CN116026361A

CN116026361A - Function positioning method, device, storage medium and computer equipment

Info

Publication number: CN116026361A
Application number: CN202211565697.8A
Authority: CN
Inventors: 祝啸风
Original assignee: Guangzhou Weride Technology Co Ltd
Current assignee: Guangzhou Weride Technology Co Ltd
Priority date: 2022-12-07
Filing date: 2022-12-07
Publication date: 2023-04-28

Abstract

The application provides a function positioning method, a device, a storage medium and computer equipment, wherein the method comprises the following steps: determining each target difference scene needing to be subjected to functional positioning, and performing original difference expression of a reference driving algorithm and a driving algorithm to be compared under each target difference scene; acquiring preset functions to be positioned; for each function to be positioned, only closing the function to be positioned of the driving algorithm to be compared, and respectively adopting each target difference scene to carry out simulation test on the driving algorithm to be compared with the closed function so as to obtain automatic driving data of the function to be positioned under each target difference scene; aiming at each target difference scene, according to the automatic driving data of each function to be positioned in the target difference scene, respectively determining the test difference performance corresponding to each function to be positioned, and determining the target function from each function to be positioned based on each test difference performance and the original difference performance corresponding to the target difference scene.

Description

Function positioning method, device, storage medium and computer equipment

Technical Field

The present disclosure relates to the field of algorithm testing, and in particular, to a functional positioning method, apparatus, storage medium, and computer device.

Background

With continuous iteration of the autopilot algorithm, the autopilot algorithm has more and more functions. In order to evaluate the advantages and disadvantages of the new version and the old version of the automatic driving algorithm, the full road section data can be adopted to respectively carry out simulation test on the new version and the old version of the automatic driving algorithm, and a plurality of difference scenes are extracted from the full road section data according to the track planning differences of the different versions of the algorithms in the simulation test. After each difference scene is obtained, a tester adopts a manual mode to respectively determine which function in the new version of the automatic driving algorithm causes each difference scene, namely, each difference scene is respectively subjected to function positioning so as to facilitate subsequent algorithm debugging, evaluation and iteration. The prior art relies entirely on manual work to perform functional positioning, and therefore has the problem of inefficiency.

Disclosure of Invention

The object of the present application is to solve at least one of the above-mentioned technical drawbacks, in particular the technical drawbacks of the prior art, such as the inefficiency of the test.

In a first aspect, an embodiment of the present application provides a function positioning method, where the method includes:

determining each target difference scene needing to be subjected to functional positioning, and performing original difference expression of a reference driving algorithm and a driving algorithm to be compared under each target difference scene;

Acquiring preset functions to be positioned;

for each function to be positioned, only closing the function to be positioned of the driving algorithm to be compared, and respectively adopting each target difference scene to carry out simulation test on the driving algorithm to be compared with the closed function so as to obtain automatic driving data of the function to be positioned in each target difference scene;

for each target difference scene, according to automatic driving data of each function to be positioned in the target difference scene, test difference performance corresponding to each function to be positioned is respectively determined, and based on each test difference performance and original difference performance corresponding to the target difference scene, a target function is determined from each function to be positioned, wherein the target function is a function which leads to the original difference performance of the target difference scene.

In one embodiment, the step of determining each target difference scene requiring functional positioning, and the original difference performance of the reference driving algorithm and the driving algorithm to be compared under each target difference scene includes:

acquiring each original difference scene, wherein the original difference scene is a scene in which the planning track difference of the reference driving algorithm and the driving algorithm to be compared meets a preset rule;

Acquiring preset difference evaluation indexes;

for each difference evaluation index, determining initial difference performance of each original difference scene under the difference evaluation index, which is used for reflecting the difference degree between the reference driving algorithm and the driving algorithm to be compared, and screening out an original difference scene with large difference degree from the original difference scenes as a target difference scene according to each initial difference performance corresponding to the difference evaluation index, and taking the initial difference performance corresponding to each target difference scene as the original difference performance corresponding to the target difference scene.

In one embodiment, each of the difference evaluation indexes includes an error reporting difference index, and each of the target difference scenes includes a first target difference scene;

the step of respectively determining initial difference performance of each original difference scene under the difference evaluation index for reflecting the difference degree between the reference driving algorithm and the driving algorithm to be compared, and screening out the original difference scene with large difference degree from the original difference scenes as a target difference scene according to each initial difference performance corresponding to the difference evaluation index comprises the following steps:

Respectively acquiring first error reporting information of the reference driving algorithm in each original difference scene;

respectively acquiring second error reporting information of the driving algorithm to be compared in each original difference scene;

according to the first error reporting information and the second error reporting information, respectively determining initial difference performance of each original difference scene under the error reporting difference index, wherein the initial difference performance is used for reflecting whether error reporting difference exists between the reference driving algorithm and the driving algorithm to be compared under the corresponding original difference scene;

and selecting each original difference scene with error reporting difference as the first target difference scene according to the initial difference performance of each original difference scene under error reporting difference indexes.

In one embodiment, the step of determining the target function from the functions to be located based on the test difference performances and the original difference performances corresponding to the target difference scene includes:

if the target difference scene is the first target difference scene, taking the test difference performance which is the same as the original difference performance corresponding to the first target difference scene in the test difference performance as a target test difference performance, and determining the target function according to the function to be positioned corresponding to each target test difference performance.

