Driving assistance system perception type selection method and device, electronic equipment and driving assistance system

Abstract

The invention relates to the technical field of vehicle control, in particular to a model selection method, device, electronic equipment and system for perception of a driving assistance system, and the method comprises the steps: obtaining the test scores of the perception performance of a plurality of sensors to be tested in one or more real scenes; according to the corresponding relationship between the type of the sensor to be tested and the evaluation weight of the perception performance in each real scene, obtaining the evaluation weight of the perception performance of each sensor to be tested in each real scene; and according to the test score of each real scene and the corresponding evaluation weight, calculating a total test score of the sensing performance of each to-be-tested sensor, and selecting a target sensor for the driving assistance system from each type of to-be-tested sensors based on the total test score. Therefore, the problem of how to accurately and efficiently select a sensor with better sensing performance for the driving assistance system is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle control, and in particular to a selection method, device, electronic equipment and system for perception of a driving assistance system. BACKGROUND

[0002] At present, there are various types of ADAS (Advanced Driver Assistance Systems) on the market, such as adaptive cruise control, automatic parking, lane keeping assistance, automatic emergency braking, etc. Different systems use different sensor combinations and algorithms to meet different driving needs and safety requirements.

[0003] In the related art, the perception performance of ADAS is evaluated by performing a perception performance test on a known target. However, the evaluation of ADAS in the related art is usually simple, cannot quantitatively evaluate the perception performance of ADAS, lacks a unified evaluation standard and weight distribution, and therefore the result is not accurate and objective, making it difficult to make a comprehensive and accurate selection decision. SUMMARY

[0004] The present application provides a selection method, device, electronic equipment and system for perception of a driving assistance system to solve the problem of how to accurately and efficiently select a sensor with better perception performance for a driving assistance system.

[0005] The first aspect of the present application provides a selection method for perception of a driving assistance system, comprising the following steps: obtaining test scores of perception performance of a plurality of to-be-tested sensors in one or more real scenes; obtaining an evaluation weight of the perception performance of each to-be-tested sensor in each real scene according to a corresponding relationship between the type of the to-be-tested sensor and the evaluation weight of the perception performance in each real scene; calculating a test total score of the perception performance of each to-be-tested sensor according to the test score and the corresponding evaluation weight of each real scene; and selecting a target sensor with perception performance for the driving assistance system from each type of to-be-tested sensor based on the test total score.

[0006] Optionally, obtaining the test scores of the perception performance of the plurality of to-be-tested sensors in the one or more real scenes comprises: obtaining perception data and true value data of the plurality of to-be-tested sensors in each real scene; and calculating the test score of each real scene according to the perception data and the true value data.

[0007] Optionally, the calculating the test score of each real scene according to the perception data and the ground truth data comprises: identifying the perception target and the perception lane line in the perception data and the ground truth data; determining respective accuracy indicators of the perception target and the perception lane line according to the perception data and the ground truth data, and obtaining respective scoring standards of the perception target and the perception lane line; and calculating the test score of each real scene according to the respective accuracy indicators and the respective scoring standards.

[0008] Optionally, the respective accuracy indicators comprise one or more of a category error, a speed error, a position error, a motion state error and a perception stability of the perception target, and a lane line distribution error and / or a lane line curvature error of the perception lane line.

[0009] Optionally, before the obtaining the test scores of the perception performance of the plurality of to-be-tested sensors in one or more real scenes, the method further comprises: setting a test time and a test mileage in each real scene; and generating a planned route and a driving time according to the real scene, the test time and the test mileage, and testing the plurality of to-be-tested sensors in the real scene according to the planned route and the driving time.

[0010] Optionally, different types of to-be-tested sensors are tested simultaneously.

[0011] Optionally, the types of the to-be-tested sensors comprise a camera, a laser radar and a millimeter wave radar.

[0012] The second aspect embodiment of the present application provides a selection device for perception of a driving assistance system, comprising: a first obtaining module configured to obtain test scores of perception performance of a plurality of to-be-tested sensors in one or more real scenes; a second obtaining module configured to obtain evaluation weights of the perception performance of each to-be-tested sensor in each real scene according to a corresponding relationship between a type of the to-be-tested sensor and the evaluation weights of the perception performance in each real scene; and an output module configured to calculate a test total score of the perception performance of each to-be-tested sensor according to the test score of each real scene and the corresponding evaluation weight, and select a target sensor for the driving assistance system from each type of to-be-tested sensor based on the test total score.

