, urban roads, sunny days, rainy days, pedestrian lanes or non-motor vehicle lanes. Under the dimension of task requirements, the target scenario can be scenes such as following a vehicle, changing lanes, overtaking, navigation positioning or path planning.

[0060] Target scene characteristics refer to a series of quantitative parameters that describe a vehicle driving in a target scene, including at least one of the following: visibility, obstacle density, road curvature, and operating speed.

[0061] Specifically, vehicle operation-related data is collected through environmental perception devices installed on the vehicle, such as cameras or radars. The vehicle operation-related data can be video data or non-video data. The acquired data is processed and analyzed to determine the target scene the vehicle is currently in. Target scene features are extracted from the vehicle operation-related data related to the target scene using image recognition algorithms or data analysis algorithms and other technologies.

[0062] S220: Determine the verification risk level according to the target scenario characteristics and the risk level classifier.

[0063] Among them, the risk level classifier refers to an algorithm model for classifying the risk level of the target scene in which the vehicle is located. For example, the risk level classifier is generated based on at least the training of the random forest classifier. In the training of the risk level classifier, a large amount of target scene feature data labeled with risk levels is used as training samples. By continuously adjusting the parameters of the random forest classifier, the risk level classifier can accurately classify new target scene features into corresponding risk levels.

[0064] Specifically, the target scene features of the target scene in which the vehicle is located are obtained, and the obtained target scene feature data are input into a risk level classifier. The risk level classifier is generated based on random forest classifier training. The risk level classifier analyzes the target scene feature data according to the pre-trained rules and outputs the verification risk level of the target scene corresponding to the target scene feature data.

[0065] S230: Extract characteristic parameters of each type of candidate sensor in the preset configuration file.

[0066] The preset configuration file refers to a pre-set structured data file for storing data. For example, the preset configuration file may store characteristic parameters such as the type, reliability, and functional weight of the candidate sensor.

[0067] Characteristic parameters refer to a series of indicators that describe sensor performance, including at least reliability parameters and functional weight parameters. Reliability parameters can reflect the sensor's ability to work stably and accurately output data under different environmental conditions. Functional weight parameters refer to the weight values ​​assigned based on the importance of the sensor to vehicle safety and / or functional realization in different scenarios, and are used to measure the importance of the sensor function.

[0068] Specifically, a pre-set configuration file is obtained, and the characteristic parameters in the preset configuration file are directly read through a preset data format parser, such as a JSON parser or an XML parser, or the extraction logic of the characteristic parameters is defined through a preset rule language such as Drools or MVEL, and the characteristic parameters of each type of candidate sensor are extracted from the preset configuration file based on the extraction logic.

[0069] S240: Randomly select any number of candidate sensors from each candidate sensor to construct a candidate sensor set.

[0070] The candidate sensor set refers to a set of any number of candidate sensors randomly selected from the candidate sensors. In the candidate sensor set, each candidate sensor has an equal probability of being selected, and there is no limit on the number of candidate sensors in the candidate sensor set.

[0071] Specifically, a random number generator or a random function in a programming language is used to randomly select any number of candidate sensors from all candidate sensors, and all the randomly selected sensors are stored together in the form of an array or a list to form a candidate sensor set. In the candidate sensor set, the same sensor is allowed to be selected multiple times.

[0072] S250 : For each candidate sensor set, determine a redundancy score of the candidate sensor set based on characteristic parameters of the candidate sensors selected in the candidate sensor set.

[0073] Among them, the redundancy score can be understood as a quantitative indicator used to evaluate the redundancy degree of the candidate sensor set. A high redundancy score indicates that even if some sensors in the candidate sensor set fail, the entire sensor set can still ensure the basic safety and functionality of the vehicle.

[0074] Specifically, a pre-set redundancy score calculation formula and characteristic parameters of all candidate sensors in each candidate sensor set are obtained. For each candidate sensor set, the characteristic parameters of all candidate sensors in each candidate sensor set are substituted into the pre-set redundancy score calculation formula to obtain a redundancy score calculation result, which is used as the redundancy score of the candidate sensor set.

[0075] S260: Mark the candidate sensor set whose redundancy score meets the preset redundancy range as a candidate sensor combination.

[0076] Specifically, the calculated redundancy score of each candidate sensor set is compared with a preset redundancy range. If the redundancy score of a candidate sensor set is within the preset redundancy range, a label of a candidate sensor combination is added to the candidate sensor set, or a flag is added to the candidate sensor set that meets the conditions, and the flag is set to "1" to indicate that the candidate sensor set is a candidate sensor combination.

[0077] S270: Determine the comprehensive utility index corresponding to each candidate sensor combination, and determine the candidate sensor combination with the largest comprehensive utility index as the candidate sensor combination to be tested.

[0078] S280: Verify the candidate sensor combination to be tested according to the verification configuration set corresponding to the verification risk level, and determine the candidate sensor combination to be tested that passes the verification as the target sensor combination of the vehicle.

