Method and system for testing and evaluating performance of high-level automatic driving system

By setting distance boundaries and scene label identification, extracting and evaluating data segments of interest in high-level autonomous driving systems, the problem of inability to comprehensively evaluate the performance of high-level autonomous driving systems in the prior art is solved, and efficient performance evaluation and safety are achieved Sexual assessment.

CN119043735APending Publication Date: 2024-11-29CHONGQING VEHICLE TEST & RES INST CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411120275.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-15
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

The prior art lacks test and evaluation technology based on actual driving scenarios for high-level autonomous driving systems, and cannot fully evaluate its performance and safety.

Method used

By setting the distance boundary of the test vehicle, extracting the data segments of interest, performing performance evaluation based on the scene labels, and statistical analysis is carried out in combination with the evaluation results, a comprehensive performance evaluation of high-level autonomous driving systems is achieved.

Benefits of technology

It improves the efficiency and accuracy of performance testing of high-level autonomous driving systems, and can comprehensively evaluate system performance in different driving scenarios to ensure safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119043735A_ABST
    Figure CN119043735A_ABST
Patent Text Reader

Abstract

The invention discloses a high-level automatic driving system performance test evaluation method and system, and the method comprises the steps: firstly, setting a distance boundary for a target according to the safety distance between an environment vehicle and a test vehicle, and when the environment vehicle enters the range of the distance boundary, starting to slice test data, and slicing is finished until the environment vehicle leaves the distance boundary range, and interested data segments are extracted. Then, scene recognition is carried out on all the interested data segments according to the operation parameters in the data segments, and the driving scene of all the interested data segments is determined. And then, according to each operation parameter of the interested data segment, a corresponding evaluation method is adopted to evaluate the performance of the high-level automatic driving system in a time period corresponding to the interested data segment, and a performance evaluation result is obtained. And finally, performing statistical analysis on performance evaluation results of the high-level automatic driving system in all driving scenes and all time periods, so as to comprehensively evaluate the overall performance of the high-level automatic driving system of the test vehicle.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of testing wheeled or tracked vehicles, and in particular to a high-level automatic driving system performance testing and evaluation method and system. Background Art

[0002] The testing of low-level autonomous driving systems usually targets a single function. For example, the AEB (automatic emergency braking) test is to place an obstacle in front of the vehicle being tested to test whether the vehicle will automatically trigger the brakes when it is about to collide. During this process, the signals sent by the vehicle controller, the vehicle's speed, braking distance and other data are recorded. These data are evaluated to determine whether the system's AEB function meets the design requirements.

[0003] High-level autonomous driving systems are usually operated based on driving tasks, such as a smart taxi completing the passenger pick-up and drop-off as a driving task. During its operation, several functions need to work synchronously at the same time, so the test verification of a single function cannot achieve the evaluation of its driving safety. It is necessary to test the vehicle based on actual driving scenarios, such as starting and stopping at traffic lights at intersections, active lane change overtaking, and pedestrian avoidance on sidewalks. The test process of a high-level autonomous driving system is the process of completing a driving task, and several of the aforementioned driving scenarios will appear randomly during this process. During the test process, it is necessary to monitor and evaluate the data of the autonomous driving system itself and the data of external test equipment, such as vehicle speed, acceleration, distance to obstacles and other parameters. However, there is currently no test and evaluation technology for high-level autonomous driving systems based on actual driving scenarios. Summary of the invention

[0004] In view of the shortcomings of the existing technology, the present invention proposes a high-level autonomous driving system performance test and evaluation method and system, which can efficiently and comprehensively evaluate the performance of the high-level autonomous driving system. The specific technical solution is as follows:

[0005] In a first aspect, a high-level autonomous driving system performance test and evaluation method is provided. In a first implementable manner of the first aspect, the method includes:

[0006] Set the distance boundary corresponding to the test vehicle, and slice the acquired test data according to the distance boundary to extract the data segment of interest;

[0007] Performing scene recognition on each of the extracted data segments of interest to determine a scene label corresponding to each of the data segments of interest;

[0008] Performing performance evaluation on each data segment of interest according to an evaluation method corresponding to the scene label, and determining an evaluation result corresponding to each data segment of interest;

[0009] The overall performance evaluation result of the test vehicle is determined by performing statistical analysis on the evaluation results of all the data segments of interest.