In one embodiment, each of the difference evaluation indicators includes a simulation score difference indicator, and each of the target difference scenes includes a second target difference scene;

respectively obtaining first simulation scores of the reference driving algorithm under each original difference scene;

respectively obtaining second simulation scores of the driving algorithm to be compared in each original difference scene;

for each original difference scene, calculating the difference between a first simulation score corresponding to the original difference scene and a second simulation score corresponding to the original difference scene, and taking the difference as the initial difference expression corresponding to the original difference scene under the simulation score difference index;

and sequencing the initial difference expressions of the original difference scenes under the simulation scoring difference indexes according to the sequence from large to small, and selecting the original difference scenes corresponding to the first N initial difference expressions as the second target difference scene, wherein N is a preset positive integer.

In one embodiment, each of the difference evaluation indexes includes a driving behavior difference index, and each of the target difference scenes includes a third target difference scene;

respectively acquiring a first vehicle state of the reference driving algorithm under each original difference scene;

respectively acquiring a second vehicle state of the driving algorithm to be compared in each original difference scene;

for each original difference scene, generating a difference score for reflecting the driving behavior difference degree of the reference driving algorithm and the driving algorithm to be compared according to a first vehicle state corresponding to the original difference scene and a second vehicle state corresponding to the original difference scene, and taking the difference score as an initial difference representation corresponding to the original difference scene under the driving behavior difference index;

And sorting initial difference expressions of the original difference scenes under the driving behavior difference indexes according to the sequence from big to small, and selecting the original difference scenes corresponding to the first N initial difference expressions as the third target difference scene, wherein N is a preset positive integer.

if the target difference scene is the second target difference scene, determining a screening interval according to the original difference performance corresponding to the second target difference scene, wherein the screening interval comprises the original difference performance;

if the target difference scene is the third target difference scene, determining a screening interval according to the original difference performance corresponding to the third target difference scene, wherein the screening interval comprises the original difference performance;

and screening out target test difference expressions falling into the screening interval from the test difference expressions, and determining the target function according to the function to be positioned corresponding to each target test difference expression.

In a second aspect, embodiments of the present application provide a functional positioning apparatus, the apparatus including:

the scene determining module is used for determining each target difference scene needing to be subjected to function positioning, and the original difference performance of the reference driving algorithm and the driving algorithm to be compared under each target difference scene;

the function acquisition module is used for acquiring each preset function to be positioned;

the simulation module is used for closing only the function to be positioned of the driving algorithm to be compared aiming at each function to be positioned, and performing simulation test on the driving algorithm to be compared with the closed function by adopting each target difference scene respectively so as to obtain automatic driving data of the function to be positioned under each target difference scene;

the positioning module is used for determining test difference performances corresponding to the functions to be positioned according to automatic driving data of the functions to be positioned in the target difference scene aiming at each target difference scene, and determining target functions from the functions to be positioned based on the test difference performances and original difference performances corresponding to the target difference scene, wherein the target functions are functions which cause the original difference performances of the target difference scene.

In a third aspect, embodiments of the present application provide a storage medium having stored therein computer readable instructions, which when executed by one or more processors, cause the one or more processors to perform the steps of the functional positioning method of any of the embodiments described above.

In a fourth aspect, embodiments of the present application provide a computer device, including: one or more processors, and memory;

the memory has stored therein computer readable instructions which, when executed by the one or more processors, perform the steps of the functional localization method of any of the embodiments described above.

In the function positioning, device, storage medium and computer equipment of the application, the computer equipment can determine the target difference scene required to be subjected to function positioning and the original difference performance of two different versions of automatic driving algorithms under each target difference scene, and acquire preset functions to be positioned. For each function to be positioned, the computer equipment can close the function to be positioned of the driving algorithm to be compared, keep the other functions on, and respectively carry out simulation test on the driving algorithm to be compared after the function is closed by adopting each target difference scene so as to acquire automatic driving data of the function to be positioned under each target difference scene. For each target difference scene, the computer device can determine test difference performance of a reference driving algorithm and a driving algorithm to be compared when one to-be-positioned function is closed in each target difference scene according to automatic driving data of each to-be-positioned function in the target difference scene, and locate a function which causes the original difference performance from each to-be-positioned function as a target function according to the original difference performance and each test difference performance corresponding to the target difference scene. Therefore, the computer equipment can automatically position the difference between two different versions of automatic driving algorithms to specific one or more functional characteristics, so that engineers can perform manual verification according to the positioned result, the number of manual screening and positioning functions can be greatly reduced, the consumed human resources can be reduced, and the testing efficiency is improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive faculty for a person skilled in the art.

FIG. 1 is a flow chart of a functional positioning method according to an embodiment;

FIG. 2 is a flowchart illustrating steps for determining a target disparity scene and an original disparity representation in one embodiment;

FIG. 3 is one of the flow diagrams of screening target difference scenes from among original difference scenes in one embodiment;

FIG. 4 is a second flow chart of a method for selecting a target difference scene from among original difference scenes according to an embodiment;

FIG. 5 is a third flow chart of a method for selecting a target difference scene from among original difference scenes according to one embodiment;

FIG. 6 is a block diagram of the functional positioning device in one embodiment;

FIG. 7 is a schematic diagram of a computer device in one embodiment.