[0079] In an embodiment of the present invention, a target scene corresponding to a vehicle is obtained, and target scene features of the target scene are extracted. A verification risk level is determined based on the target scene features and a risk level classifier. Characteristic parameters of each type of candidate sensor in a preset configuration file are extracted. An arbitrary number of candidate sensors are randomly selected from each candidate sensor to construct a candidate sensor set. For each candidate sensor set, a redundancy score of the candidate sensor set is determined based on the characteristic parameters of the candidate sensors selected from the candidate sensor set. The candidate sensor sets whose redundancy scores meet a preset redundancy range are marked as candidate sensor combinations. A comprehensive utility index corresponding to each candidate sensor combination is determined, and the candidate sensor combination with the largest comprehensive utility index is determined as the candidate sensor combination to be tested. The candidate sensor combination to be tested is verified according to a verification configuration set corresponding to the verification risk level, and the candidate sensor combination to be tested that passes the verification is determined as the target sensor combination for the vehicle. The embodiments of the present invention utilize a risk level classifier to determine and verify risk levels according to unified standards, thereby eliminating the subjectivity of human judgment and improving the reliability of sensor selection. By calculating a redundancy score, candidate sensor combinations that meet a preset redundancy range are screened, ensuring appropriate redundancy of the candidate sensor combinations and avoiding cost waste caused by sensor redundancy. Pre-filtering candidate sensors using the redundancy score avoids verifying candidate sensor combinations on a large number of sensors that do not meet the requirements, thereby improving sensor selection efficiency.

[0080] Furthermore, the embodiment of the present invention further refines step S250. Specifically, it refines how to determine the redundancy score, including: determining the redundancy score of the candidate sensor set based on the characteristic parameters of the candidate sensors selected from the candidate sensor set, including: substituting the reliability parameter and the function weight parameter in the characteristic parameters of the candidate sensors selected from the candidate sensor set into a preset redundancy evaluation formula; and using the evaluation result of the preset redundancy evaluation formula as the redundancy score of the candidate sensor set; wherein the preset redundancy evaluation formula at least includes:

[0081]

[0082] Among them, R j represents the redundancy score of the j-th candidate sensor set, p represents the number of types of candidate sensors selected in the j-th candidate sensor set, k represents the k-th type of candidate sensor selected in the j-th candidate sensor set, and η k represents the reliability parameter of the k-th candidate sensor selected from the j-th candidate sensor set, γ k It represents the function weight parameter of the k-th candidate sensor selected from the j-th candidate sensor set.

[0083] Specifically, obtain the preset redundancy evaluation formula And the characteristic parameters of all candidate sensors in the candidate sensor set, for each candidate sensor characteristic parameter, extract the reliability parameter η of each candidate sensor k and the function weight parameter γ k , the reliability parameters η of all candidate sensors in the candidate sensor set k and the function weight parameter γ k Substitute into the preset redundancy evaluation formula In the example, we get the evaluation result R for the candidate sensor set. j , the evaluation result R j as the redundancy score of the candidate sensor set.

[0084] Example 3

[0085] Figure 3 A flowchart of another scenario-based sensor selection method is provided for the third embodiment of the present invention. This embodiment of the present invention refines the above embodiment. Specifically, it refines the specific steps of how to determine the candidate sensor combination to be tested and how to determine the target sensor combination.

[0086] like Figure 3 As shown, another scenario-based sensor selection method may include the following specific steps:

[0087] S310: Determine the verification risk level corresponding to the target scene characteristics of the vehicle and obtain characteristic parameters of each type of candidate sensor.

[0088] S320: Select at least one candidate sensor combination from each candidate sensor based on a preset redundancy range and each characteristic parameter.

[0089] S330 : For each candidate sensor combination, determine the sensing coverage of the candidate sensors in the candidate sensor combination through Monte Carlo simulation calibration.

[0090] Among them, the perception coverage range refers to the spatial range in which the candidate sensor can detect the target object. The perception coverage range of the sensor can be described by parameters such as distance, angle and resolution. The perception coverage range of the sensor is affected by parameters such as distance, angle and resolution. The size of the perception coverage range can affect the sensor's ability to perceive the surrounding environment.

[0091] Specifically, for each candidate sensor combination, the Monte Carlo simulation method is used to set a large number of random environmental conditions and sensor operating parameters to simulate the sensor's perception in different scenarios. By statistically analyzing multiple simulation results, the perception coverage of each candidate sensor in the combination is determined.

[0092] S340: Determine a comprehensive utility index of the corresponding candidate sensor combination according to the sensing coverage of each candidate sensor.

[0093] Specifically, for each candidate sensor combination, the corresponding perception coverage range is substituted into a preset utility evaluation formula, the formula is calculated, and the utility evaluation result of the preset utility evaluation formula is obtained, and the utility evaluation result is used as the comprehensive utility index of the candidate sensor combination.

[0094] S350: Determine the maximum comprehensive utility index among the comprehensive utility indexes, and record the candidate sensor combination corresponding to the maximum comprehensive utility index as the candidate sensor combination to be tested.

[0095] Specifically, among the comprehensive utility indices of all candidate sensor combinations, the comprehensive utility indices of all candidate sensor combinations are compared, and the comprehensive utility indices with the largest value are found, and the candidate sensor combination corresponding to the comprehensive utility indices with the largest value is marked as the candidate sensor combination to be tested.

[0096] S360: Extract the verification configuration set associated with the verification risk level in the preset test configuration file.

