Target fusion performance evaluation method and device for intelligent driving and electronic equipment

Abstract

The invention provides a target fusion performance evaluation method and device for intelligent driving and electronic equipment, and belongs to the technical field of automatic driving, and the method comprises the steps: obtaining perception data collected by a plurality of vehicle-mounted sensors; performing data analysis on the sensing data acquired by the plurality of vehicle-mounted sensors to obtain readable data; fusing the readable data corresponding to different vehicle-mounted sensors to obtain a fusion signal of a single target in a continuous time period; performing time alignment on sensing data acquired by a single sensor corresponding to a single target and a fusion signal of the single target in a continuous time period; determining a deviation ratio between the fusion signal and the sensing data acquired by the single vehicle-mounted sensor based on the fusion signal and the sensing data acquired by the single vehicle-mounted sensor; and determining a target fusion performance evaluation result based on the deviation rate. The target perception fusion performance evaluation method can solve the problem that an existing target perception fusion performance evaluation method is relatively fixed in use scene of performance evaluation and is not flexible enough.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a method, device and electronic equipment for evaluating target fusion performance of intelligent driving. Background Art

[0002] The existing performance evaluation methods for target perception fusion are divided into the following two types: The first method is to use a tape measure or other distance measurement tools to measure the actual distance between the vehicle and the target object when the vehicle and the target object are both stationary. Then, the actual distance between the vehicle and the target object is compared with the sensor perception value or the perception fusion value to evaluate the performance of a single sensor or the perception fusion algorithm. The second method is to install measurement equipment on the ego vehicle and the target vehicle to obtain their relative distance or speed when the ego vehicle and the target object are respectively stationary or moving. The true value obtained by this method is more accurate and can effectively evaluate the performance of the perception value and fusion algorithm.

[0003] The first method needs to be performed in a stationary state and can only evaluate distance, not other characteristic values ​​such as speed. In addition, it can only evaluate one target object at a time. In addition, the recorded true value and perceived value are at a time, and fluctuations in values ​​during the measurement or perception process cannot be ruled out.

[0004] The second method requires lane lines to assist in measuring lateral distance or speed. If the lane lines have obvious curvature, more intensive lane line point collection is required, otherwise it will affect the lateral accuracy. The installation and debugging of the equipment requires the assistance of professionals, and the installation and debugging are complicated. In addition, only the target objects with the corresponding equipment installed can detect the true value, and it is impossible to perform performance evaluation on an unlimited number of targets, which limits the test scenarios and scope.

[0005] In summary, the existing performance evaluation methods for target perception fusion have the problem that the usage scenarios of performance evaluation are relatively fixed and not flexible enough. Summary of the invention

[0006] In view of this, it is necessary to provide a target fusion performance evaluation method, device and electronic device for intelligent driving to solve the technical problem that the existing target perception fusion performance evaluation method has a relatively fixed and inflexible use scenario of performance evaluation.

[0007] In order to solve the above problems, on the one hand, the present invention provides a method for evaluating target fusion performance of intelligent driving, comprising: Acquire perception data collected by multiple vehicle-mounted sensors; Analyze the perception data collected by multiple vehicle-mounted sensors to obtain readable data; The readable data corresponding to different vehicle-mounted sensors are fused to obtain the fusion signal of a single target in a continuous time period; Temporally align the perception data collected by a single sensor corresponding to a single target with the fusion signal of the single target in a continuous time period; After completing the time alignment, determining a deviation rate between the fused signal and the perception data collected by the single onboard sensor based on the fused signal and the perception data collected by the single onboard sensor; Based on the deviation rate, a target fusion performance evaluation result is determined.

[0008] In a possible implementation, determining, based on the fused signal and the perception data collected by the single vehicle-mounted sensor, a deviation rate between the fused signal and the perception data collected by the single vehicle-mounted sensor includes: The difference between the fusion signal and the perception data collected by the single vehicle-mounted sensor is divided by the perception data collected by the single vehicle-mounted sensor to obtain the deviation rate between the fusion signal and the perception data collected by the single vehicle-mounted sensor.

[0009] In a possible implementation, the target fusion performance evaluation method for intelligent driving further includes: In the case where the deviation rate is greater than the target deviation threshold, the abnormal cause of the deviation is determined based on the perception data moment corresponding to the time when the deviation rate is greater than the target deviation threshold.