Detailed Description

The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.

In one embodiment, the present application provides a functional positioning method. The following embodiments will be described by taking the application of the method to a computer device as an example, and it will be understood that the computing device refers to a device having a data processing function, and may be, but not limited to, a personal notebook, a desktop notebook, a single server or a server cluster, etc. As shown in fig. 1, the method of the present application may include the steps of:

s102: and determining each target difference scene needing to be subjected to functional positioning, and performing original difference expression of a reference driving algorithm and a driving algorithm to be compared under each target difference scene.

The reference driving algorithm and the driving algorithm to be compared can be different versions of automatic driving algorithms. The target difference scene is a simulation difference scene requiring function positioning, and the planned track difference of the reference driving algorithm and the driving algorithm to be compared under the same simulation difference scene meets the preset rule. The preset rule may be any rule used for determining a difference scene in the prior art, which is not particularly limited in this application.

The computer device may determine each target difference scene and determine the original difference performance of the two different versions of the autopilot algorithm, the reference and the to-be-compared driving algorithms, under each target difference scene, respectively. Compared with the reference driving algorithm, the driving algorithm to be compared generally adds one or more functions newly, or iterates one or more functional features, and the original difference performance can reflect the difference degree of the reference driving algorithm and the driving algorithm to be compared with the iterated functions, which are started by the new added functions, in the corresponding target difference scene.

It should be noted that, the specific value of the original difference expression may be determined according to a specific difference evaluation index, where the difference evaluation index may include or not include a track difference index, which is not specifically limited in this application. For example, when the difference evaluation index is an acceleration difference index, the original difference performance refers to acceleration differences of different versions of the autopilot algorithm in the same target difference scene.

S104: and acquiring preset functions to be positioned.

The function to be positioned may be a function that needs to determine whether to cause an original differential performance corresponding to any target differential scene, and may be a new function of the driving algorithm to be compared with the reference driving algorithm or a function in which algorithm iteration occurs. The number of the respective functions to be positioned may be one or more, which is not particularly limited in this application, and may be preset by an engineer.

S106: and aiming at each function to be positioned, only closing the function to be positioned of the driving algorithm to be compared, and respectively adopting each target difference scene to carry out simulation test on the driving algorithm to be compared with the closed function so as to obtain the automatic driving data of the function to be positioned in each target difference scene.

Specifically, for each preset function to be positioned, the computer device may close the function to be positioned of the driving algorithm to be compared, and keep the other functions open, so as to obtain branches of the driving algorithm to be compared. The computer equipment can respectively adopt each target difference scene to carry out simulation test on the branches of the driving algorithm to be compared so as to obtain the automatic driving data of the branches of the driving algorithm to be compared in each target difference scene, namely the automatic driving data corresponding to the function to be positioned, and the automatic driving data can record the behavior state and/or the driving state of the simulated vehicle controlled by the branches of the driving algorithm to be compared in each target difference scene.

For example, the preset functions to be positioned are a function a and a function B, and the target difference scenes are a scene a and a scene B, respectively. Under the condition, the computer equipment can close the function A of the driving algorithm to be compared, keep the function B and the rest functions on, and respectively adopt the scene A and the scene B to carry out simulation test on the driving algorithm to be compared of the closed function A so as to obtain the automatic driving data corresponding to the function A. The computer equipment can also close the function B of the driving algorithm to be compared, keep the function A and the other functions open, and respectively adopt the scene A and the scene B to carry out simulation test on the driving algorithm to be compared of the closed function B so as to obtain automatic driving data corresponding to the function B.

S108: for each target difference scene, according to automatic driving data of each function to be positioned in the target difference scene, test difference performance corresponding to each function to be positioned is respectively determined, and based on each test difference performance and original difference performance corresponding to the target difference scene, a target function is determined from each function to be positioned, wherein the target function is a function which leads to the original difference performance of the target difference scene.

Through S106, the computer may obtain autopilot data for each function to be positioned in each target discrepancy scenario. In other words, for the same target discrepancy scenario, it may correspond to autopilot data for each function to be located. For each target difference scene, the computer device can respectively determine test difference performance corresponding to each function to be positioned according to the automatic driving data of each function to be positioned in the target difference scene. The test difference performance can reflect the difference degree of the driving algorithm to be compared and the reference driving algorithm under the same target difference scene after the corresponding function to be positioned in the driving algorithm to be compared is closed.

For each target difference scene, the computer device locates one or more functions that may cause the original difference performance as target functions in the respective functions to be located according to the test difference performance and the original difference performance corresponding to the target difference scene, in other words, the original difference performance of the target difference scene has a high probability caused by the one or more target functions.

In one embodiment, after determining the target function corresponding to each target difference scene, the computer device may bind each target difference scene with its corresponding target function, and push the target difference scene to the corresponding development engineer according to the corresponding target function, so that the development engineer may check in detail whether the work results in the target difference scene.

In the application, the computer equipment can automatically position the difference between two different versions of automatic driving algorithms to a specific functional feature or a plurality of functional features, so that engineers can perform manual verification according to the positioned result, the number of functions for manual screening and positioning can be greatly reduced, consumed human resources can be reduced, and the testing efficiency is improved.