[0013] The third aspect embodiment of the present application provides an electronic device, comprising: a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the selection method for perception of a driving assistance system as described above.

[0014] The fourth aspect of the present application provides a selection system for perception of a driving assistance system, comprising: a test vehicle, wherein the test vehicle is provided with a driving assistance system and a true value system, wherein the driving assistance system comprises a plurality of to-be-tested sensors, and the true value system is used to obtain true value data in one or more real scenes when the test vehicle is tested; an electronic device, configured to obtain test scores of perception performance of the plurality of to-be-tested sensors in the one or more real scenes based on the true value data; obtain an evaluation weight of the perception performance of each to-be-tested sensor in each real scene according to a corresponding relationship between the type of the to-be-tested sensor and the evaluation weight of the perception performance in each real scene; calculate a test total score of the perception performance of each to-be-tested sensor according to the test score and the corresponding evaluation weight of each real scene, and select a target sensor with perception performance for the driving assistance system from to-be-tested sensors of each type based on the test total score.

[0015] Therefore, the present application has at least the following beneficial effects:

[0016] The embodiments of the present application can more accurately understand the actual performance of each sensor by obtaining test scores of perception performance of a plurality of to-be-tested sensors in real scenes, can ensure that different types of sensors are treated fairly in the evaluation process by setting reasonable evaluation weights according to the type of the to-be-tested sensor, and can calculate a test total score of each to-be-tested sensor by combining the test score and the corresponding evaluation weight of each real scene, so as to comprehensively evaluate the overall performance of the sensor, select a sensor with better perception performance, and ensure that the driving assistance system is equipped with the sensor with the best performance. Therefore, the technical problems of how to accurately and efficiently select a sensor with better perception performance for the driving assistance system are solved.

[0017] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0018] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, taken in conjunction with the accompanying drawings, in which:

[0019] Figure 1 A flowchart of a selection method for perception of a driving assistance system according to an embodiment of the present application is provided;

[0020] Figure 2 A structural diagram of a selection system for perception of a driving assistance system according to an embodiment of the present application is provided;

[0021] Figure 3 A flowchart of a selection method for perception of a driving assistance system according to an embodiment of the present application is provided;

[0022] Figure 4A schematic diagram of a selection device for a driving assistance system according to an embodiment of the present application;

[0023] Figure 5 A structural schematic diagram of an electronic device according to an embodiment of the present application;

[0024] Figure 6 A schematic diagram of a selection system for a driving assistance system according to an embodiment of the present application;

[0025] Figure 7 A structural diagram of a selection system for a driving assistance system according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] Embodiments of the present application are described in detail below with reference to the accompanying drawings, in which the same or similar components have the same or similar designations and functions throughout various figures and / or portions of the figures. The embodiments described below are examples in which the present application is applied to explain the present application, and are not to be understood as limiting the present application.

[0027] The selection method, device, electronic device and system for a driving assistance system according to the embodiments of the present application are described below with reference to the accompanying drawings. In view of the problem that the driving assistance system mentioned in the above background art is difficult to be quantitatively evaluated in various application scenarios, the present application provides a selection method for a driving assistance system, in which the performance test scores of a plurality of to-be-tested sensors in real scenes are obtained, so that the actual performance of each sensor can be more accurately understood, reasonable evaluation weights are set according to the types of the to-be-tested sensors, so that different types of sensors can be treated fairly during the evaluation process, and the test total scores of each to-be-tested sensor can be calculated by combining the test scores of each real scene and the corresponding evaluation weights, so as to comprehensively evaluate the overall performance of the sensor, select a sensor with better perception performance, and ensure that the driving assistance system is equipped with the best sensor. Thus, the problems in the related art that the perception performance of different types of sensors in different real scenes cannot be quantitatively evaluated, and the better sensor is selected for the driving assistance system based on the evaluation result are solved.

[0028] Specifically, Figure 1 A flowchart of a selection method for a driving assistance system according to an embodiment of the present application.

[0029] As Figure 1 shown, the selection method for a driving assistance system includes the following steps:

[0030] In step S101, the test scores of the perception performance of a plurality of to-be-tested sensors in one or more real scenes are obtained.

[0031] The to-be-tested sensor can be a sensor device that has not been comprehensively evaluated or selected and needs to be tested for performance, the type of the to-be-tested sensor can include a camera, a laser radar, a millimeter wave radar, etc., the real scene can be various environments and conditions that the sensor can encounter in actual application, the real scene can include a scene composed of a road type, a weather condition, and a traffic condition, etc., and the perception performance can be the ability of the sensor to capture, process, and analyze environmental information.