[0010] In a possible implementation, the target fusion performance evaluation method for intelligent driving determines the target fusion performance evaluation result based on the deviation rate, including: Among the deviation rates corresponding to multiple vehicle-mounted sensors, the deviation rates within the target deviation range are excluded to obtain the remaining deviation rate, and the target fusion performance evaluation result is determined based on the remaining deviation rate.

[0011] In a possible implementation, the perception data collected by a single sensor corresponding to a single target is time-aligned with the fusion signal of the single target in a continuous time period, including: The perception data collected by a single sensor corresponding to a single target is interpolated with the fusion signal of the single target in a continuous time period to achieve time alignment.

[0012] In a possible implementation, the deviation rate includes: a deviation rate corresponding to a lateral and longitudinal distance, a lateral and longitudinal speed, a target type, or a confidence level.

[0013] In one possible implementation, multiple vehicle-mounted sensors include: cameras, millimeter-wave radars, and lidars.

[0014] On the other hand, the present invention also provides a target fusion performance evaluation device for intelligent driving, comprising: An acquisition module, used to acquire perception data collected by multiple vehicle-mounted sensors; The parsing module is used to parse the perception data collected by multiple vehicle-mounted sensors to obtain readable data; A fusion module is used to fuse the readable data corresponding to different vehicle-mounted sensors to obtain a fusion signal of a single target in a continuous time period; An alignment module is used to time-align the perception data collected by a single sensor corresponding to a single target with the fusion signal of the single target in a continuous time period; a deviation calculation module, configured to determine, after completing the time alignment, a deviation rate between the fused signal and the perception data collected by the single onboard sensor based on the fused signal and the perception data collected by the single onboard sensor; An evaluation module is used to determine a target fusion performance evaluation result based on the deviation rate.

[0015] On the other hand, the present invention also provides an electronic device, including a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps of the target fusion performance evaluation method for intelligent driving as described in any one of the above items.

[0016] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the target fusion performance evaluation method for intelligent driving as described in any one of the above items.

[0017] The beneficial effects of adopting the above implementation mode are: the target fusion performance evaluation method, device and electronic device for intelligent driving provided by the present invention obtain a fusion signal of a single target in a continuous time period by fusing the readable data corresponding to different vehicle-mounted sensors; the perception data collected by a single sensor corresponding to a single target is time-aligned with the fusion signal of the single target in a continuous time period; after completing the time alignment, based on the fusion signal and the perception data collected by the single vehicle-mounted sensor, the deviation rate between the fusion signal and the perception data collected by the single vehicle-mounted sensor is determined; based on the deviation rate, a statistical analysis is performed to observe the influence of different influencing factors on the deviation rate, and the target fusion performance evaluation result is determined, thereby raising questions, solving problems, and forming conclusions. Fusion performance evaluation can be performed even when the vehicle and the target object are in motion respectively, and there is no need to measure the lateral distance or speed, which requires lane lines to assist. Compared with existing means, it is easy to implement, flexible in scenarios, and highly automated, thereby solving the technical problem that the existing target perception fusion performance evaluation method has a relatively fixed and inflexible performance evaluation scenario. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A flowchart of an embodiment of a method for evaluating target fusion performance of intelligent driving provided by the present invention; Figure 2 This is a numerical timing diagram of the perception fusion index provided by the present invention; Figure 3 A flowchart of another embodiment of the target fusion performance evaluation method for intelligent driving provided by the present invention; Figure 4 A time series diagram of the perception fusion index deviation rate provided by the present invention; Figure 5 The average value curve of the fusion and perception deviation rate for different time periods provided by the present invention; Figure 6 A principle block diagram of an embodiment of a target fusion performance evaluation device for intelligent driving provided by the present invention; Figure 7 A schematic structural diagram of an embodiment of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0021] In the description of the embodiments of the present application, unless otherwise specified, “plurality” means two or more.

[0022] The terms "including" and "having" and any variations thereof in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product or equipment comprising a series of steps or modules is not necessarily limited to those steps or modules clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, methods, products or equipment.

[0023] The naming or numbering of the steps in the embodiments of the present invention does not mean that the steps in the method flow must be executed in the time / logical sequence indicated by the naming or numbering. The execution order of the named or numbered process steps can be changed according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved.