[0097] A preset test profile is a pre-created structured document used to store various test-related parameters. It contains the target scenario's verification risk level and the corresponding verification configuration set. Different verification risk levels correspond to different verification configuration sets within the preset test profile.

[0098] A verification configuration set refers to a series of verification standards set for different verification risk levels, used to verify the performance of the candidate sensor combination under different verification risk levels. The verification configuration set includes at least a verification threshold, a verification interface configuration, and a verification data configuration. The verification threshold refers to the critical value set in the verification configuration set to determine whether the working data of the candidate sensor combination under test meets the requirements. If the working data exceeds or falls below the verification threshold, the verification result of the candidate sensor combination under test is considered to have failed. The verification interface configuration refers to the standardized parameters, protocols, and test specifications set for the hardware physical interface and software communication interface of each sensor in the candidate sensor combination under test during the verification process of the candidate sensor combination under test, to ensure the consistency of the verification environment with the actual vehicle interface. The verification data configuration defines the type and / or format of the test data required during the verification process, and is used to simulate the data in the actual operating environment.

[0099] Specifically, by matching keywords or structured language queries, a verification configuration set associated with the verification risk level is extracted from the preset test configuration file. The verification configuration set contains at least key information such as verification threshold, verification interface configuration and verification data configuration, which is used to guide subsequent test verification work.

[0100] S370: Deploy a test environment for the candidate sensor combination to be tested according to the verification interface configuration and the verification data configuration of the verification configuration set.

[0101] Among them, the test environment refers to a specific environment built for testing the candidate sensor combination to be tested. For example, the test environment includes at least one of the following: a digital twin simulation environment, a hardware-in-the-loop test environment, and a real vehicle test environment.

[0102] Specifically, according to the verification interface configuration and verification data configuration of the verification configuration set, a test environment for the candidate sensor combination to be tested is deployed in environments such as digital twin simulation, hardware-in-the-loop testing, or actual vehicle testing.

[0103] S380: Collect working data of the candidate sensor combination to be tested working in the test environment.

[0104] Specifically, the candidate sensor combination to be tested is run in the deployed test environment, and various working data generated during the working process of the sensor combination are collected in real time using hardware-triggered collection or software-timed collection methods.

[0105] S390: Compare the verification threshold of the verification configuration set with the working data of each candidate sensor combination to be tested to obtain a verification result of the candidate sensor combination to be tested.

[0106] The verification result refers to the conclusion drawn by comparing the verification threshold with the working data of the candidate sensor combination to be tested. The verification result can be classified as pass or fail, and is used to determine whether the candidate sensor combination to be tested is suitable as the target sensor combination of the vehicle. When the verification result is pass, the candidate sensor combination to be tested is used as the target sensor combination of the vehicle.

[0107] Specifically, the verification threshold specified in the verification configuration set is compared one by one with the working data of each candidate sensor combination to be tested to obtain a comparison result, which is used as the verification result of the candidate sensor combination to be tested. The verification result can be divided into two categories: pass and fail. If the working data meets the verification threshold, the verification result of the candidate sensor combination to be tested is considered to be pass; otherwise, it is considered to be fail.

[0108] S3100: Set the candidate sensor combination to be tested with a passed verification result as the target sensor combination for the vehicle.

[0109] Specifically, among the verification results of all candidate sensor combinations to be tested, the candidate sensor combinations to be tested with a passing verification result are screened out, and the candidate sensor combinations to be tested with a passing verification result are set as the target sensor combination of the vehicle.

[0110] In an embodiment of the present invention, a verification risk level corresponding to a target scenario characteristic of a vehicle is determined, and characteristic parameters of each type of candidate sensor are obtained. At least one candidate sensor combination is selected from each candidate sensor based on a preset redundancy range and the characteristic parameters. For each candidate sensor combination, a perception coverage range of the candidate sensors within the candidate sensor combination is determined through Monte Carlo simulation calibration. A comprehensive utility index corresponding to the candidate sensor combination is determined based on the perception coverage range of each candidate sensor. A maximum comprehensive utility index within each comprehensive utility index is determined, and the candidate sensor combination corresponding to the maximum comprehensive utility index is recorded as the candidate sensor combination to be tested. A verification configuration set associated with the verification risk level is extracted from a preset test configuration file. A test environment for the candidate sensor combination to be tested is deployed according to the verification interface configuration and verification data configuration of the verification configuration set. Working data of the candidate sensor combination to be tested operating in the test environment is collected. A verification threshold of the verification configuration set is compared with the working data of each candidate sensor combination to obtain a verification result for the candidate sensor combination to be tested. The candidate sensor combination to be tested that passes the verification result is set as the target sensor combination for the vehicle. In the embodiments of the present invention, the sensor's perception coverage can be quickly and efficiently acquired through Monte Carlo simulation, thereby reducing cost investment and shortening the selection cycle. By comparing the verification threshold with the working data, the candidate sensor combinations to be tested corresponding to the working data that meets the verification threshold are screened out, and the candidate sensor combinations to be tested that meet the performance standards under actual working conditions can be accurately locked in, effectively improving the reliability of sensor selection.