[0010] In combination with the first implementable manner of the first aspect, in a second implementable manner of the first aspect, setting a distance boundary corresponding to the test vehicle includes:

[0011] Obtain the longitudinal speed, longitudinal acceleration, lateral speed, and lateral acceleration corresponding to the current test vehicle and the preceding vehicle;

[0012] The longitudinal distance from the boundary is calculated according to the longitudinal speeds and longitudinal accelerations corresponding to the test vehicle and the preceding vehicle, and the lateral distance from the boundary is calculated according to the lateral speeds and lateral accelerations corresponding to the test vehicle and the preceding vehicle.

[0013] In combination with the first implementable manner of the first aspect, in a third implementable manner of the first aspect, performing scene recognition on the extracted data segment of interest includes:

[0014] Performing data analysis on the data of interest to respectively determine the discrimination parameter values ​​corresponding to each stage of the data segment of interest;

[0015] The discrimination conditions corresponding to each scene are traversed in sequence until the discrimination parameter values ​​corresponding to each stage of the data segment of interest meet the discrimination conditions corresponding to the scene, and the scene label corresponding to the scene is marked on the data segment of interest.

[0016] In combination with the third implementable method of the first aspect, in the fourth implementable method of the first aspect, if the discrimination parameter values ​​corresponding to each stage of the data segment of interest do not meet the discrimination conditions corresponding to all scenes, then the discrimination conditions corresponding to the new scene are generated based on the discrimination parameter values ​​of each stage of the data segment of interest.

[0017] In combination with the first implementable manner of the first aspect, in a fifth implementable manner of the first aspect, performing performance evaluation on the data segment of interest according to an evaluation method corresponding to the scene label includes:

[0018] Based on the corresponding operating parameter values ​​in the data segment of interest, an evaluation algorithm corresponding to the scene label is used to calculate the evaluation criteria corresponding to each operating parameter;

[0019] According to the evaluation criteria corresponding to each operating parameter and in combination with the weight coefficient of each operating parameter corresponding to the scene label, the evaluation result corresponding to the data segment of interest is calculated.

[0020] In combination with the first implementable manner of the first aspect, in a sixth implementable manner of the first aspect, statistical analysis is performed on the evaluation results of all the data segments of interest, including:

[0021] Performing statistical analysis on the evaluation results corresponding to all the data segments of interest to determine the scene evaluation results corresponding to each scene label included in the test data;

[0022] The overall performance evaluation result of the test vehicle is calculated based on the scene evaluation results corresponding to all the scene labels and the scene weight coefficients corresponding to the scene labels.

[0023] In a second aspect, a high-level autonomous driving system performance test and evaluation system is provided. In a first achievable manner of the second aspect, the system includes:

[0024] A data extraction module is configured to set a distance boundary corresponding to the test vehicle, and slice the acquired test data according to the distance boundary to extract the data segment of interest;

[0025] A scene label recognition module, configured to perform scene recognition on each of the extracted data segments of interest, and determine a scene label corresponding to each of the data segments of interest;

[0026] A data segment evaluation module, configured to perform performance evaluation on each data segment of interest according to an evaluation method corresponding to the scene label, and determine an evaluation result corresponding to each data segment of interest;

[0027] The overall performance analysis module is configured to determine the overall performance evaluation result of the test vehicle by performing statistical analysis on the evaluation results of all the data segments of interest.

[0028] In combination with the first implementable manner of the second aspect, in a second implementable manner of the second aspect, the data extraction module includes:

[0029] A data acquisition unit configured to acquire longitudinal speed, longitudinal acceleration and lateral speed, lateral acceleration corresponding to the current test vehicle and the preceding vehicle;

[0030] The boundary setting unit is configured to calculate the longitudinal distance from the boundary according to the longitudinal speeds and longitudinal accelerations corresponding to the test vehicle and the preceding vehicle, and to calculate the lateral distance from the boundary according to the lateral speeds and lateral accelerations corresponding to the test vehicle and the preceding vehicle.