[0024] Reference to an "embodiment" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0025] The present invention provides a method, device and electronic device for evaluating target fusion performance of intelligent driving, which are described below respectively.

[0026] like Figure 1 As shown, the present invention provides a method for evaluating target fusion performance of intelligent driving, comprising: S101, acquiring perception data collected by multiple vehicle-mounted sensors; S102, analyzing the perception data collected by multiple vehicle-mounted sensors to obtain readable data; S103, fusing the readable data corresponding to different vehicle-mounted sensors to obtain a fused signal of a single target in a continuous time period; S104, time-aligning the perception data collected by a single sensor corresponding to a single target with the fusion signal of the single target in a continuous time period; S105, after completing the time alignment, determining a deviation rate between the fused signal and the perception data collected by the single vehicle-mounted sensor based on the fused signal and the perception data collected by the single vehicle-mounted sensor; S106. Determine a target fusion performance evaluation result based on the deviation rate.

[0027] It is understandable that the collected data is parsed into readable text or pictures, for example, the BLF file collected by VECTOR is parsed into a CSV file, or the signal data in ROSBAG is parsed into a CSV file and pictures, etc. In order to ensure the uniqueness of the data, the parsed text data contains timestamp information, and the timestamp of each frame of data is unique.

[0028] To ensure data integrity, data analysis should include necessary information about the evaluation index. If it is necessary to evaluate the lateral and longitudinal distances, lateral and longitudinal speeds, and target types of target fusion, the analyzed data must include the lateral and longitudinal distances, lateral and longitudinal speeds, and target type information of a single sensor and the fused target.

[0029] To ensure the relevance of the data, each target instance in the fusion data should also contain the target instance of the associated sensor data. For example, the camera target B and the millimeter-wave radar target C associated with the fusion target A can be obtained from the parsed data.

[0030] To ensure the accuracy of the data, the parsed data should be sampled and checked to see if the values ​​are consistent with the original data.

[0031] In some embodiments, determining the deviation rate between the fused signal and the perception data collected by the single vehicle-mounted sensor based on the fused signal and the perception data collected by the single vehicle-mounted sensor includes: The difference between the fusion signal and the perception data collected by the single vehicle-mounted sensor is divided by the perception data collected by the single vehicle-mounted sensor to obtain the deviation rate between the fusion signal and the perception data collected by the single vehicle-mounted sensor.

[0032] It can be understood that after the above steps, the corresponding relationship between the fusion signal of a single target object and the signal of a single sensor in a continuous time period can be obtained. This is regarded as a unit, called an event. At the same time, the start event and the end event of the event can also be obtained. The video data, raw data, and parsed data in the time period can be exported in batches for subsequent retrospective analysis. The indicators to be analyzed, such as horizontal and vertical distances, horizontal and vertical speeds, target types, confidence, etc., are plotted in a time series curve according to their relationship with time, such as Figure 2The difference percentage between the fusion signal (i.e., fusion output value) and the signal collected by a single sensor (i.e., perception output value) (i.e., camera signal, millimeter-wave radar signal, etc.) is calculated using the following formula:

[0033] In some embodiments, the target fusion performance evaluation method for intelligent driving further includes: In the case where the deviation rate is greater than the target deviation threshold, the abnormal cause of the deviation is determined based on the perception data moment corresponding to the time when the deviation rate is greater than the target deviation threshold.

[0034] In some embodiments, determining a target fusion performance evaluation result based on the deviation rate includes: Among the deviation rates corresponding to multiple vehicle-mounted sensors, the deviation rates within the target deviation range are excluded to obtain the remaining deviation rate, and the target fusion performance evaluation result is determined based on the remaining deviation rate.

[0035] In some embodiments, time-aligning the perception data collected by a single sensor corresponding to a single target with the fusion signal of the single target in a continuous time period includes: The perception data collected by a single sensor corresponding to a single target is interpolated with the fusion signal of the single target in a continuous time period to achieve time alignment.