[0032] It can be understood that the embodiments of the present application can obtain test scores of the perception performance of the plurality of to-be-tested sensors in one or more real scenes, and by testing the sensors in actual environments, the perception performance of the sensors in various real scenes can be directly reflected, facilitating subsequent comparison and evaluation of the performance of the sensors.

[0033] In the embodiments of the present application, obtaining the test scores of the perception performance of the plurality of to-be-tested sensors in one or more real scenes includes: obtaining perception data and ground truth data of the plurality of to-be-tested sensors in each real scene; and calculating a test score of each real scene according to the perception data and the ground truth data.

[0034] The perception data can be environmental information captured and processed by the sensor in the real scene, and the ground truth data can be verified accurate environmental information corresponding to the perception data.

[0035] It can be understood that by obtaining the perception data and the ground truth data and calculating the test scores, the embodiments of the present application can quantitatively show the performance difference of the sensor in a specific scene, so that the user can intuitively compare the performance of different sensors.

[0036] It should be noted that the perception data and the ground truth data are detection data of surrounding targets and surrounding lane lines, etc., which are data detected by different systems at the same time, the perception data is data detected by the sensor, and the ground truth data is data detected by a ground truth system, the ground truth system can accurately detect sensors of surrounding targets and / or surrounding lane lines, etc., and the ground truth data serves as reference data for evaluating the perception data.

[0037] Specifically, the ground truth system can include one or more of a laser radar, a millimeter wave radar, a monitoring camera, a combined inertial navigation, and a multi-sensor, wherein the ground truth system acquires size information through the laser radar, acquires speed information through the continental millimeter wave radar, acquires position information through the combined inertial navigation, and acquires category and lane line information through the laser radar and the monitoring camera, and this multi-sensor fusion scheme ensures high precision of the ground truth system. The time of each sensor is synchronized, the satellite time is taken as a reference, and a time stamp is sent through a standard interface, so that the time of each sensor is finally synchronized.

[0038] Therefore, the embodiment of the present application can collect ground truth data by using a ground truth system in the same real scene, accurately determine whether the perception data is accurate by comparing the perception data and the ground truth data, and thus accurately calculate the test score of each real scene by using the ground truth data and the perception data, which has high accuracy and practicability and is helpful for comprehensively evaluating the performance of the sensor.

[0039] In the embodiment of the present application, the calculation of the test score of each real scene according to the perception data and the ground truth data includes: identifying the perception target and the perception lane line in the perception data and the ground truth data; determining respective accuracy indexes of the perception target and the perception lane line according to the perception data and the ground truth data, and obtaining respective scoring standards of the perception target and the perception lane line; and calculating the test score of each real scene according to the respective accuracy indexes and the respective scoring standards.

[0040] The perception target can be a specific object or object recognized by the sensor in the real scene, such as a vehicle, a pedestrian, a road marking, a traffic signal, etc.; the perception lane line can be a lane boundary line recognized by the sensor on the road; the respective accuracy indexes can include one or more of a category error, a speed error, a position error, a motion state error and a perception stability of the perception target, and a lane line distribution error and / or a lane line curvature error of the perception lane line.

[0041] The category error of the perception target can be an error occurred when the sensor identifies the category of the perception target, for example, the sensor can incorrectly identify a car as a pedestrian or a bicycle, etc.; the speed error can be the difference between the speed of the perception target measured by the sensor and the real speed; the position error can be the deviation between the position of the target perceived by the sensor and the real position; the motion state error can be the error generated when the sensor judges the motion state of the perception target; the perception stability can be the consistency of the identification of the same perception target by the sensor at different times or under different conditions; the lane line distribution error can be the deviation between the position of the lane line recognized by the sensor and the real position of the lane line; and the lane line curvature error can be the difference between the curvature of the lane line measured by the sensor and the real curvature.

[0042] It can be understood that the embodiment of the present application can identify the perception target and the perception lane line in the perception data and the ground truth data, determine the accuracy indexes and the scoring standards of the perception target and the perception lane line according to these data, and calculate the test score of each real scene, so as to quantify the perception accuracy score of the perception target and the perception lane line by using the accuracy indexes and the scoring standards, and realize the quantitative scoring of different real scenes in the same dimension, and accurately evaluate the perception performance of the sensor.

[0043] Specifically, as shown in Table 1, when the vehicle travels to the specified scene route according to the system planned time, the scene and time information are recorded, the perception target related parameters are started to be stored and extracted, and the frequency of comparing the measured perception and true value parameters according to the preset time interval is recorded corresponding score. The preset time interval can be 1 minute or 2 minutes, etc., which can be specifically selected according to the test requirements.