[0111] Furthermore, the embodiment of the present invention further refines step S340. Specifically, the specific steps for determining the comprehensive utility index are refined, including: for each candidate sensor combination, substituting the corresponding sensing coverage range into a preset utility evaluation formula; obtaining the utility evaluation result of the preset utility evaluation formula as the comprehensive utility index of the candidate sensor combination; wherein the preset utility evaluation formula includes at least:

[0112]

[0113] Among them, F j represents the utility evaluation result of the jth group of candidate sensor combinations, n represents the total number of candidate sensors in the jth group of candidate sensor combinations, S i represents the sensing coverage of the i-th candidate sensor in the j-th candidate sensor combination, w i represents the category weight of the i-th candidate sensor in the j-th candidate sensor combination, C j,total represents the overall solution cost of the jth group of candidate sensor combinations, α and β represent the efficiency weight coefficient and cost weight coefficient of the preset utility evaluation formula respectively.

[0114] Among them, the preset utility evaluation formula refers to a pre-set mathematical formula for calculating the utility evaluation results of the candidate sensor combination. The preset utility evaluation formula comprehensively considers multiple factors such as perception coverage and cost. The weight coefficient in the preset utility evaluation formula can adjust the degree of influence of each factor on the utility evaluation result.

[0115] The utility evaluation result refers to the value calculated by substituting the relevant parameters of the candidate sensor combination to be tested into the preset utility evaluation formula. The higher the value, the better the comprehensive utility of the candidate sensor combination to be tested.

[0116] Category weights refer to the weights assigned to different types of candidate sensors based on their importance in the sensor combination. Sensor types that have a greater impact on perception performance in specific scenarios can be assigned higher category weights.

[0117] The efficiency weight coefficient is used to adjust the impact of the perception coverage-related term on the utility evaluation results in the preset utility evaluation formula. A larger value for the efficiency weight coefficient indicates a more significant impact of perception coverage on the utility evaluation results.

[0118] The cost weight coefficient is a factor used to adjust the impact of cost-related items on the utility evaluation results within the preset utility evaluation formula. A larger cost weight coefficient indicates a more significant impact of the overall solution cost on the utility evaluation results.

[0119] Specifically, for each candidate sensor combination, the overall solution cost C of each candidate sensor combination is obtained. j,total , the type weight w of each candidate sensor in each candidate sensor combination i , the preset efficiency weight coefficient α and cost weight coefficient β of each candidate sensor combination are obtained to obtain the preset utility evaluation formula The overall program cost C j,total , category weight w i When the efficiency weight coefficient α and the cost weight coefficient β are known, the perception coverage S of the candidate sensor in the candidate sensor combination is determined by Monte Carlo simulation calibration. i , the sensing coverage S i A preset utility evaluation formula is input to obtain a calculation result of the preset utility evaluation formula, and the calculation result is used as a comprehensive utility index of the candidate sensor combination.

[0120] Example 4

[0121] Figure 4 This is a flowchart of a sensor selection method for a commercial vehicle autonomous driving scenario provided by the fourth embodiment of the present invention; Figure 5 A redundancy calculation flow chart provided in Example 4 of the present invention;

[0122] Figure 6 A sensor interface deployment flowchart provided for the fourth embodiment of the present application; Figure 7 A three-level verification candidate sensor combination flowchart provided for the fourth embodiment of the present application; Figure 8 A non-dominated sorting genetic algorithm flowchart provided for the fourth embodiment of the present application.

[0123] As Figure 4 shown, a sensor selection method applied to a commercial vehicle automatic driving scene can include the following specific steps:

[0124] Randomly select any number of candidate sensors within each candidate sensor to construct a candidate sensor set, explicitly define the reliability parameter and the function weight parameter of each candidate sensor set, and calculate the redundancy score of each candidate sensor set according to the formula , wherein R j represents the redundancy score of the jth candidate sensor set, P represents the number of types of candidate sensors selected in the jth candidate sensor set, k represents the kth type of candidate sensor selected in the jth candidate sensor set, η k represents the reliability parameter of the kth type of candidate sensor selected in the jth candidate sensor set, γ k represents the function weight parameter of the kth type of candidate sensor selected in the jth candidate sensor set, and the preset redundancy range of the redundancy score of each candidate sensor set is defined as [R max ≥ R j ≥ R min ], R max is the maximum redundancy threshold, and R min is the minimum redundancy threshold. The candidate sensor set corresponding to the redundancy score greater than the minimum redundancy threshold and less than the maximum redundancy threshold is marked as a candidate sensor combination. The minimum redundancy threshold and the maximum redundancy threshold constitute the preset redundancy range. The reliability parameter and the function weight parameter constitute the characteristic parameter. If the redundancy score does not satisfy the redundancy range, an alarm is triggered and the candidate sensor set is determined again. The above redundancy score calculation process can be as shown in Figure 5 . For each candidate sensor combination, the comprehensive utility index of the candidate sensor combination is calculated according to , wherein F j represents the utility evaluation result of the jth candidate sensor combination, n represents the total number of candidate sensors in the jth candidate sensor combination, S i represents the perception coverage range of the ith candidate sensor in the jth candidate sensor combination, w i represents the type weight of the ith candidate sensor in the jth candidate sensor combination, and C j,totalrepresents the overall solution cost of the jth group of candidate sensor combinations, α and β represent the efficiency weight coefficient and cost weight coefficient of the preset utility evaluation formula respectively, and the maximum comprehensive utility index among the comprehensive utility indexes is determined using the non-dominated sorting genetic algorithm. The candidate sensor combination corresponding to the maximum comprehensive utility index is recorded as the candidate sensor combination to be tested; a scenario feature matrix is ​​constructed according to the scenario in which the commercial vehicle is to travel, and the matrix parameters of the scenario feature matrix may include: visibility, obstacle density, road curvature and operating speed, and the scenario feature matrix is ​​input into a risk level classifier generated based on random forest classifier training, and the risk level classifier outputs the scenario risk level based on the scenario feature matrix; configuration parameters associated with the scenario risk level are extracted from the preset test configuration file of the vehicle, wherein the configuration parameters may include the positioning pin accuracy of the mechanical interface layer, the anti-vibration threshold of the locking device, the waterproof shell type, the temperature control range of the active cooling duct, and may also include the adaptive power supply bus voltage, maximum power, and dynamic adjustment of power supply priority of the electrical interface layer, and may also include the unified data protocol and low-latency conversion threshold of the communication interface layer; such as Figure 6 As shown in , the candidate sensor combination to be tested can be deployed and configured according to the configuration parameters. After the deployment and configuration are completed, the candidate sensor combination to be tested can be added to the sensor network, and the candidate sensor combination to be tested that has not completed the deployment and configuration can be diagnosed for faults; after the deployment and configuration are completed, as shown in Figure 7 As shown, the candidate sensor combination to be tested is subjected to three-level verification, including: digital twin simulation verification, hardware-in-the-loop test verification and real vehicle verification, among which digital twin simulation verification refers to simulating extreme scenarios to verify the missed detection rate of the sensor combination, hardware-in-the-loop test verification refers to simulating sensor failures and testing the redundant switching response time, and real vehicle verification refers to further judging whether the candidate sensor combination meets the standards through real vehicle road test data. The real vehicle road test data feeds back to the optimization algorithm to obtain a verification report after three-level verification. The verification report in which various indicators (such as sensor coverage) meet the standards is recorded as a verification result of the three-level verification. The candidate sensor combination to be tested with a passed verification result is set as the target sensor combination for commercial vehicles, and the target sensor combination is deployed in mass production. If the verification result fails, the selection is returned.

[0125] like Figure 8As shown, in the step of determining the to-be-tested candidate sensor combination, the non-dominated sorting genetic algorithm can be used to determine the to-be-tested candidate sensor combination, wherein the algorithm flow of the non-dominated sorting genetic algorithm comprises: initializing each to-be-tested candidate sensor combination, calculating the comprehensive utility index of each to-be-tested candidate sensor combination, taking the comprehensive utility index as the objective function of the non-dominated sorting genetic algorithm, performing fast non-dominated sorting, calculating the crowding degree and other operations on the objective function, so as to ensure the diversity and optimization effect of the to-be-tested candidate sensor combination, through continuous selection of parent individuals, crossover and mutation and other genetic operations, when the termination condition is met, that is, the maximum target function, a three-dimensional Pareto front solution set is generated, and the three-dimensional Pareto front solution set is the optimal to-be-tested candidate sensor.

[0126] Further, the embodiment of the present application provides a sensor combination selection step in three specific target scenarios. In the target scenario of intercity expressway freight transportation, the target scenario characteristics of expressway freight transportation include an operating speed of 80-100 km / h, a cost of 5000 yuan, and safety requirements of expressway freight transportation. The random forest classifier is used to obtain the verification risk level (including 5 levels, Lv1, Lv2, Lv3, Lv4 and Lv5) of the intercity expressway freight transportation scene as Lv3. Through the dynamic selection engine (non-dominated sorting genetic algorithm), the finally selected to-be-tested candidate sensor combination is: 2 4D millimeter wave radars, 1 long-focus camera and 2 short-wave infrared sensors. According to the verification risk level, the to-be-tested candidate sensor combination is installed through the platform interface according to the quick disassembly structure (tolerance ±0.5 mm), and the adaptive power supply bus is used to preferentially allocate 200W power to the radar. Finally, through three-level verification, the rain and fog scene miss detection rate of the to-be-tested candidate sensor combination in digital twin simulation verification is <3%, the power supply switching response in hardware-in-the-loop test verification is <100ms, and the road test is 5000km without failure in real vehicle verification. The to-be-tested candidate sensor combination is used as the target sensor combination for mass production and deployment, which can achieve the effect of reducing the total cost by 22%, the system delay <150ms and the rain and fog miss detection rate meeting the standard.

[0127] For the mining dump truck scenario, the target scenario characteristics include dust concentration, vibration, protection level IP69K, and continuous working time. The random forest classifier is used to obtain the verification risk level of Lv5 for the intercity highway freight yard. Through redundancy strategy calculation, the candidate sensor combination to be tested uses a dust-resistant Class 4 mechanical lidar and a vibration compensation module with a compensation accuracy of ±0.1g. According to the verification risk level, rubber-metal composite shock-absorbing brackets and dust filter ducts (PM2.5 filtration >99%) are installed during platform interface deployment, and shielded twisted pair cables are used for the communication interface to resist electromagnetic interference. The point cloud distortion rate in the verification phase was less than 2% after simulated dust penetration testing. After hardware-in-the-loop testing, there was no looseness under 10g acceleration on the vibration table. During actual vehicle verification, the average daily failure rate was reduced to 0.7 times. This candidate sensor combination to be tested was used as the target sensor combination for mass production deployment, which can reduce the cloud distortion rate from 12% to 1.8%.