[0031] In combination with the first implementable manner of the second aspect, in a third implementable manner of the second aspect, the scene label recognition module includes:

[0032] A parameter analysis unit, configured to perform data analysis on the data of interest, and respectively determine the discrimination parameter values ​​corresponding to each stage of the data segment of interest;

[0033] The label discrimination unit is configured to sequentially traverse the discrimination conditions corresponding to each scene until the discrimination parameter values ​​corresponding to each stage of the data segment of interest meet the discrimination conditions corresponding to the scene, and mark the scene label corresponding to the scene for the data segment of interest.

[0034] In combination with the first implementable manner of the second aspect, in a fourth implementable manner of the second aspect, the data segment evaluation module includes:

[0035] A calculation unit configured to calculate, based on the corresponding operating parameter values ​​in the data segment of interest, an evaluation algorithm corresponding to the scene label to obtain an evaluation standard corresponding to each operating parameter;

[0036] The evaluation unit is configured to calculate the evaluation result corresponding to the data segment of interest according to the evaluation criteria corresponding to each operating parameter and the weight coefficient of each operating parameter corresponding to the scene label.

[0037] Beneficial effects: By adopting the high-level automatic driving system performance test and evaluation method and system of the present invention, all the data segments of interest in the test data of the test vehicle can be extracted through the set distance boundary, thereby reducing the amount of data analysis and improving the efficiency of performance test evaluation. By performing scene recognition on all the extracted data segments of interest, the driving scene to which each data segment of interest belongs can be identified, thereby determining the data segments corresponding to different driving scenes. Based on the operating parameters included in the data segments, the performance of the high-level automatic driving system in different driving scenarios can be evaluated by using the corresponding evaluation method, thereby determining the performance of the high-level automatic driving system in different driving scenarios. By statistically analyzing the performance evaluation results of the high-level automatic driving system in different driving scenarios, the purpose of comprehensively evaluating the performance of the high-level automatic driving system can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the specific implementation of the present invention, the following will briefly introduce the drawings required for use in the specific implementation. In all the drawings, each element or part is not necessarily drawn according to the actual scale.

[0039] Figure 1 A flowchart of a high-level autonomous driving system performance test and evaluation method provided by an embodiment of the present invention;

[0040] Figure 2 A system block diagram of a high-level autonomous driving system performance test and evaluation system provided by an embodiment of the present invention;

[0041] Figure 3 A schematic diagram of a distance boundary provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following embodiments of the technical solution of the present invention are described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and cannot be used to limit the protection scope of the present invention.

[0043] like Figure 1 The flowchart of the high-level automated driving system performance test and evaluation method shown in FIG. 1 includes:

[0044] Step 1: Set the distance boundary corresponding to the test vehicle, and slice the acquired test data according to the distance boundary to extract the data segment of interest;

[0045] Step 2: Perform scene recognition on each of the extracted data segments of interest to determine the scene label corresponding to each of the data segments of interest;

[0046] Step 3: Perform performance evaluation on each data segment of interest according to the evaluation method corresponding to the scene label, and determine the evaluation result corresponding to each data segment of interest;

[0047] Step 4: Determine the overall performance evaluation result of the test vehicle by performing statistical analysis on the evaluation results of all the data segments of interest.

[0048] Specifically, first, the distance boundary can be set as the target based on the safe distance between the surrounding vehicles and the test vehicle. When the surrounding vehicle enters the distance boundary range, the test data is sliced ​​until the surrounding vehicle leaves the distance boundary range, and the slicing ends, thereby extracting the interesting data segments of the test vehicle for subsequent automatic driving system performance evaluation. Then, the scene recognition can be performed on each extracted interesting data segment according to the operating parameters, the driving scene corresponding to each interesting data segment can be determined, and each interesting data segment can be marked with a corresponding scene label.