[0036] It can be understood that time alignment refers to the time alignment of the data of a single sensor of a single target and the data of target fusion. For example, the frequency of target fusion data is 10HZ, the frequency of front camera data is 33HZ, and the frequency of front millimeter wave radar data is 13HZ. If the frequency of fusion data is used for alignment, it is necessary to find the same data as the fusion data in the camera and millimeter wave radar data. Of course, most of the time, absolutely consistent data cannot be found. At this time, the camera and millimeter wave radar data need to be interpolated. Interpolation algorithms such as linear interpolation algorithm, nearest neighbor interpolation algorithm, bilinear quadratic interpolation algorithm, etc. Interpolation algorithm is not the focus of the present invention. The present invention is not limited to the difference method listed. All methods that can complete the interpolation function are included in the present invention. Determining a single target is determined based on the ID of the fusion target. For example, when the continuous frame threshold is set to 3, if the ID is the same, it is considered to be the same target. If the ID is different, it is considered to be a different target. According to the target of a single sensor associated with the fusion target, they can be interpolated in time to obtain an evaluation unit based on the fusion ID, that is, the curve, deviation and expectation of the fusion data for the single sensor data are evaluated in the same time period of the fusion ID.

[0037] In some embodiments, the deviation rate includes: the deviation rate corresponding to the lateral and longitudinal distances, the lateral and longitudinal speeds, the target type or the confidence level.

[0038] Multiple on-board sensors, including cameras, millimeter-wave radars, and lidars.

[0039] In some embodiments, the method provided by the present invention is applied to the performance evaluation of target fusion, comprising the following implementation steps: data collection, data parsing, time alignment, data analysis, and report generation, such as Figure 3 The following mainly describes the specific implementation content of each step.

[0040] Data collection: This step mainly collects data collected by various sensors (cameras, millimeter-wave radars, lidars, etc.) and perception fusion data based on the test variables covered by the test cases. It also involves relevant data such as body data, planning, control, and decision-making. The test-related variable factors are listed in Table 1. Of course, the relevant factors are not limited to those listed, but also include other factors not included in the list. Table 1 is just an example. The collected use cases will try to include situations that affect the perception scene, so that the collected data is as comprehensive and specific as possible, making the subsequent statistical conclusions more reliable.

[0041] Table 1: Test related variables

[0042] Data parsing: Parse the collected data into readable text or pictures, for example, parse the BLF file collected by VECTOR into a CSV file, or parse the signal data in ROSBAG into a CSV file and pictures, etc. In order to ensure the uniqueness of the data, the parsed text data contains timestamp information, and the timestamp of each frame of data is unique.

[0043] To ensure data integrity, data analysis should include necessary information about the evaluation index. If it is necessary to evaluate the lateral and longitudinal distances, lateral and longitudinal speeds, and target types of target fusion, the analyzed data must include the lateral and longitudinal distances, lateral and longitudinal speeds, and target type information of a single sensor and the fused target.

[0044] To ensure the relevance of the data, each target instance in the fusion data should also contain the target instance of the associated sensor data. For example, the camera target B and the millimeter-wave radar target C associated with the fusion target A can be obtained from the parsed data.

[0045] To ensure the accuracy of the data, the parsed data should be sampled and checked to see if the values ​​are consistent with the original data.

[0046] Time alignment: Time alignment refers to the time alignment of the data of a single sensor of a single target and the data of target fusion. For example, the frequency of target fusion data is 10HZ, the frequency of front camera data is 33HZ, and the frequency of front millimeter wave radar data is 13HZ. If the frequency of fusion data is used for alignment, then the data with the same time as the fusion data needs to be found in the camera and millimeter wave radar data. Of course, absolutely consistent data cannot be found in most of the time. At this time, the camera and millimeter wave radar data need to be interpolated. Interpolation algorithms such as linear interpolation algorithm, nearest neighbor interpolation algorithm, bilinear quadratic interpolation algorithm, etc. Interpolation algorithm is not the focus of the present invention. The present invention is not limited to the difference method listed. All methods that can complete the interpolation function are included in the present invention. The determination of a single target is determined based on the ID of the fusion target. For example, when the continuous frame threshold is set to 3, if the ID is the same, it is considered to be the same target. If the ID is different, it is considered to be a different target. According to the target of a single sensor associated with the fusion target, they can be interpolated in time to obtain an evaluation unit based on the fusion ID, that is, the curve, deviation and expectation of the fusion data for the single sensor data are evaluated in the same time period of the fusion ID.