In one embodiment, as shown in fig. 2, the step of determining the respective target difference scenes required for functional positioning, and the original difference performance of the reference driving algorithm and the driving algorithm to be compared under each target difference scene includes:

s202: and acquiring each original difference scene, wherein the original difference scene is a scene in which the planning track difference of the reference driving algorithm and the driving algorithm to be compared meets a preset rule.

The preset rule involved in this step may be any rule used in the prior art for determining a difference scene, which is not specifically limited in this application.

In one embodiment, the computer device may perform a simulation test on the reference driving algorithm using the complete road segment data to obtain planned trajectory data output by the reference driving algorithm. And the computer equipment can carry out simulation test on the driving algorithm to be compared by adopting the complete road section data so as to obtain planning track data output by the driving algorithm to be compared. The computer equipment can calculate the planned track difference of each simulation frame by frame according to the planned track data output by the reference driving algorithm and the planned track data output by the driving algorithm to be compared, and screen one or more original difference scenes from the complete road section data according to the planned track difference.

S204: and obtaining preset difference evaluation indexes.

The difference evaluation index may be an index for evaluating a behavior state difference and/or a driving behavior difference of the reference driving algorithm and the driving algorithm to be compared under the same simulation test scene. The number of the indexes and the specific indexes of each difference evaluation index can be determined according to the actual situation, and the method is not particularly limited.

S206: for each difference evaluation index, determining initial difference performance of each original difference scene under the difference evaluation index, which is used for reflecting the difference degree between the reference driving algorithm and the driving algorithm to be compared, and screening out an original difference scene with large difference degree from the original difference scenes as a target difference scene according to each initial difference performance corresponding to the difference evaluation index, and taking the initial difference performance corresponding to each target difference scene as the original difference performance corresponding to the target difference scene.

Specifically, since the existence of the trajectory planning difference does not necessarily represent that the driving algorithm to be compared has the disadvantage problem to be solved, after each original difference scene is obtained, the computer device may determine an initial difference performance of each original difference scene under each difference evaluation index, where the initial difference performance may reflect the difference degrees of the reference driving algorithm and the driving algorithm to be compared under the corresponding original difference scene and the corresponding difference evaluation index.

For each difference evaluation index, the computer device can determine the difference degree of the reference driving algorithm and the driving algorithm to be compared under each original difference scene and the difference evaluation index according to each initial difference performance corresponding to the difference evaluation index, screen the original difference scene with large difference degree under the difference evaluation index as a target difference scene corresponding to the difference evaluation index according to the difference degree, and respectively take the initial difference performance corresponding to each target difference scene as the original difference performance of the target difference scene.

Therefore, the original difference scene which can truly reflect the problem can be screened out from a plurality of original difference scenes with planning track differences and used as a target difference scene, and subsequent steps are executed, so that accurate comparison can be realized, subsequent calculated amount is reduced, and the testing efficiency is further improved.

In one embodiment, each of the difference evaluation indicators includes an error reporting difference indicator. That is, among the individual difference evaluation indexes, one of the difference evaluation indexes is a reported error difference index. The error reporting difference index is an index for evaluating the error reporting difference of the system under the same original difference scene of the reference driving algorithm and the driving algorithm to be compared. Each of the target difference scenes includes a first target difference scene, that is, the first target difference scene is included in each of the target difference scenes, and the first target difference scene refers to an original difference scene in which a system error report difference exists.

As shown in fig. 3, the step of determining initial difference expressions of each original difference scene under the difference evaluation index for reflecting the difference degree between the reference driving algorithm and the driving algorithm to be compared, and screening an original difference scene with a large difference degree from the original difference scenes as a target difference scene according to the initial difference expressions corresponding to the difference evaluation index includes:

s302: and respectively acquiring first error reporting information of the reference driving algorithm in each original difference scene.

The first error reporting information of the reference driving algorithm in the original difference scene can be used for reflecting whether the reference driving algorithm sends out a system error reporting in the original difference scene. Further, if there is a problem or road condition that the reference driving algorithm is difficult to safely solve in the simulation process, the reference driving algorithm may send out a system error report.

S304: and respectively acquiring second error reporting information of the driving algorithm to be compared in each original difference scene.

Similar to the first error reporting information, the second error reporting information of the driving algorithm to be compared in the original difference scene can be used for reflecting whether the driving algorithm to be compared sends out a system error report in the original difference scene.

S306: according to the first error reporting information and the second error reporting information, respectively determining initial difference performance of each original difference scene under the error reporting difference indexes, wherein the initial difference performance is used for reflecting whether error reporting differences exist in the reference driving algorithm and the driving algorithm to be compared under the corresponding original difference scenes.

Specifically, for each original difference scene, the computer device may determine, according to the first error reporting information, whether the reference driving algorithm sends out a systematic error report in the original difference scene, and determine, according to the second error reporting information, whether the driving algorithm to be compared sends out a systematic error report in the original difference scene, so that the computer device may determine, according to the first error reporting information and the second error reporting information, whether an error reporting difference exists between the reference driving algorithm and the driving algorithm to be compared in the original difference scene, that is, determine an initial difference performance of the original difference scene under an error reporting difference index. The error reporting difference is a situation that one algorithm in the reference driving algorithm and the driving algorithm to be compared sends out system error reporting for the same original difference scene by the pointer and the other algorithm does not send out system error reporting.