[0049] Afterwards, the performance of the high-level autonomous driving system of the test vehicle in the time period corresponding to each data segment of interest can be evaluated according to the various operating parameters in the data segment of interest through the evaluation method corresponding to the scene label, and the performance evaluation results of the high-level autonomous driving system in different driving scenarios and different time periods can be obtained. Finally, the performance evaluation results of the high-level autonomous driving system in all driving scenarios and all time periods are statistically analyzed to comprehensively evaluate the overall performance of the high-level autonomous driving system of the test vehicle.

[0050] It should be understood that since the performance test and evaluation of high-level autonomous driving systems is mainly used to evaluate the safety of vehicles, the test data when the test vehicle is in a safe state is not very useful for the performance test and evaluation of high-level autonomous driving systems. To this end, the test data of the test vehicle in a dangerous state can be sliced ​​for subsequent system evaluation tests by setting a distance boundary, thereby reducing the amount of data analysis and improving the efficiency of test evaluation.

[0051] In this embodiment, optionally, in step 1, setting a distance boundary corresponding to the test vehicle includes:

[0052] Obtain the longitudinal speed, longitudinal acceleration, lateral speed, and lateral acceleration corresponding to the current test vehicle and the preceding vehicle;

[0053] The longitudinal distance from the boundary is calculated according to the longitudinal speeds and longitudinal accelerations corresponding to the test vehicle and the preceding vehicle, and the lateral distance from the boundary is calculated according to the lateral speeds and lateral accelerations corresponding to the test vehicle and the preceding vehicle.

[0054] Specifically, if Figure 3 As shown, the lateral distance and longitudinal distance of the test vehicle from the boundary can be calculated based on the longitudinal speed, longitudinal acceleration and lateral speed, lateral acceleration corresponding to the test vehicle and the preceding vehicle in the acquired test data.

[0055] The specific calculation formula of the longitudinal distance is as follows:

[0056]

[0057] Among them, v1 and a1 are the longitudinal velocity and longitudinal acceleration of the test vehicle, respectively, and v2 and a2 are the longitudinal velocity and longitudinal acceleration of the preceding vehicle, respectively.

[0058] The specific calculation formula of the horizontal distance is as follows:

[0059]

[0060] Among them, v1′ and a1′ are the lateral velocity and lateral acceleration of the test vehicle, respectively; v2′ and a2′ are the lateral velocity and lateral acceleration of the preceding vehicle, respectively.

[0061] In this embodiment, optionally, in step 2, scene recognition is performed on the extracted data segment of interest, including:

[0062] Performing data analysis on the data of interest to respectively determine the discrimination parameter values ​​corresponding to each stage of the data segment of interest;

[0063] The discrimination conditions corresponding to each scene are traversed in sequence until the discrimination parameter values ​​corresponding to each stage of the data segment of interest meet the discrimination conditions corresponding to the scene, and the scene label corresponding to the scene is marked on the data segment of interest.

[0064] Specifically, in this embodiment, the longitudinal distance, lateral distance and relative speed of the vehicle can be selected as the discrimination parameters for the recognition of the driving scene. The driving scene can also be divided into three stages: the starting state, the intermediate state and the ending state, and the discrimination conditions of different driving scenes are pre-set according to the three stages. The discrimination conditions include conditional thresholds of different discrimination parameters corresponding to the three stages of the vehicle in the driving scene.

[0065] For example, the following table shows the judgment conditions corresponding to two driving scenarios: ACC and Cut-in:

[0066]

[0067]

[0068] When performing scene recognition on a data segment of interest, the discriminant parameters in the data segment of interest can be analyzed to determine the discriminant parameter values ​​corresponding to the data segment of interest in these three stages, and the discriminant parameter values ​​corresponding to the three stages are matched with the condition thresholds of each stage corresponding to the driving scene in turn. If the discriminant parameter values ​​corresponding to the three stages all meet the discriminant conditions, the data segment of interest belongs to the driving scene. On the contrary, as long as the condition threshold of any stage is not met, it is necessary to match the discriminant condition corresponding to the next driving scene until the discriminant parameter values ​​corresponding to the three stages all meet the discriminant conditions.