[0047] Data analysis: After the above steps, we can get the corresponding relationship between the fusion signal of a single target object and the signal of a single sensor in a continuous time period. This is regarded as a unit, called an event. At the same time, we can also get the start event and end event of the event. The video data, raw data, and parsed data in this time period are exported in batches for subsequent retrospective analysis. The indicators to be analyzed, such as horizontal and vertical distance, horizontal and vertical speed, target type, confidence, etc., are plotted against time to form a time series curve, such as Figure 2 The difference percentage between the fusion signal (i.e., fusion output value) and the signal collected by a single sensor (i.e., perception output value) (i.e., camera signal, millimeter-wave radar signal, etc.) is calculated using the following formula:

[0048] Taking the fusion output value and camera output value of the longitudinal distance x as an example, the deviation rate between the fusion output value and the camera perception output value is calculated as follows:

[0049] in, represents the deviation rate between fusion and camera, represents the fusion output value, It represents the camera perception output value, other signal indicators such as lateral distance, lateral and longitudinal speed, target type, or fusion output, which is similar to the calculation formula of other sensors such as millimeter wave radar.

[0050] There are single-source sensor and multi-source sensor fusion in fusion. For example, for single-source sensor fusion, only one sensor (camera or millimeter-wave radar or other) detects the target object for the fusion input, while for multi-source sensor fusion, more than one sensor (camera and millimeter-wave radar or other) detects the target object. The fusion type is represented by perception_type. Taking the forward 1V1R (front camera and front millimeter-wave radar) fusion as an example, the fusion types involved are listed. Perception_type 1 indicates the fusion output of a single millimeter-wave radar, perception_type 2 indicates the fusion output of a single camera, and perception_type 3 indicates the fusion output of the camera and millimeter-wave radar. If it is multi-source sensor fusion (perception_type is 3), in addition to evaluating the deviation rate between the fusion and sensor indicators, the deviation rate between the camera and the millimeter-wave radar will also be evaluated. The calculation formula is similar. Taking the longitudinal distance x as an example, the formula is listed as follows:

[0051] in, Indicates the deviation rate between the camera and the millimeter-wave radar, Indicates the vertical distance output by the camera. The calculation formulas for other signal indicators such as the longitudinal distance, lateral distance, lateral and longitudinal speed, and target type output by the millimeter-wave radar are similar.

[0052] like Figure 4 It is a time series diagram of the deviation rate of perception fusion indicators. Figure 2 Analyze the differences in indicators between fusion and other sensors, and complete analysis in specific scenarios based on the time point. For example, Figure 2 From the vertical distance curve, we can see that fusion_x and radar_x are more consistent, while from the horizontal distance curve, we can see that fusion_y and camera_y are more consistent, which means that in terms of fusion strategy, the vertical distance is more biased towards radar, while the horizontal distance is more biased towards vision. This observation is consistent with the actual fusion logic. Figure 4 The abnormal point on the map (value far greater than 0) is at 10:44:09.502, and the deviation rate of the horizontal distance is greatly abnormal. Figure 2At the same time and position of the same image, it is found that the reason is that the visual or millimeter-wave radar data is close to zero at that time (that is, the denominator in the formula is close to 0), resulting in an abnormally large deviation rate. For this reason, it is necessary to set a threshold to avoid this special situation. Set the lateral distance threshold to Y, the longitudinal distance threshold to X, the lateral distance threshold to VY, and the longitudinal distance threshold to VX to exclude all deviation rates within this range. For example, if Y is set to 0.1m, X is set to 5m, VY is set to 0.1m / s, and VX is set to 5m / s, then the targets with a lateral distance between -0.1m and 0.1m, or a longitudinal distance between -5m and 5m, or a lateral speed between -0.1m / s and 0.1m / s, or a longitudinal speed between -5m / s and 5m / s should be excluded. After excluding outliers, the change in the deviation rate can be analyzed from multiple aspects that affect the fusion performance, such as the acquisition period, target distance, vehicle speed, target speed, and other aspects. In general, the deviation between vision and radar is greater than the deviation between fusion and a single sensor, which is related to the strategy followed by the fusion data, that is, the forward 1V1R fusion is roughly a compromise between vision and millimeter-wave radar. Figure 5 In the above heat map, only the three groups of deviation rates of longitudinal distance in different collection periods are analyzed. There are several points that need to be analyzed in detail. The first is that the deviation rate of fusion and vision is less than 0 as a whole, while the deviation rate of fusion and radar is greater than 0 as a whole, which means that in terms of the longitudinal distance value, millimeter wave radar < fusion < vision. The second point is that the deviation rate of fusion and radar is too large at 12 noon, which is contrary to the fact that fusion is more inclined to radar in longitudinal distance. It is necessary to further analyze the influence of other variables during this period, such as the distance and speed of the target or the speed change of the vehicle.