S308: and selecting each original difference scene with error reporting difference as the first target difference scene according to the initial difference performance of each original difference scene under error reporting difference indexes.

Through S306, the computer device may determine whether the error reporting difference exists in the reference driving algorithm and the driving algorithm to be compared in each original difference scene according to the initial difference performance of each original difference scene under the error reporting difference index, and the computer device may declare the original difference scene with the error reporting difference as the first target difference scene.

For example, the original difference scene includes a scene C and a scene D, where the reference driving algorithm and the driving algorithm to be compared have a fault reporting difference, and the reference driving algorithm and the driving algorithm to be compared do not have a fault reporting difference, and the computer device may select the scene C as the first target difference scene.

In this embodiment, by setting error reporting difference indexes in each difference evaluation index and selecting an original difference scene with error reporting difference as a target difference scene, an original difference scene capable of truly reflecting a problem is screened out from a plurality of original difference scenes with planned track differences to serve as a target difference scene, so that accurate comparison can be realized, subsequent calculation amount can be reduced, and testing efficiency can be further improved.

In one embodiment, the step of determining the target function from the functions to be located based on the respective test difference expressions and the original difference expressions corresponding to the target difference scene includes:

It can be appreciated that in S108, if the currently processed target difference scene is the first target difference scene, the computer device may determine, according to the autopilot data of each function to be located in the first target difference scene, the test difference performance corresponding to each function to be located. The test difference performance corresponding to each function to be positioned can be used for reflecting whether error reporting difference exists between the reference driving algorithm and the driving algorithm to be compared which closes the function to be positioned in the first target difference scene.

The computer device may take as the target test differential performance a test differential performance identical to an original differential performance of the first target differential scene, and determine, among the respective to-be-positioned functions, a target function that causes the original differential performance according to the to-be-positioned functions corresponding to the respective target test differential performance.

Further, the original difference performance of the first target difference scene may also record an algorithm identification of an automatic driving algorithm in which a system error occurs and/or an algorithm identification of an automatic driving algorithm in which a system error is not sent. Each test difference performance can also record the algorithm identification of the automatic driving algorithm with the system error and/or the algorithm identification of the automatic driving algorithm without the system error. The computer device may take, as the target test differential performance, a test differential performance in which the error reporting differential exists and the algorithm of the issuing system is consistent with the original differential performance, and determine the target function according to the test differential performance.

Therefore, the target function causing error reporting difference can be accurately positioned from the functions to be positioned, so that the number of functions for manual screening and positioning can be greatly reduced, the consumed manpower resource is further reduced, and the testing efficiency is improved.

In one embodiment, each of the differential evaluation indicators comprises a simulation score differential indicator. That is, among the individual difference evaluation indexes, one of the difference evaluation indexes is a simulation score difference index. The simulation score difference index is an index for evaluating the simulation score difference of the reference driving algorithm and the driving algorithm to be compared under the same original difference scene. In one embodiment, the simulation score difference indicator may be used to evaluate the safety score difference, comfort score difference, and blocking traffic score difference of different versions of the autopilot algorithm in the same original difference scene.

Each of the target difference scenes includes a second target difference scene, that is, a second target difference scene is included in each of the target difference scenes, and the second target difference scene refers to an original difference scene in which the simulation score difference satisfies the scene filtering rule.

As shown in fig. 4, the step of determining initial difference expressions of each original difference scene under the difference evaluation index for reflecting the difference degree between the reference driving algorithm and the driving algorithm to be compared, and screening an original difference scene with a large difference degree from the original difference scenes as a target difference scene according to the initial difference expressions corresponding to the difference evaluation index includes:

s402: and respectively acquiring a first simulation score of the reference driving algorithm under each original difference scene.

Wherein, the first simulation score of the reference driving algorithm in the original difference scene can be used for reflecting the driving performance of the reference driving algorithm in the original difference scene. In one embodiment, the first simulation score may include a first security score, a first comfort score, and a first blocking traffic score.

S404: and respectively obtaining second simulation scores of the driving algorithm to be compared in each original difference scene.

Similar to the first simulation score, the second simulation score of the driving algorithm to be compared in the original difference scene may be used to reflect the driving performance of the driving algorithm to be compared in the original difference scene. In one embodiment, the second simulation score may include a second security score, a second comfort score, and a second blocking traffic score.

S406: and aiming at each original difference scene, calculating the difference between a first simulation score corresponding to the original difference scene and a second simulation score corresponding to the original difference scene, and taking the difference as the initial difference expression corresponding to the original difference scene under the simulation score difference index.

Specifically, for each original difference scene, the computer device may calculate a difference between the two simulation scores according to a first simulation score of the reference driving algorithm under the original difference scene and a second simulation score of the driving algorithm to be compared under the original difference scene, and use the calculated difference as an initial difference performance of the original difference scene corresponding to the simulation score difference index.

Further, if the simulation score difference indicators are used to evaluate the safety score difference, the comfort score difference, and the blocking traffic score difference of the autopilot algorithm of different versions under the same original difference scene, the first simulation score may include a first safety score, a first comfort score, and a first blocking traffic score, and the second simulation score may include a second safety score, a second comfort score, and a second blocking traffic score, then for each original difference scene, the computer device may calculate a first difference between the first safety score and the second safety score corresponding to the original difference scene, a second difference between the first comfort score and the second comfort score corresponding to the original difference scene, and a third difference between the first blocking traffic score and the second blocking traffic score corresponding to the original difference scene, and use the first difference, the second difference, and the third difference as an initial difference representation corresponding to the original difference scene.