[0128] In the urban smart bus scenario, target scenario characteristics include pedestrian recognition, platooning control, V2X collaboration requirements, and 5G dynamic loading rates in urban BRT scenarios. A random forest classifier was used to determine the verification risk level for the urban smart bus scenario, which was Lv4. The dynamic selection engine output a candidate sensor combination for testing, including a solid-state lidar and four millimeter-wave radars. The verification phase simulated sudden pedestrian crossings using a digital twin. Virtual V2X signals were injected during hardware-in-the-loop testing. Combined with actual vehicle testing during peak traffic conditions (AEB triggering distance of 15m), the candidate sensor combination was used as the target sensor combination for mass production deployment. This enabled multi-vehicle collaboration in complex urban scenarios, reducing the risk of data leakage by 90%. During the verification phase, 20 typical scenarios (such as platform stops) were preloaded simultaneously. The candidate sensor combination achieved a pedestrian trajectory prediction model accuracy of 92.7% and a platoon spacing control accuracy of ±0.2m.

[0129] Furthermore, an embodiment of the present invention provides a judgment standard for determining the verification risk level based on target scene characteristics, as shown in Table 1.

[0130] Table 1 Verification risk level judgment criteria

[0131]

[0132] Furthermore, an embodiment of the present invention provides a table of correspondence between target scenarios and verification risk levels, as shown in Table 2.

[0133] Table 2. Correspondence between target scenarios and verification risk levels

[0134]

[0135] Example 5

[0136] Figure 9This is a structural diagram of a scenario-based sensor selection device provided in Example 5 of the present invention. Figure 9 As shown, the scenario-based sensor selection device specifically includes: a data acquisition module 510 , a data acquisition module 520 , a utility selection module 530 and a test verification module 540 .

[0137] The data acquisition module 510 is used to determine the verification risk level corresponding to the target scene characteristics of the vehicle and obtain characteristic parameters of each type of candidate sensor;

[0138] a redundancy screening module 520 for selecting at least one candidate sensor combination from among the candidate sensors based on a preset redundancy range and various characteristic parameters;

[0139] The utility selection module 530 is configured to determine the comprehensive utility index corresponding to each candidate sensor combination, and determine the candidate sensor combination with the largest comprehensive utility index as the candidate sensor combination to be tested;

[0140] The test verification module 540 is configured to verify the candidate sensor combination to be tested according to the verification configuration set corresponding to the verification risk level, and determine the candidate sensor combination to be tested that passes the verification as the target sensor combination of the vehicle.

[0141] Furthermore, the data acquisition module 510 includes: a feature acquisition unit, used to acquire the target scene corresponding to the vehicle and extract the target scene features of the target scene, wherein the target scene features include at least one of the following: visibility, obstacle density, road curvature and operating speed; a level determination unit, used to determine the verification risk level based on the target scene features and the risk level classifier, wherein the risk level classifier is generated based on random forest classifier training; a parameter extraction unit, used to extract characteristic parameters of each type of candidate sensor in the preset configuration, wherein the characteristic parameters include reliability parameters and functional weight parameters.

[0142] Furthermore, the redundancy screening module 520 includes: a sensor selection unit, which is used to randomly select any number of candidate sensors from each candidate sensor to construct a candidate sensor set; a score determination unit, which is used to determine the redundancy score of the candidate sensor set based on the characteristic parameters of the candidate sensors selected from the candidate sensor set for each candidate sensor set; and a sensor marking unit, which is used to mark the candidate sensor set whose redundancy score meets the preset redundancy range as a candidate sensor combination.

[0143] The scoring determination unit is specifically configured to substitute the reliability parameter and the function weight parameter in the characteristic parameters of the candidate sensor selected from the candidate sensor set into a preset redundancy evaluation formula; and use the evaluation result of the preset redundancy evaluation formula as the redundancy score of the candidate sensor set; wherein the preset redundancy evaluation formula at least includes:

[0144]

[0145] Among them, R j represents the redundancy score of the j-th candidate sensor set, P represents the number of types of candidate sensors selected in the j-th candidate sensor set, k represents the k-th type of candidate sensor selected in the j-th candidate sensor set, η k represents the reliability parameter of the kth candidate sensor selected from the jth group of candidate sensors, γ k It represents the function weight parameter of the k-th candidate sensor selected from the j-th candidate sensor set.

[0146] Furthermore, the utility selection module 530 includes: a range determination unit, which is used to determine the perception coverage range of the candidate sensors in the candidate sensor combination through Monte Carlo simulation calibration for each candidate sensor combination; an index determination unit, which is used to determine the comprehensive utility index of the corresponding candidate sensor combination according to the perception coverage range of each candidate sensor; and a combination determination unit, which is used to determine the maximum comprehensive utility index among the comprehensive utility indexes, and record the candidate sensor combination corresponding to the maximum comprehensive utility index as the candidate sensor combination to be tested.