[0053] The above is only a quantitative analysis of the deviation rate between fusion and perception values ​​under specified influencing conditions. These examples are intended to illustrate the various distribution results produced under the defined deviation rate. Many questions can be raised from them, and some of them are the same as the actual situation and can be explained, while some are contrary to the actual situation and require more in-depth analysis.

[0054] Similarly, we can also conduct statistical analysis on the deviation rate of each event to obtain statistical data such as expectation, variance, median value, etc. within the time period, conduct influencing factor analysis on the deviation rate in different dimensions, raise questions, answer questions, and finally form conclusions.

[0055] Generate report: Through the above analysis methods, draw conclusions and form a complete performance evaluation report.

[0056] like Figure 6 As shown, the present invention also provides a target fusion performance evaluation device 600 for intelligent driving, comprising: An acquisition module 601 is used to acquire perception data collected by multiple vehicle-mounted sensors; The parsing module 602 is used to parse the perception data collected by multiple vehicle-mounted sensors to obtain readable data; A fusion module 603 is used to fuse the readable data corresponding to different vehicle-mounted sensors to obtain a fusion signal of a single target in a continuous time period; An alignment module 604 is used to time-align the perception data collected by a single sensor corresponding to a single target with the fusion signal of the single target in a continuous time period; A deviation calculation module 605 is used to determine the deviation rate between the fused signal and the perception data collected by the single vehicle-mounted sensor based on the fused signal and the perception data collected by the single vehicle-mounted sensor after completing the time alignment; The evaluation module 606 is used to determine a target fusion performance evaluation result based on the deviation rate.

[0057] The target fusion performance evaluation device for intelligent driving provided in the above embodiment can implement the technical solution described in the above intelligent driving target fusion performance evaluation method embodiment. The specific implementation principles of the above modules or units can refer to the corresponding contents in the above intelligent driving target fusion performance evaluation method embodiment, which will not be repeated here.

[0058] like Figure 7 As shown, the present invention also provides an electronic device 700. The electronic device 700 includes a processor 701, a memory 702 and a display 703. Figure 7 Only some components of the electronic device 700 are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0059] In some embodiments, the memory 702 may be an internal storage unit of the electronic device 700, such as a hard disk or memory of the electronic device 700. In other embodiments, the memory 702 may also be an external storage device of the electronic device 700, such as a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard), etc. equipped on the electronic device 700.

[0060] Furthermore, the memory 702 may include both an internal storage unit of the electronic device 700 and an external storage device. The memory 702 is used to store application software installed in the electronic device 700 and various data.

[0061] In some embodiments, the processor 701 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 702, such as the target fusion performance evaluation method for intelligent driving in the present invention.

[0062] In some embodiments, the display 703 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 703 is used to display information on the electronic device 700 and to display a visual user interface. The components 701-703 of the electronic device 700 communicate with each other via a system bus.

[0063] In some embodiments of the present invention, when the processor 701 executes the intelligent driving target fusion performance evaluation program in the memory 702, the following steps may be implemented: Acquire perception data collected by multiple vehicle-mounted sensors; Analyze the perception data collected by multiple vehicle-mounted sensors to obtain readable data; The readable data corresponding to different vehicle-mounted sensors are fused to obtain the fusion signal of a single target in a continuous time period; Temporally align the perception data collected by a single sensor corresponding to a single target with the fusion signal of the single target in a continuous time period; After completing the time alignment, determining a deviation rate between the fused signal and the perception data collected by the single onboard sensor based on the fused signal and the perception data collected by the single onboard sensor; Based on the deviation rate, a target fusion performance evaluation result is determined.

[0064] It should be understood that when the processor 701 executes the intelligent driving target fusion performance evaluation program in the memory 702, in addition to the above functions, other functions can also be implemented. For details, please refer to the description of the corresponding method embodiment above.