[0069] In this embodiment, optionally, if the discrimination parameter values ​​corresponding to each stage of the data segment of interest do not satisfy the discrimination conditions corresponding to all scenes, a discrimination condition corresponding to a new scene is generated based on the discrimination parameter values ​​of each stage of the data segment of interest.

[0070] Specifically, if after traversing all the discrimination conditions corresponding to the stored driving scenarios, the data segment of interest still does not match the corresponding driving scenario. At this time, the discrimination parameter values ​​corresponding to the data segment of interest that does not match the driving scenario can be used as new discrimination conditions to generate a new driving scenario. In this way, the number of driving scenarios can be expanded to comprehensively evaluate the performance of a high-level autonomous driving system.

[0071] In this embodiment, optionally, in step 3, the performance of the data segment of interest is evaluated according to an evaluation method corresponding to the scene label, including:

[0072] Based on the corresponding operating parameter values ​​in the data segment of interest, an evaluation algorithm corresponding to the scene label is used to calculate the evaluation criteria corresponding to each operating parameter;

[0073] According to the evaluation criteria corresponding to each operating parameter and in combination with the weight coefficient of each operating parameter corresponding to the scene label, the evaluation result corresponding to the data segment of interest is calculated.

[0074] Specifically, first, a corresponding evaluation method may be selected according to a scene label corresponding to a data segment of interest, and evaluation criteria corresponding to different operating parameters may be calculated according to values ​​of various operating parameters in the data segment of interest.

[0075] For example, the average relative speed of the test vehicle in the corresponding time period can be calculated based on the relative speed in the data segment of interest, and then the evaluation result corresponding to the relative speed can be calculated based on the average relative speed. The specific calculation formula is as follows:

[0076]

[0077] Among them, V rel is the average relative speed of the test vehicle, V ref is the set average relative vehicle speed target value, and d1 is the corresponding evaluation result of the average vehicle speed target parameter.

[0078] The average relative distance between the test vehicle and other vehicles in the corresponding time period can also be calculated based on the relative distance in the data segment of interest, and then the evaluation result corresponding to the relative distance can be calculated based on the average relative distance. The specific calculation formula is as follows:

[0079]

[0080] Among them, D rel is the average relative distance of the test vehicles, D ref is the set average relative distance target value. d2 is the corresponding evaluation result of the target parameter of average relative distance.

[0081] After the evaluation criteria corresponding to different operating parameters are calculated, the evaluation criteria corresponding to all operating parameters can be integrated according to the weight coefficients corresponding to the different operating parameters to obtain the performance evaluation result corresponding to the data segment of interest. The specific calculation formula is as follows:

[0082]

[0083] Among them, d i ,ω i are the evaluation criteria and weight coefficients corresponding to the i-th operating parameter, and N is the number of operating parameters involved in the data segment of interest.

[0084] It should be understood that different driving scenarios correspond to different evaluation methods, different operating parameters are involved in the evaluation of the data segments of interest, and different algorithms for the evaluation criteria.

[0085] For example, another driving scenario may evaluate the performance criteria of the data segment of interest based on the maximum vehicle deceleration and the minimum vehicle spacing.

[0086] The performance standard based on the maximum deceleration of the vehicle is calculated as follows:

[0087]

[0088] V r ' ef =V1-V2

[0089] Among them, V r ' el To measure the maximum deceleration of the vehicle, V r ' ef is the maximum deceleration target value. V1 and V2 are the speeds of the test vehicle and the preceding vehicle in the intermediate state, respectively. d3 is the corresponding evaluation result of the maximum deceleration target parameter.

[0090] The calculation formula based on the performance standard of minimum vehicle spacing is as follows:

[0091]

[0092] Among them, D r ' el is the minimum distance between test vehicles, D r ' ef is the set target value of the minimum vehicle spacing. d4 is the corresponding evaluation result of the target parameter of the minimum spacing.

[0093] In this embodiment, optionally, in step 4, statistical analysis is performed on the evaluation results of all the data segments of interest, including:

[0094] Performing statistical analysis on the evaluation results corresponding to all the data segments of interest to determine the scene evaluation results corresponding to each scene label included in the test data;

[0095] The overall performance evaluation result of the test vehicle is calculated based on the scene evaluation results corresponding to all the scene labels and the scene weight coefficients corresponding to the scene labels.