S408: and sequencing the initial difference expressions of the original difference scenes under the simulation scoring difference indexes according to the sequence from large to small, and selecting the original difference scenes corresponding to the first N initial difference expressions as the second target difference scene, wherein N is a preset positive integer.

Specifically, for the initial difference performance corresponding to the original difference scene under the simulation scoring difference index, the larger the value of the initial difference performance is, the larger the difference degree is. For the simulation scoring difference indexes, the computer equipment can sort the initial difference expressions corresponding to the simulation scoring difference indexes according to a sorting mode from large to small. And taking the original difference scenes corresponding to the first N (total N) initial difference expressions after sequencing as second target difference scenes.

Further, if the simulation score difference index is used for evaluating the safety score difference, the comfort score difference and the blocking traffic score difference of the automatic driving algorithm of different versions under the same original difference scene, the computer device may sort the first differences corresponding to the original difference scenes according to the order from big to small, and select the original difference scenes corresponding to the first N first differences after sorting as the second target difference scene; sorting the second difference values corresponding to the original difference scenes according to the sequence from big to small, and selecting the original difference scenes corresponding to the first N second difference values after sorting as second target difference scenes; and sorting the third difference values corresponding to the original difference scenes according to the sequence from big to small, and selecting the original difference scenes corresponding to the first N third difference values after sorting as second target difference scenes.

Therefore, N original difference scenes with the largest simulation scoring difference can be used as target difference scenes respectively, and subsequent steps are executed, so that the number of functions of manual screening and positioning can be greatly reduced, the consumed human resources can be further reduced, and the testing efficiency can be improved.

In one embodiment, each of the differential evaluation indexes includes a driving behavior differential index, that is, one of the differential evaluation indexes is a driving behavior differential index. The driving behavior difference evaluation index is used for evaluating the vehicle state difference of the simulated vehicle controlled by the reference driving algorithm and the simulated vehicle controlled by the driving algorithm to be compared. Each target difference scene comprises a third target difference scene, wherein the third target difference scene refers to an original difference scene of which the driving behavior difference meets the scene screening rule.

As shown in fig. 5, the step of determining initial difference expressions of each original difference scene under the difference evaluation index for reflecting the difference degree between the reference driving algorithm and the driving algorithm to be compared, and screening an original difference scene with a large difference degree from the original difference scenes as a target difference scene according to the initial difference expressions corresponding to the difference evaluation index includes:

S502: and respectively acquiring the first vehicle state of the reference driving algorithm under each original difference scene.

In one embodiment, the first vehicle state may include a light state, a driving track, a wheel angle state, whether to enter an automatic parking state, a steering wheel state, a horn blast state, and a brake change state of the simulated vehicle under each of the original differential scenarios controlled by the reference driving algorithm.

S504: and respectively acquiring a second vehicle state of the driving algorithm to be compared in each original difference scene.

In one embodiment, the second vehicle state may include a lighting state, a driving track, a wheel angle state, whether to enter an automatic parking state, a steering wheel state, a horn blast state, and a brake change state of the simulated vehicle under each of the original difference scenes controlled by the driving algorithm to be compared.

S506: and generating a difference score for reflecting the driving behavior difference degree of the reference driving algorithm and the driving algorithm to be compared according to the first vehicle state corresponding to the original difference scene and the second vehicle state corresponding to the original difference scene aiming at each original difference scene, and taking the difference score as the initial difference representation of the original difference scene corresponding to the driving behavior difference index.

Specifically, for each original discrepancy scenario, the computer device may synthesize various vehicle states, generating a discrepancy score that can reflect the discrepancy program of the reference driving algorithm and the driving algorithm to be compared in terms of driving behavior, the discrepancy score being positively correlated with the degree of driving behavior discrepancy. The computer device may use the calculated difference score as an initial difference representation of the original difference scene corresponding to the driving behavior difference index.

S508: and sorting initial difference expressions of the original difference scenes under the driving behavior difference indexes according to the sequence from big to small, and selecting the original difference scenes corresponding to the first N initial difference expressions as the third target difference scene, wherein N is a preset positive integer.

Specifically, for the initial difference performance corresponding to the original difference scene under the driving behavior difference index, the larger the value of the initial difference performance is, the larger the difference degree is. For the driving behavior difference indexes, the computer equipment can rank the initial difference expressions corresponding to the driving behavior difference indexes in a ranking manner from large to small. And taking the original difference scenes corresponding to the first N (total N) initial difference expressions after sequencing as a third target difference scene.

Therefore, N original difference scenes with the largest driving behavior difference can be used as target difference scenes respectively, and subsequent steps are executed, so that the number of functions of manual screening and positioning can be greatly reduced, the consumed human resources can be further reduced, and the testing efficiency can be improved.

It can be appreciated that in S108, if the currently processed target difference scene is the second target difference scene or the third target difference scene, the computer device may determine, according to the autopilot data of each function to be positioned in the currently processed target difference scene, the test difference performance corresponding to each function to be positioned. The test difference performance corresponding to each function to be positioned can be used for reflecting the simulation scoring difference degree of the reference driving algorithm and the driving algorithm to be compared for closing the function to be positioned in the second target difference scene or the driving behavior difference degree in the third target difference scene.