[0044] Table 1: Score table of perception target

[0045]

[0046] At the same time, the perception lane line related parameters are stored and extracted, the frequency of comparing the measured perception and true value parameters according to the preset interval time is recorded corresponding score, and the score table of the perception lane line is as shown in Table 2 below.

[0047] Table 2: Score table of perception lane line

[0048]

[0049] Therefore, the embodiment of the application quantifies the scores of different real scenes based on the above score table and the accurate index calculated according to the true value data, so as to realize the quantitative scoring of different real scenes in the same dimension, realize the accurate evaluation of the perception performance of the sensor, and improve the accuracy of the evaluation.

[0050] In the embodiment of the application, before the plurality of to-be-tested sensors, there further comprises: setting a test time and a test mileage in each real scene; generating a planned route and a driving time according to the real scene, the test time and the test mileage, and testing the plurality of to-be-tested sensors in the real scene according to the planned route and the driving time.

[0051] The test time can be a time length set for performance evaluation of the to-be-tested sensor in each real scene, such as one hour, etc., the test mileage can be a total distance traveled by the to-be-tested sensor in the real scene within the test time, such as ten kilometers, etc., the planned route can be a driving path formulated for the to-be-tested sensor based on the real scene, the test time and the test mileage, and the driving time can be a time required for the to-be-tested sensor to travel according to the planned route.

[0052] It can be understood that the embodiments of the present application ensure the sufficiency and effectiveness of the test process by setting the test time and test mileage under each real scene, generating a planned route and driving time according to the real scene, the test time and the test mileage, wherein the planned route can include various road types, traffic conditions and environmental conditions, etc., to provide a more real, complex and comprehensive scene for the test, and the plurality of sensors to be tested are tested in the real scene according to the planned route and the driving time, which can directly reflect the performance of the sensors in actual application. Through the test in the real scene, the actual data of the sensors in different scenes can be obtained, so that the performance of the sensors can be accurately evaluated.

[0053] Specifically, one autonomous vehicle development project evaluates five different types of sensors: lidar, camera, millimeter wave radar, ultrasonic radar, and infrared sensor.

[0054] The test time is set to 3 hours and the test mileage is set to 30 kilometers. A city road route containing various road types and traffic conditions is planned. The driving time is adjusted according to the road conditions and test speed to ensure that the test mileage of 30 kilometers is completed within 3 hours.

[0055] The five sensors to be tested are installed in appropriate positions of the autonomous vehicle and necessary calibration and initialization work is performed to ensure that the sensors can work normally before the test starts. At the start of the test, the five sensors will work simultaneously to collect and output perception data in real time. The test vehicle travels according to the planned route and driving time to simulate various situations in real driving scenarios. During the test, the perception data of all sensors is collected and compared with the true value data. By calculating the accuracy index and scoring standard of each sensor, their performance in the same scene can be quantitatively evaluated.

[0056] Thus, multiple types of sensors to be tested can be tested simultaneously, which can significantly improve the test efficiency and compare the performance of different sensors in the same real scene.

[0057] In step S102, the evaluation weight of the perception performance of each sensor to be tested in each real scene is obtained according to the correspondence between the type of the sensor to be tested and the evaluation weight of the perception performance in each real scene.

[0058] The type of the sensor to be tested can be various different types of sensors tested in the evaluation process, such as lidar, millimeter wave radar, camera, etc., and the evaluation weight can be the proportion of different functions and performance indicators in the overall evaluation, such as safety performance, driving experience and intelligent level, etc. Evaluation indicators can be determined according to the true value system performance.

[0059] It should be noted that different types of sensors such as lidar, millimeter wave radar, camera, etc. have different types of sensing data, for example, the sensing data of the camera is image data, and therefore the light requirement is relatively high. In the daylight and other sufficient light conditions, the sensing performance of each type of camera is similar, but in the night and other insufficient light conditions, the sensing performance of each type of camera usually has a large difference. In order to select a camera with better night sensing performance from a plurality of types of cameras within a target cost range, the embodiment of the present application gives a higher weight to the real scene under the condition of insufficient light such as night when setting the weight of each real scene, so as to meet the selection demand of the camera, so that the weight of the camera in each real scene can be reasonably set from the actual application scene, the accuracy of the camera sensing evaluation is improved, and the reliability of the subsequent selection is improved. The weight setting of lidar, millimeter wave radar and camera in each real scene is similar, and the weight can also be reasonably set in combination with the actual application scene. To avoid redundancy, no further description is given.