[0096] Specifically, after obtaining the corresponding evaluation results corresponding to all data segments of interest, the evaluation results of each data segment of interest can be classified and counted according to the scene label to determine the scene evaluation result corresponding to each scene label. The specific calculation formula is as follows:

[0097]

[0098] in, is the scene evaluation result corresponding to the l-th scene, M is the number of interesting data segments corresponding to the l-th scene, d p is the evaluation result corresponding to the pth data segment of interest.

[0099] After calculating the scenario evaluation results corresponding to each driving scenario, the overall performance evaluation results of the high-level autonomous driving system of the test vehicle can be calculated by combining the scenario weight coefficients corresponding to each driving scenario. The specific calculation formula is as follows:

[0100]

[0101] Where L is the number of driving scenarios, α l is the scene weight coefficient corresponding to the lth scene.

[0102] In this way, the performance of the high-level autonomous driving system can be comprehensively evaluated based on the performance evaluation results of all driving scenarios involved in the test vehicle during the test process, thereby achieving the purpose of comprehensively evaluating the performance of the high-level autonomous driving system.

[0103] like Figure 2 The system block diagram of the high-level autonomous driving system performance test and evaluation system shown in the figure includes:

[0104] A data extraction module is configured to set a distance boundary corresponding to the test vehicle, and slice the acquired test data according to the distance boundary to extract the data segment of interest;

[0105] A scene label recognition module, configured to perform scene recognition on each of the extracted data segments of interest, and determine a scene label corresponding to each of the data segments of interest;

[0106] A data segment evaluation module, configured to perform performance evaluation on each data segment of interest according to an evaluation method corresponding to the scene label, and determine an evaluation result corresponding to each data segment of interest;

[0107] The overall performance analysis module is configured to determine the overall performance evaluation result of the test vehicle by performing statistical analysis on the evaluation results of all the data segments of interest.

[0108] Specifically, the evaluation system consists of a data extraction module, a scene label recognition module, a data segment evaluation module, and an overall performance analysis module. Among them, the data extraction module can set a distance boundary for the target based on the safe distance between the surrounding vehicles and the test vehicle, and extract the data segments of interest from the test vehicle based on the distance boundary for subsequent automatic driving system performance evaluation. The scene label recognition module can perform scene recognition on each extracted data segment of interest based on the operating parameters, determine the driving scene corresponding to each data segment of interest, and mark each data of interest with a corresponding scene label.

[0109] The data segment evaluation module can evaluate the performance of the high-level autonomous driving system of the test vehicle in the time period corresponding to each data segment of interest according to the various operating parameters in the data segment of interest, through the evaluation method corresponding to the scene label, and obtain the performance evaluation results of the high-level autonomous driving system in different driving scenarios and different time periods. The overall performance analysis module can perform statistical analysis on the performance evaluation results of the high-level autonomous driving system in all driving scenarios and all time periods, and comprehensively evaluate the overall performance of the high-level autonomous driving system of the test vehicle.

[0110] In this embodiment, optionally, the data extraction module includes:

[0111] A data acquisition unit configured to acquire longitudinal speed, longitudinal acceleration and lateral speed, lateral acceleration corresponding to the current test vehicle and the preceding vehicle;

[0112] The boundary setting unit is configured to calculate the longitudinal distance from the boundary according to the longitudinal speeds and longitudinal accelerations corresponding to the test vehicle and the preceding vehicle, and to calculate the lateral distance from the boundary according to the lateral speeds and lateral accelerations corresponding to the test vehicle and the preceding vehicle.

[0113] Specifically, the data extraction module includes a data acquisition unit and a boundary setting unit. The data acquisition unit can extract the longitudinal speed, longitudinal acceleration, lateral speed, and lateral acceleration of the test vehicle and the preceding vehicle from the acquired test data. The boundary setting unit can calculate the lateral distance and longitudinal distance of the test vehicle from the boundary using the above algorithm based on the extracted relevant parameter data.