If the currently processed target difference scene is the second target difference scene or the third target difference scene, the computer device may determine the screening interval according to the original difference performance corresponding to the second target difference scene or the third target difference scene. Further, the screening interval may be a continuous interval including the original differential representation.

Aiming at each test difference expression corresponding to the currently processed target difference interval, the computer equipment can take the test difference expression with the value falling into the screening interval as the target test difference expression, and determine the target function in each function to be positioned according to the function to be positioned corresponding to each target test difference expression.

Therefore, the target function which causes the simulation scoring difference and/or the driving behavior difference can be accurately positioned from the functions to be positioned, so that the number of functions for manual screening and positioning can be greatly reduced, the consumed manpower resource is further reduced, and the testing efficiency is improved.

The following describes a function positioning device provided in an embodiment of the present application, and the function positioning device described below and the function positioning method described above may be referred to correspondingly.

In one embodiment, the present application provides a functional positioning apparatus 600. As shown in fig. 6, the apparatus 600 includes a scene determination module 610, a function acquisition module 620, a simulation module 630, and a positioning module 640.

Wherein:

the scene determining module 610 is configured to determine each target difference scene that needs to be functionally positioned, and an original difference performance of the reference driving algorithm and the driving algorithm to be compared under each target difference scene;

a function obtaining module 620, configured to obtain each preset function to be located;

the simulation module 630 is configured to close only the function to be located of the driving algorithm to be compared for each function to be located, and perform a simulation test on the driving algorithm to be compared with the closed function by using each target difference scene, so as to obtain automatic driving data of the function to be located under each target difference scene;

The positioning module 640 is configured to determine, for each target difference scene, according to autopilot data of each function to be positioned in the target difference scene, a test difference performance corresponding to each function to be positioned, and determine, from each function to be positioned, a target function based on each test difference performance and an original difference performance corresponding to the target difference scene, where the target function is a function that causes the original difference performance of the target difference scene.

In one embodiment, the scene determination module 610 includes an original difference scene acquisition unit, a difference evaluation index acquisition unit, and a scene filtering unit. The original difference scene acquisition unit is used for acquiring each original difference scene, wherein the original difference scene is a scene in which the planned track difference of the reference driving algorithm and the driving algorithm to be compared meets a preset rule. The difference evaluation index acquisition unit is used for acquiring each preset difference evaluation index. The scene screening unit is used for respectively determining initial difference performance of each original difference scene under the difference evaluation index for reflecting the difference degree between the reference driving algorithm and the driving algorithm to be compared according to each initial difference performance corresponding to the difference evaluation index, screening the original difference scene with large difference degree from each original difference scene to serve as a target difference scene, and respectively taking the initial difference performance corresponding to each target difference scene as the original difference performance corresponding to the target difference scene.

In one embodiment, each of the difference evaluation indicators includes a error reporting difference indicator, and each of the target difference scenes includes a first target difference scene.

The scene screening unit comprises a first error reporting information acquisition unit, a second error reporting information acquisition unit, an error reporting difference determination unit and a first selection unit. The first error information acquisition unit is used for respectively acquiring first error information of the reference driving algorithm in each original difference scene. The second included information acquisition unit is used for respectively acquiring second error reporting information of the driving algorithm to be compared in each original difference scene. The error reporting difference determining unit is used for respectively determining initial difference performance of each original difference scene under the error reporting difference index according to the first error reporting information and the second error reporting information, and the initial difference performance is used for reflecting whether error reporting difference exists between the reference driving algorithm and the driving algorithm to be compared under the corresponding original difference scenes. The first selection unit is used for selecting each original difference scene with error reporting difference as the first target difference scene according to the initial difference performance of each original difference scene under the error reporting difference index.

In one embodiment, the positioning module 640 includes a first positioning unit. The first positioning unit is configured to, when the target difference scene is the first target difference scene, use, as a target test difference performance, a test difference performance that is the same as an original difference performance corresponding to the first target difference scene in each of the test difference performances, and determine, according to the to-be-positioned function corresponding to each of the target test difference performances, the target function.

In one embodiment, each of the difference evaluation indicators comprises a simulation score difference indicator, and each of the target difference scenes comprises a second target difference scene.

The scene screening unit comprises a first simulation score acquisition unit, a second simulation score acquisition unit, a simulation score difference determination unit and a second selection unit. The first simulation score acquisition unit is used for respectively acquiring first simulation scores of the reference driving algorithm under each original difference scene. The second simulation score acquisition unit is used for respectively acquiring second simulation scores of the driving algorithm to be compared in each original difference scene. The simulation score difference determining unit is used for calculating the difference between a first simulation score corresponding to the original difference scene and a second simulation score corresponding to the original difference scene for each original difference scene, and taking the difference as the initial difference expression corresponding to the original difference scene under the simulation score difference index. The second selection unit is used for sorting initial difference expressions of the original difference scenes under the simulation scoring difference indexes according to the sequence from large to small, and selecting the original difference scenes corresponding to the first N initial difference expressions as the second target difference scene, wherein N is a preset positive integer.