[0060] Therefore, the embodiment of the present application can set the evaluation weight of the sensing performance of each to-be-tested sensor in each real scene according to the type of the to-be-tested sensor. By reasonably setting the weight, the performance difference of different sensors in different scenes can be more accurately reflected, and the evaluation result is more in line with the actual application demand.

[0061] In step S103, the test total score of the sensing performance of each to-be-tested sensor is calculated according to the test score of each real scene and the corresponding evaluation weight, and a target sensor with sensing performance is selected from each type of to-be-tested sensor for the driving assistance system based on the test total score.

[0062] It can be understood that the embodiment of the present application realizes accurate evaluation of the sensing performance of each to-be-tested sensor in multiple scenes by comprehensively considering the test score of each real scene and the corresponding evaluation weight. Based on the test total score of each to-be-tested sensor, a sensor with better sensing performance can be selected from each type of sensor, so as to ensure that the driving assistance system can utilize better sensor technology to improve its sensing ability in various driving scenes. At the same time, after the evaluation by the test total score, a target sensor with better performance is selected from each type of to-be-tested sensor, so as to complete the selection of the sensor.

[0063] Therefore, the embodiment of the present application realizes accurate and objective evaluation of the sensor by generating the evaluation result of the sensing performance through the total score, wherein the evaluation result can be in a qualitative manner, such as excellent, good, general or poor, or in a quantitative manner, such as a performance level represented by a percentage or a specific value, thereby facilitating the selection of a target sensor with better performance from each type of sensor to be tested, and improving the reliability of sensor selection.

[0064] For example, in order to select a camera, and it is expected to select a camera with better sensing performance in insufficient light conditions such as at night, four real scenes of urban road (day), highway (day), urban road (night), and highway (night) are set as shown in Table 3, and the weights of different real scenes are set according to the selection expectation as shown in Table 3, so that the sensing performance of the camera in the four scenes can be comprehensively evaluated.

[0065] Table 3: Evaluation table of camera sensing performance

[0066] Real scene Test score Weight Total score Urban road (day) 80 10％ 8 Urban road (night) 50 50％ 25 Highway (day) 75 10％ 7.5 Highway (night) 40 30％ 12 Total score 52.5

[0067] When the total score is more than 90 points, it can be evaluated as excellent; 80-89 points can be evaluated as good; 70-79 points can be evaluated as medium; 60-69 points can be evaluated as passing; and less than 60 points can be evaluated as failing.

[0068] Therefore, the evaluation of the sensing performance of the camera in Table 3 is not passing. The embodiment of the present application can test multiple cameras at the same time in the same scene, select the target camera with the highest total score and passing total score, and complete the selection of the camera. The testing of laser radar and millimeter wave radar and other types of sensors is the same as that of the camera, and different types of sensors are tested at the same time in the same real scene. When the total score of different types of sensors in the same real scene is calculated, the weights are set according to the test requirements of different types of sensors, so as to realize the simultaneous test selection of different types of sensors, and the individualized test selection, improve the accuracy and reliability of the selection, and improve the user experience.

[0069] According to the selection method of the driving assistance system perception provided in the embodiments of the present application, by obtaining the perception performance test scores of a plurality of to-be-tested sensors in real scenes, the actual performance of each sensor can be more accurately understood, reasonable evaluation weights are set according to the types of the to-be-tested sensors, so that different types of sensors can be treated fairly in the evaluation process, and the test total score of each to-be-tested sensor can be calculated by combining the test scores of each real scene and the corresponding evaluation weights, so as to comprehensively evaluate the overall performance of the sensor, select a sensor with better perception performance, and ensure that the driving assistance system is equipped with a sensor with better performance.

[0070] In the following, the perception performance evaluation of the sensor will be taken as an example, as shown in the following table, the test system of the embodiments of the present application can include a true value system, a to-be-tested sensor, an evaluation system and a report output system, wherein the evaluation system and the report output system can be arranged in an electronic device, for evaluating the perception performance according to the test data after the test is completed, and outputting an evaluation report. The specific process is as follows: Figure 2

[0071] (1) The true value system can include one or more of a laser radar, a millimeter wave radar, a monitoring camera, a combined inertial navigation and a multi-sensor, wherein the true value system obtains size information through the laser radar, obtains speed information through the continental millimeter wave radar, obtains position information through the combined inertial navigation, and obtains category and lane line information through the laser radar and the monitoring camera. This multi-sensor fusion scheme ensures the high accuracy of the true value system. The time synchronization of each sensor is based on the satellite time obtained, and the time stamp is sent through a standard interface, so that the time synchronization of each sensor is finally realized.