[0114] In this embodiment, optionally, the scene label recognition module includes:

[0115] A parameter analysis unit, configured to perform data analysis on the data of interest, and respectively determine the discrimination parameter values ​​corresponding to each stage of the data segment of interest;

[0116] The label discrimination unit is configured to sequentially traverse the discrimination conditions corresponding to each scene until the discrimination parameter values ​​corresponding to each stage of the data segment of interest meet the discrimination conditions corresponding to the scene, and mark the scene label corresponding to the scene for the data segment of interest.

[0117] Specifically, the scene label recognition module includes a parameter analysis unit and a label discrimination unit. Among them, the parameter analysis unit can perform data analysis on the discrimination parameters in the data segment of interest, and determine the discrimination parameter values ​​corresponding to the starting state, the intermediate state and the terminal state respectively. The label discrimination unit can match the discrimination parameter values ​​corresponding to the three stages with the condition thresholds corresponding to each stage of the driving scene in turn. If the discrimination parameter values ​​corresponding to the three stages all meet the discrimination conditions, the data segment of interest belongs to the driving scene. On the contrary, as long as the condition threshold of any stage is not met, it is necessary to match the discrimination conditions corresponding to the next driving scene until the discrimination parameter values ​​corresponding to the three stages all meet the discrimination conditions.

[0118] In this embodiment, optionally, the data segment evaluation module includes:

[0119] A calculation unit configured to calculate, based on the corresponding operating parameter values ​​in the data segment of interest, an evaluation algorithm corresponding to the scene label to obtain an evaluation standard corresponding to each operating parameter;

[0120] The evaluation unit is configured to calculate the evaluation result corresponding to the data segment of interest according to the evaluation criteria corresponding to each operating parameter and the weight coefficient of each operating parameter corresponding to the scene label.

[0121] Specifically, the data segment evaluation module includes a calculation unit and an evaluation unit. The calculation unit can select a corresponding evaluation method according to the scene label corresponding to the data segment of interest, and calculate the evaluation criteria corresponding to different operating parameters according to the values ​​of various operating parameters in the data segment of interest. The evaluation unit can integrate the evaluation criteria corresponding to all operating parameters according to the weight coefficients corresponding to different operating parameters to obtain the performance evaluation result corresponding to the data segment of interest.

[0122] After calculating the performance evaluation results corresponding to each data segment of interest, the overall performance analysis module can classify and count the evaluation results of each data segment of interest according to the scene label, and determine the scene evaluation results corresponding to each scene label. Then, the overall performance evaluation results of the high-level autonomous driving system of the test vehicle are calculated in combination with the scene weight coefficient corresponding to each driving scene, so as to achieve the purpose of comprehensively evaluating the performance of the high-level autonomous driving system.

[0123] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A high-level autonomous driving system performance test and evaluation method, It is characterized in that include: Set the distance boundary corresponding to the test vehicle, and slice the acquired test data according to the distance boundary to extract the data segment of interest; Performing scene recognition on each of the extracted data segments of interest to determine a scene label corresponding to each of the data segments of interest; Performing performance evaluation on each data segment of interest according to an evaluation method corresponding to the scene label, and determining an evaluation result corresponding to each data segment of interest; The overall performance evaluation result of the test vehicle is determined by performing statistical analysis on the evaluation results of all the data segments of interest.

2. The high-level autonomous driving system performance test and evaluation method according to claim 1, It is characterized in that Set the distance boundary corresponding to the test vehicle, including: Obtain the longitudinal speed, longitudinal acceleration, lateral speed, and lateral acceleration corresponding to the current test vehicle and the preceding vehicle; The longitudinal distance from the boundary is calculated according to the longitudinal speeds and longitudinal accelerations corresponding to the test vehicle and the preceding vehicle, and the lateral distance from the boundary is calculated according to the lateral speeds and lateral accelerations corresponding to the test vehicle and the preceding vehicle.