[0147] The indicator determination unit is specifically configured to substitute, for each candidate sensor combination, the corresponding sensing coverage range into a preset utility evaluation formula; and obtain a utility evaluation result of the preset utility evaluation formula as a candidate sensor combination to be tested for the candidate sensor combination; wherein the preset utility evaluation formula includes at least:

[0148]

[0149] Among them, F j represents the utility evaluation result of the jth group of candidate sensor combinations, n represents the total number of candidate sensors in the jth group of candidate sensor combinations, S i represents the sensing coverage of the i-th candidate sensor in the j-th candidate sensor combination, w i represents the category weight of the i-th candidate sensor in the j-th candidate sensor combination, C j,total represents the overall solution cost of the jth group of candidate sensor combinations, α and β represent the efficiency weight coefficient and cost weight coefficient of the preset utility evaluation formula, respectively.

[0150] Optionally, the test verification module 540 includes: a test verification unit, used to extract a verification configuration set associated with the verification risk level in a preset test configuration file, wherein the verification configuration set includes at least a verification threshold, a verification interface configuration and a verification data configuration; an environment deployment unit, used to deploy the test environment of the candidate sensor combination to be tested according to the verification interface configuration and the verification data configuration of the verification configuration set, wherein the test environment includes at least one of a digital twin simulation environment, a hardware-in-the-loop test environment and a real vehicle test environment; a data acquisition unit, used to collect working data of the candidate sensor combination to be tested working in the test environment; a result acquisition unit, used to compare the verification threshold of the verification configuration set with the working data of each candidate sensor combination to be tested to obtain a verification result of the candidate sensor combination to be tested; a target combination determination unit, used to set the candidate sensor combination to be tested whose verification result is passed as the target sensor combination of the vehicle.

[0151] The scenario-based sensor selection device provided in the embodiment of the present invention can execute the scenario-based sensor selection method provided in any embodiment of the present invention, and has the corresponding beneficial effects of executing the method.

[0152] Example 6

[0153] A sixth embodiment of the present invention provides an electronic device, a computer-readable storage medium, and a computer program product for executing a scenario-based sensor selection method.

[0154] Figure 10 A schematic diagram of the structure of an electronic device that can be used to implement any of the scene-based sensor selection methods of the embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown in the embodiments of the present invention, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of the present invention described and / or required herein.

[0155] like Figure 10As shown, the electronic device includes at least one processor 11, and a memory, such as a Read-Only Memory (ROM) 12, a Random Access Memory (RAM) 13, etc., connected to the at least one processor 11 in communication. The memory stores computer programs executable by the at least one processor 11, and the processor 11 can perform various appropriate actions and processes according to the computer programs stored in the ROM 12 or loaded into the RAM 13 from the storage unit 18. Various programs and data required for operation of the electronic device can also be stored in the RAM 13. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An Input / Output (I / O) interface 15 is also connected to the bus 14.

[0156] Various components in the electronic device are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc., an output unit 17, such as various types of displays, a speaker, etc., a storage unit 18, such as a magnetic disk, an optical disk, etc., and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0157] The processor 11 can be various general and / or special purpose processing components having processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit, a graphics processing unit, various special-purpose artificial intelligence computing chips, various processors running machine learning model algorithms, a digital signal processor, and any appropriate processor, controller, microcontroller, etc. The processor 11 performs various methods and processes described above, such as the scene-based sensor selection method.

[0158] In some embodiments, the scene-based sensor selection method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the scene-based sensor selection method can be performed. Alternatively, in other embodiments, the processor 11 can be configured as the scene-based sensor selection method by any other appropriate means, such as by means of firmware.

[0159] Various implementations of the systems and techniques described above in the embodiments of the present invention may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays, application specific integrated circuits, application specific standard products, system-on-chip systems, load programmable logic devices, computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: being implemented in one or more computer programs that are executable and / or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0160] The computer programs for implementing the methods of the embodiments of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer programs are executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0161] In the context of an embodiment of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, RAM, ROM, an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0162] To provide interaction with a user, the systems and techniques described herein can be implemented on a device having: a display device (e.g., a cathode ray tube or liquid crystal display monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0163] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include local area networks, wide area networks, blockchain networks, and the Internet.

[0164] A computing system may include clients and servers. The clients and servers are generally remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a host product within a cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosts and virtual private server services.

[0165] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0166] The above specific embodiments do not constitute a limitation on the scope of protection of the embodiments of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A scenario-based sensor selection method, characterized in that: The method comprises: Determine the verification risk level corresponding to the target scenario characteristics of the vehicle and obtain the characteristic parameters of each type of candidate sensor; selecting at least one set of candidate sensor combinations from among the candidate sensors based on a preset redundancy range and each of the characteristic parameters; Determining the comprehensive utility index corresponding to each candidate sensor combination, and determining the candidate sensor combination having the largest comprehensive utility index as the candidate sensor combination to be tested; The candidate sensor combination to be tested is verified according to the verification configuration set corresponding to the verification risk level, and the candidate sensor combination to be tested that passes the verification is determined as the target sensor combination of the vehicle.