[0072] (2) The to-be-tested sensor can be installed in the “measured system installation area” through a support, and the perception information and different scene performance information tested are time-stamped and sent to the evaluation system through CAN messages.

[0073] (3) The evaluation system and the report output system are composed of an industrial computer (high-precision map / evaluation system software), a storage device and a display. When the vehicle drives to the specified scene route according to the system planning time, the evaluation system records the scene and time information, and starts to store and extract the perception target and lane line related parameters at the same time. The corresponding scores are recorded according to the frequency of comparing the measured perception and the true value parameters every minute.

[0074] After the road test is completed, the evaluation system will give a score in different time periods according to different scenes, for evaluating the performance of the perception in different situations, and the evaluation system will also give a comprehensive score according to the following formula:

[0075] Total score = Scene 1*K1 + Scene 2*K2 + Scene 3*K3 + Scene 4*K4 + … Scene n*Kn,​

[0076] Wherein, scene n represents the n-th scene, Kn represents the weight of scene n, which can be adjusted by the function of ADAS application.

[0077] The evaluation system passes all scores to the report output system, which presents them to the user on a display screen.

[0078] The selection method of the driving assistance system perception according to the system shown in the embodiment of the application will be described below. Figure 2 As shown in the figure, the specific steps are as follows: Figure 3

[0079] First step: select the sensor to be tested, install the sensor to be tested in the "measured system installation area", and arrange it within ±50 mm. The embodiment of the application can support 3 cameras and 2 millimeter wave radars for evaluation at the same time.

[0080] Second step: based on the vehicle coordinate, calibrate the measured sensor, develop CAN message, and send the target and lane line information perceived by the measured sensor to the bus to pass to the subsequent evaluation system.

[0081] Third step: planning of the test route, relying on the pre-evaluation direction and focus. Turn on the perception evaluation system, after the system self-checking is passed, a selection interface of the measured sensor will pop up, select the sensor to be tested on the interface, and click OK. The system will pop up a list of pre-planned routes, enter the number of kilometers in the blank, fill 0 for scenes that do not need to be tested, and click OK. The high-precision map in the evaluation system will automatically plan the driving route and the corresponding driving time according to the demand.

[0082] Fourth step: drive according to the recommended time point and route of the planned route.

[0083] Fifth step: after the road test is completed, click to generate a report to view the report results directly on the screen.

[0084] In summary, the embodiment of the application can build a high-precision true value system based on a real car on any social road, simultaneously evaluate multiple sensors in different scenes, and directly use quantitative scores in the same dimension to evaluate the sensor perception, which has the characteristics of high true value precision, no site limitation, simultaneous multi-sensing and multi-scene comparison, score-based comparison results, and fast results.

[0085] Secondly, the selection device of the driving assistance system perception according to the embodiment of the application is described with reference to the accompanying drawings.

[0086] Figure 4 is a block schematic diagram of the selection device of the driving assistance system perception according to the embodiment of the application.

[0087] As​Figure 4 As shown, the selection device for the perception of the driving assistance system 10 comprises a first acquisition module 100, a second acquisition module 200 and an output module 300.

[0088] The first acquisition module 100 is configured to acquire test scores of the perception performance of a plurality of to-be-tested sensors in one or more real scenes; the second acquisition module 200 is configured to acquire an evaluation weight of the perception performance of each to-be-tested sensor in each real scene according to a corresponding relationship between the type of the to-be-tested sensor and the evaluation weight of the perception performance in each real scene; and the output module 300 is configured to calculate a test total score of the perception performance of each to-be-tested sensor according to the test score and the corresponding evaluation weight of each real scene, and select a target sensor with the perception performance for the driving assistance system from each type of to-be-tested sensor based on the test total score.

[0089] It should be noted that the foregoing description of the selection method for the perception of the driving assistance system is also applicable to the selection device for the perception of the driving assistance system, which will not be described here again.

[0090] The selection device for the perception of the driving assistance system provided by the embodiment of the present application can more accurately understand the actual performance of each sensor by acquiring the test scores of the perception performance of a plurality of to-be-tested sensors in real scenes, can ensure that different types of sensors are treated fairly in the evaluation process by setting reasonable evaluation weights according to the types of the to-be-tested sensors, can calculate the test total score of each to-be-tested sensor by combining the test score and the corresponding evaluation weight of each real scene, thereby comprehensively evaluating the overall performance of the sensor, selecting a sensor with better perception performance, and ensuring that the driving assistance system is equipped with the sensor with the best performance.