3. The high-level autonomous driving system performance test and evaluation method according to claim 1, It is characterized in that Performing scene recognition on the extracted data segment of interest, including: Performing data analysis on the data of interest to respectively determine the discrimination parameter values ​​corresponding to each stage of the data segment of interest; The discrimination conditions corresponding to each scene are traversed in sequence until the discrimination parameter values ​​corresponding to each stage of the data segment of interest meet the discrimination conditions corresponding to the scene, and the scene label corresponding to the scene is marked on the data segment of interest.

4. The high-level autonomous driving system performance test and evaluation method according to claim 3, It is characterized in that If the discrimination parameter values ​​corresponding to each stage of the data segment of interest do not satisfy the discrimination conditions corresponding to all scenes, the discrimination conditions corresponding to the new scene are generated based on the discrimination parameter values ​​of each stage of the data segment of interest.

5. The high-level autonomous driving system performance test and evaluation method according to claim 1, It is characterized in that The performance of the data segments of interest is evaluated according to the evaluation method corresponding to the scene label, including: Based on the corresponding operating parameter values ​​in the data segment of interest, an evaluation algorithm corresponding to the scene label is used to calculate the evaluation criteria corresponding to each operating parameter; According to the evaluation criteria corresponding to each operating parameter and in combination with the weight coefficient of each operating parameter corresponding to the scene label, the evaluation result corresponding to the data segment of interest is calculated.

6. The high-level autonomous driving system performance test and evaluation method according to claim 1, It is characterized in that Perform statistical analysis on the evaluation results of all the data segments of interest, including: Performing statistical analysis on the evaluation results corresponding to all the data segments of interest to determine the scene evaluation results corresponding to each scene label included in the test data; The overall performance evaluation result of the test vehicle is calculated based on the scene evaluation results corresponding to all the scene labels and the scene weight coefficients corresponding to the scene labels.

7. A high-level autonomous driving system performance test and evaluation system, It is characterized in that include: A data extraction module is configured to set a distance boundary corresponding to the test vehicle, and slice the acquired test data according to the distance boundary to extract the data segment of interest; A scene label recognition module, configured to perform scene recognition on each of the extracted data segments of interest, and determine a scene label corresponding to each of the data segments of interest; A data segment evaluation module, configured to perform performance evaluation on each data segment of interest according to an evaluation method corresponding to the scene label, and determine an evaluation result corresponding to each data segment of interest; The overall performance analysis module is configured to determine the overall performance evaluation result of the test vehicle by performing statistical analysis on the evaluation results of all the data segments of interest.

8. The high-level autonomous driving system performance test and evaluation system according to claim 7, It is characterized in that The data extraction module comprises: A data acquisition unit configured to acquire longitudinal speed, longitudinal acceleration and lateral speed, lateral acceleration corresponding to the current test vehicle and the preceding vehicle; The boundary setting unit is configured to calculate the longitudinal distance from the boundary according to the longitudinal speeds and longitudinal accelerations corresponding to the test vehicle and the preceding vehicle, and to calculate the lateral distance from the boundary according to the lateral speeds and lateral accelerations corresponding to the test vehicle and the preceding vehicle.

9. The high-level autonomous driving system performance test and evaluation system according to claim 7, It is characterized in that The scene label recognition module includes: A parameter analysis unit configured to perform data analysis on the data of interest and respectively determine the discrimination parameter values ​​corresponding to each stage of the data segment of interest; The label discrimination unit is configured to sequentially traverse the discrimination conditions corresponding to each scene until the discrimination parameter values ​​corresponding to each stage of the data segment of interest meet the discrimination conditions corresponding to the scene, and mark the scene label corresponding to the scene for the data segment of interest.

10. The high-level autonomous driving system performance test and evaluation method according to claim 7, It is characterized in that The data segment evaluation module includes: A calculation unit configured to calculate, based on the corresponding operating parameter values ​​in the data segment of interest, an evaluation algorithm corresponding to the scene label to obtain an evaluation standard corresponding to each operating parameter; The evaluation unit is configured to calculate the evaluation result corresponding to the data segment of interest according to the evaluation criteria corresponding to each operating parameter and the weight coefficient of each operating parameter corresponding to the scene label.