In one embodiment, each of the difference evaluation indexes includes a driving behavior difference index, and each of the target difference scenes includes a third target difference scene.

The scene filtering unit includes a first vehicle state acquisition unit, a second vehicle state acquisition unit, a state difference determination unit, and a third selection unit. The first vehicle state acquisition unit is used for respectively acquiring the first vehicle state of the reference driving algorithm under each original difference scene. The second vehicle state acquisition unit is used for respectively acquiring second vehicle states of the driving algorithm to be compared in each original difference scene. The state difference determining unit is used for generating a difference score for reflecting the driving behavior difference degree of the reference driving algorithm and the driving algorithm to be compared according to the first vehicle state corresponding to the original difference scene and the second vehicle state corresponding to the original difference scene, and taking the difference score as the initial difference representation of the original difference scene corresponding to the driving behavior difference index. The third selection unit is used for respectively sequencing initial difference expressions of the original difference scenes under the driving behavior difference indexes according to the sequence from big to small, and selecting the original difference scenes corresponding to the first N initial difference expressions as the third target difference scene, wherein N is a preset positive integer.

In one embodiment, the positioning module 640 includes a first interval determination unit, a second interval determination unit, and a second positioning unit. The first interval determining unit is configured to determine a screening interval according to an original difference performance corresponding to the second target difference scene if the target difference scene is the second target difference scene, where the screening interval includes the original difference performance. The second interval determining unit is configured to determine a screening interval according to an original difference performance corresponding to the third target difference scene if the target difference scene is the third target difference scene, where the screening interval includes the original difference performance. The second positioning unit is used for screening out target test difference expressions falling into the screening interval from the test difference expressions, and determining the target function according to the function to be positioned corresponding to each target test difference expression.

In one embodiment, the present application further provides a storage medium having stored therein computer readable instructions, which when executed by one or more processors, cause the one or more processors to perform the steps of the functional positioning method of any of the embodiments described above.

In one embodiment, the present application also provides a computer device. The computer device has stored therein computer readable instructions which, when executed by one or more processors, cause the one or more processors to perform the steps of the functional localization method described in any of the embodiments above.

Fig. 7 is a schematic diagram of an internal structure of a computer device according to an embodiment of the present application, and in one example, the computer device may be a server. Referring to FIG. 7, a computer device 900 includes a processing component 902 that further includes one or more processors, and memory resources represented by memory 901, for storing instructions, such as applications, executable by the processing component 902. The application program stored in the memory 901 may include one or more modules each corresponding to a set of instructions. Further, the processing component 902 is configured to execute instructions to perform the steps of the functional localization method according to any of the embodiments described above.

The computer device 900 may also include a power component 903 configured to perform power management of the computer device 900, a wired or wireless network interface 904 configured to connect the computer device 900 to a network, and an input output (I/O) interface 905. The computer device 900 may operate based on an operating system stored in memory 901, such as Windows Server TM, mac OS XTM, unix, linux, free BSDTM, or the like.

It will be appreciated by those skilled in the art that the internal structure of the computer device shown in the present application is merely a block diagram of some of the structures related to the aspects of the present application and does not constitute a limitation of the computer device to which the aspects of the present application apply, and that a particular computer device may include more or less components than those shown in the figures, or may combine some of the components, or have a different arrangement of the components.

Finally, it is further noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element. Herein, "a," "an," "the," and "the" may also include plural forms, unless the context clearly indicates otherwise. Plural means at least two cases such as 2, 3, 5 or 8, etc. "and/or" includes any and all combinations of the associated listed items.

In the present specification, each embodiment is described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and may be combined according to needs, and the same similar parts may be referred to each other.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of functional localization, the method comprising:

acquiring preset functions to be positioned;

2. The method for locating a function according to claim 1, wherein the step of determining each target difference scene for which the function locating is required, and the original difference performance of the reference driving algorithm and the driving algorithm to be compared under each target difference scene, comprises:

acquiring preset difference evaluation indexes;

3. The function positioning method according to claim 2, wherein each of the difference evaluation indexes includes a error reporting difference index, and each of the target difference scenes includes a first target difference scene;

And selecting each original difference scene with error reporting difference as the first target difference scene according to the initial difference performance of each original difference scene under the error reporting difference index.

4. A function positioning method according to claim 3, wherein the step of determining a target function from among the respective functions to be positioned based on the respective test difference expressions and the original difference expressions corresponding to the target difference scene includes:

5. The function positioning method according to claim 2, wherein each of the difference evaluation indexes includes a simulation score difference index, and each of the target difference scenes includes a second target difference scene;

6. The function positioning method according to claim 2, wherein each of the difference evaluation indexes includes a driving behavior difference index, and each of the target difference scenes includes a third target difference scene;

7. The function positioning method according to claim 5 or 6, wherein the step of determining a target function from among the respective functions to be positioned based on the respective test difference expressions and the original difference expressions corresponding to the target difference scene includes:

8. A functional positioning apparatus, the apparatus comprising:

9. A storage medium having stored therein computer readable instructions which, when executed by one or more processors, cause the one or more processors to perform the steps of the functional localization method of any one of claims 1 to 7.

10. A computer device, comprising: one or more processors, and memory;

stored in the memory are computer readable instructions which, when executed by the one or more processors, perform the steps of the functional localization method of any one of claims 1 to 7.