2. The method according to claim 1, characterized in that Determining the verification risk level corresponding to the target scene characteristics of the vehicle and obtaining characteristic parameters of each type of candidate sensor includes: Acquire a target scene corresponding to the vehicle, and extract target scene features of the target scene, wherein the target scene features include at least one of the following: visibility, obstacle density, road curvature, and operating speed; Determining the verification risk level according to the target scenario characteristics and a risk level classifier, wherein the risk level classifier is generated based on random forest classifier training; The characteristic parameters of each type of candidate sensor in a preset configuration file are extracted, wherein the characteristic parameters include reliability parameters and function weight parameters.

3. The method according to claim 1, characterized in that The selecting at least one set of candidate sensor combinations from among the candidate sensors based on the preset redundancy range and the characteristic parameters includes: Randomly selecting any number of the candidate sensors from among the candidate sensors to construct a candidate sensor set; For each candidate sensor set, determining a redundancy score of the candidate sensor set based on the characteristic parameters of the candidate sensor selected in the candidate sensor set; The candidate sensor set whose redundancy score meets the preset redundancy range is marked as the candidate sensor combination.

4. The method according to claim 3, characterized in that The determining the redundancy score of the candidate sensor set based on the characteristic parameters of the candidate sensors selected from the candidate sensor set includes: Substituting the reliability parameter and the function weight parameter in the characteristic parameters of the candidate sensor selected from the candidate sensor set into a preset redundancy evaluation formula; Using the evaluation result of the preset redundancy evaluation formula as the redundancy score of the candidate sensor set; The preset redundancy evaluation formula includes at least: Among them, R j represents the redundancy score of the j-th candidate sensor set, p represents the number of types of candidate sensors selected in the j-th candidate sensor set, k represents the k-th type of candidate sensor selected in the j-th candidate sensor set, and η k represents the reliability parameter of the k-th candidate sensor selected from the j-th candidate sensor set, γ k It represents the function weight parameter of the k-th candidate sensor selected from the j-th candidate sensor set.

5. The method according to claim 1, characterized in that: The determining of the comprehensive utility index corresponding to each candidate sensor combination and determining the candidate sensor combination having the largest comprehensive utility index as the candidate sensor combination to be tested includes: For each candidate sensor combination, determining the sensing coverage of the candidate sensors in the candidate sensor combination through Monte Carlo simulation calibration; Determining the comprehensive utility index corresponding to the candidate sensor combination according to the sensing coverage of each candidate sensor; The maximum comprehensive utility index among the comprehensive utility indexes is determined, and the candidate sensor combination corresponding to the maximum comprehensive utility index is recorded as the candidate sensor combination to be tested.

6. The method according to claim 5, characterized in that Determining the comprehensive utility index corresponding to the candidate sensor combination according to the sensing coverage of each candidate sensor includes: For each candidate sensor combination, substituting the corresponding sensing coverage into a preset utility evaluation formula; Obtaining a utility evaluation result of the preset utility evaluation formula as the comprehensive utility index of the candidate sensor combination; The preset utility evaluation formula at least includes: Among them, F j represents the utility evaluation result of the jth group of candidate sensor combinations, n represents the total number of candidate sensors in the jth group of candidate sensor combinations, S i represents the sensing coverage of the i-th candidate sensor in the j-th candidate sensor combination, w i represents the category weight of the i-th candidate sensor in the j-th candidate sensor combination, C j,total represents the overall solution cost of the jth group of candidate sensor combinations, α and β represent the efficiency weight coefficient and cost weight coefficient of the preset utility evaluation formula respectively.

7. The method according to claim 1, characterized in that: The verifying the candidate sensor combination to be tested according to the verification configuration set corresponding to the verification risk level, and determining the candidate sensor combination to be tested that passes the verification as the target sensor combination of the vehicle, includes: Extracting the verification configuration set associated with the verification risk level in a preset test configuration file, wherein the verification configuration set at least includes a verification threshold, a verification interface configuration, and a verification data configuration; Deploying a test environment for the candidate sensor combination to be tested according to the verification interface configuration and the verification data configuration of the verification configuration set, wherein the test environment includes at least one of a digital twin simulation environment, a hardware-in-the-loop test environment, and a real vehicle test environment; Collecting working data of the candidate sensor combination to be tested working in the test environment; Comparing the verification threshold of the verification configuration set with the working data of each of the candidate sensor combinations to be tested to obtain a verification result of the candidate sensor combination to be tested; The candidate sensor combination to be tested with a verification result of passing is set as the target sensor combination of the vehicle.

8. A scenario-based sensor selection device, characterized in that: The device comprises: A data acquisition module is used to determine the verification risk level corresponding to the target scene characteristics of the vehicle and obtain the characteristic parameters of each type of candidate sensor; a redundancy screening module, configured to select at least one set of candidate sensor combinations from among the candidate sensors based on a preset redundancy range and each of the characteristic parameters; a utility selection module, configured to determine a comprehensive utility index corresponding to each candidate sensor combination, and determine the candidate sensor combination having the largest comprehensive utility index as the candidate sensor combination to be tested; A test verification module is used to verify the candidate sensor combination to be tested according to the verification configuration set corresponding to the verification risk level, and determine the candidate sensor combination to be tested that passes the verification as the target sensor combination of the vehicle.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the scenario-based sensor selection method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the scenario-based sensor selection method according to any one of claims 1 to 4 when executed.

Images