[0091] Figure 5 The electronic device provided by the embodiment of the present application is shown in a structural schematic diagram. The electronic device can comprise:

[0092] The memory 501, the processor 502 and the computer program stored in the memory 501 and executable on the processor 502.

[0093] The processor 502 implements the selection method for the perception of the driving assistance system provided in the foregoing embodiments when executing the program.

[0094] Further, the electronic device further comprises:

[0095] The communication interface 503 is configured to communicate between the memory 501 and the processor 502.

[0096] The memory 501 is configured to store the computer program executable on the processor 502.

[0097] The memory 501 may include a high-speed RAM (Random Access Memory) memory, and may also include a non-volatile memory, such as at least one disk memory.

[0098] If the memory 501, the processor 502, and the communication interface 503 are implemented independently, the communication interface 503, the memory 501, and the processor 502 can be connected to each other via a bus and communicate with each other. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0099] Optionally, in a specific implementation, if the memory 501, the processor 502 and the communication interface 503 are integrated on a chip, the memory 501, the processor 502 and the communication interface 503 can communicate with each other through an internal interface.

[0100] The processor 502 may be a CPU (Central Processing Unit), or an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present application.

[0101] The embodiment of the present application also provides a selection system for driver assistance system perception, such as Figure 6 As shown, the driving assistance system perception selection system 20 includes: a test vehicle 210 and an electronic device 220.

[0102] The test vehicle 210 is provided with a driving assistance system and a true value system, wherein the driving assistance system includes a plurality of to-be-tested sensors, and the true value system is used to obtain true value data in one or more real scenes when the test vehicle 210 is tested; the electronic device 220 is used to obtain a test score of the perception performance of the plurality of to-be-tested sensors in the one or more real scenes based on the true value data; according to a corresponding relationship between the type of the to-be-tested sensor and the evaluation weight of the perception performance in each real scene, an evaluation weight of the perception performance of each to-be-tested sensor in each real scene is obtained; a test total score of the perception performance of each to-be-tested sensor is calculated according to the test score and the corresponding evaluation weight of each real scene, and a target sensor of the perception performance is selected for the driving assistance system from each type of to-be-tested sensor based on the test total score.

[0103] For example, a high-precision true value system is installed on a vehicle, and a measured system installation position is reserved, wherein, as shown in the figure, Figure 7 a laser radar, a monitoring camera and a combined inertial navigation are arranged on a special bracket installed on the roof, have the effects of stable installation, simple and elegant, vehicle body streamline design, easy installation and removal, and do not need to punch and groove the test vehicle; a millimeter wave radar is installed in the central position of the front bumper of the vehicle; an industrial computer, a storage device and a display are arranged at the position of the rear seat and the trunk of the vehicle. After time synchronization, two groups of data are sent to the industrial computer evaluation system for data extraction, comparative analysis, and the results are transmitted to the report output system for user review and export.

[0104] According to the selection system for the perception of the driving assistance system provided in the embodiments of the present application, the test scores of the perception performance of the plurality of to-be-tested sensors in the real scenes are obtained, so that the actual performance of each sensor can be more accurately understood, the reasonable evaluation weight is set according to the type of the to-be-tested sensor, so that the sensors of different types can be treated fairly in the evaluation process, and the test total score of each to-be-tested sensor is calculated by combining the test score of each real scene and the corresponding evaluation weight, so that the overall performance of the sensor can be comprehensively evaluated, the sensor with better perception performance is selected, and the sensor with the best performance is ensured to be equipped in the driving assistance system.

[0105] In the description of the application, reference to "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" etc. means that a particular feature, structure, material or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. The illustrative appearances of the above-mentioned terms in various places in the specification are not necessarily referred to the same embodiment or example nor are separate or alternative embodiments or examples of the application mutually exclusive of one another. Moreover, the particular features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples. Furthermore, the application encompasses different embodiments or examples of the application to the extent permitted by the law.

[0106] Furthermore, the terms "first", "second", etc. are used herein only to describe different instances, features, steps, etc. and do not imply a relative importance or a specific order of the features being described. Thus, features defined with "first", "second" etc. can include at least one of the features. In the description of the application, the meaning of "N" is at least two, such as, for example, two, three and the like, unless otherwise expressly specified.

[0107] Any process or method described in a flowchart or otherwise described herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing specific logic functions (or steps) of the process, and the various embodiments of the application can include additional or fewer steps performing the same or equivalent functions as those described in the same or different order as those described. Those skilled in the art will appreciate that the order of the steps in the processes described herein is immaterial so long as the desired result is achieved.

[0108] It should be understood that parts of the application can be implemented in hardware, software, firmware or a combination thereof. In the above embodiments, the steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. As with a hardware implementation, the methods can also be implemented with any of a number of technologies, e.g., a number of technologies, including but not limited to discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays, field programmable gate arrays, or the like.

[0109] Those skilled in the art can understand that all or part of the steps carried out by the above-mentioned embodiment method can be completed by programs instructing related hardware, and the programs can be stored in a computer readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

Claims

1. A method for selecting a driver assistance system perception, characterized in that: The following steps are involved: Obtain test scores of the perception performance of multiple sensors under test in one or more real scenarios; Obtaining the evaluation weight of the perception performance of each sensor to be tested in each real scene according to the corresponding relationship between the type of the sensor to be tested and the evaluation weight of the perception performance in each real scene; A total test score of the perception performance of each sensor to be tested is calculated according to the test score of each real scenario and the corresponding evaluation weight, and a target sensor for the driving assistance system is selected from each type of sensors to be tested based on the total test score.

2. The method for selecting a driver assistance system according to claim 1, wherein: Obtaining test scores of the perception performance of the plurality of sensors to be tested in one or more real scenarios includes: Acquire the perception data and true value data of the multiple sensors to be tested in each real scene; A test score for each real scene is calculated based on the perception data and the true value data.

3. The method for selecting a driver assistance system according to claim 2, wherein: Calculating the test score of each real scene according to the perception data and the true value data includes: Identifying a perceived target and a perceived lane line in the perception data and the ground truth data; Determining respective accuracy indicators of the perceived target and the perceived lane line based on the perception data and the true value data, and obtaining respective scoring criteria for the perceived target and the perceived lane line; The test score of each real scenario is calculated according to the respective accuracy indicators and the respective scoring criteria.

4. The method for selecting a driver assistance system perception according to claim 3, characterized in that: The respective accuracy indicators include one or more of the category error, speed error, position error, motion state error and perception stability of the perceived target, and the lane line distribution error and / or lane line curvature error of the perceived lane line.

5. The method for selecting a driver assistance system perception according to claim 1, characterized in that: Before obtaining the test scores of the perception performance of multiple sensors under test in one or more real scenarios, it also includes: Set the test time and test mileage for each real scenario; A planned route and driving time are generated according to the real scenario, the test time, and the test mileage, and the multiple sensors to be tested are tested in the real scenario according to the planned route and the driving time.

6. The method for selecting a driver assistance system according to claim 5, wherein: Different types of sensors under test are tested simultaneously.

7. The method for selecting a driver assistance system perception according to claim 1, characterized in that: The types of sensors to be tested include cameras, lidars, and millimeter-wave radars.

8. A device for selecting a driver assistance system perception, characterized in that: include: A first acquisition module is used to obtain test scores of the perception performance of multiple sensors under test in one or more real scenarios; A second acquisition module is used to obtain the evaluation weight of the perception performance of each sensor to be tested in each real scene according to the corresponding relationship between the type of the sensor to be tested and the evaluation weight of the perception performance in each real scene; An output module is used to calculate a total test score of the perception performance of each sensor to be tested based on the test score of each real scenario and the corresponding evaluation weight, and select a target sensor for the driving assistance system from each type of sensor to be tested based on the total test score.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the selection method for driver assistance system perception as described in any one of claims 1 to 7.

10. A driver assistance system perception selection system, characterized in that: include: A test vehicle, wherein the test vehicle is provided with a driving assistance system and a truth system, wherein the driving assistance system includes a plurality of sensors to be tested, and the truth system is used to obtain truth data in one or more real scenarios when testing the test vehicle; An electronic device is used to obtain, based on the true value data, test scores of the perception performance of multiple sensors to be tested in one or more real scenarios; obtain the evaluation weight of the perception performance of each sensor to be tested in each real scenario based on the correspondence between the type of the sensor to be tested and the evaluation weight of the perception performance in each real scenario; calculate the total test score of the perception performance of each sensor to be tested based on the test score and the corresponding evaluation weight of each real scenario, and select a target sensor for the driving assistance system from each type of sensor to be tested based on the total test score.

Images