[0065] Furthermore, the embodiment of the present invention does not specifically limit the type of the electronic device 700 mentioned, and the electronic device 700 may be a portable electronic device such as a mobile phone, a tablet computer, a personal digital assistant (PDA), a wearable device, a laptop computer, etc. Exemplary embodiments of portable electronic devices include but are not limited to portable electronic devices equipped with IOS, Android, Microsoft or other operating systems. The above-mentioned portable electronic device may also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, the electronic device 700 may not be a portable electronic device, but a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0066] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to perform the target fusion performance evaluation method for intelligent driving provided by the above methods, the method comprising: Acquire perception data collected by multiple vehicle-mounted sensors; Analyze the perception data collected by multiple vehicle-mounted sensors to obtain readable data; The readable data corresponding to different vehicle-mounted sensors are fused to obtain the fusion signal of a single target in a continuous time period; Temporally align the perception data collected by a single sensor corresponding to a single target with the fusion signal of the single target in a continuous time period; After completing the time alignment, determining a deviation rate between the fused signal and the perception data collected by the single onboard sensor based on the fused signal and the perception data collected by the single onboard sensor; Based on the deviation rate, a target fusion performance evaluation result is determined.

[0067] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0068] The target fusion performance evaluation method, device and electronic device for intelligent driving provided by the present invention are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A target fusion performance evaluation method for intelligent driving, characterized in that: include: Acquire perception data collected by multiple vehicle-mounted sensors; Analyze the perception data collected by multiple vehicle-mounted sensors to obtain readable data; The readable data corresponding to different vehicle-mounted sensors are fused to obtain the fusion signal of a single target in a continuous time period; Temporally align the perception data collected by a single sensor corresponding to a single target with the fusion signal of the single target in a continuous time period; After completing the time alignment, determining a deviation rate between the fused signal and the perception data collected by the single onboard sensor based on the fused signal and the perception data collected by the single onboard sensor; Based on the deviation rate, a target fusion performance evaluation result is determined.

2. The target fusion performance evaluation method for intelligent driving according to claim 1 is characterized in that: Determining a deviation rate between the fused signal and the perception data collected by the single vehicle-mounted sensor based on the fused signal and the perception data collected by the single vehicle-mounted sensor includes: The difference between the fusion signal and the perception data collected by the single vehicle-mounted sensor is divided by the perception data collected by the single vehicle-mounted sensor to obtain the deviation rate between the fusion signal and the perception data collected by the single vehicle-mounted sensor.

3. The target fusion performance evaluation method for intelligent driving according to claim 1 is characterized in that: Also includes: In the case where the deviation rate is greater than the target deviation threshold, the abnormal cause of the deviation is determined based on the perception data moment corresponding to the time when the deviation rate is greater than the target deviation threshold.

4. The target fusion performance evaluation method for intelligent driving according to claim 1 is characterized in that: Based on the deviation rate, a target fusion performance evaluation result is determined, including: Among the deviation rates corresponding to multiple vehicle-mounted sensors, the deviation rates within the target deviation range are excluded to obtain the remaining deviation rate, and the target fusion performance evaluation result is determined based on the remaining deviation rate.

5. The target fusion performance evaluation method for intelligent driving according to claim 1 is characterized in that: The perception data collected by a single sensor corresponding to a single target is time-aligned with the fusion signal of a single target in a continuous time period, including: The perception data collected by a single sensor corresponding to a single target is interpolated with the fusion signal of the single target in a continuous time period to achieve time alignment.

6. The target fusion performance evaluation method for intelligent driving according to claim 1 is characterized in that: The deviation rate includes: the deviation rate corresponding to the horizontal and vertical distance, the horizontal and vertical speed, the target type or the confidence level.

7. The target fusion performance evaluation method for intelligent driving according to any one of claims 1 to 6, characterized in that: Multiple on-board sensors, including cameras, millimeter-wave radars, and lidars.

8. A target fusion performance evaluation device for intelligent driving, characterized in that: include: An acquisition module, used to acquire perception data collected by multiple vehicle-mounted sensors; The parsing module is used to parse the perception data collected by multiple vehicle-mounted sensors to obtain readable data; A fusion module is used to fuse the readable data corresponding to different vehicle-mounted sensors to obtain a fusion signal of a single target in a continuous time period; An alignment module is used to time-align the perception data collected by a single sensor corresponding to a single target with the fusion signal of the single target in a continuous time period; a deviation calculation module, configured to determine, after completing the time alignment, a deviation rate between the fused signal and the perception data collected by the single onboard sensor based on the fused signal and the perception data collected by the single onboard sensor; An evaluation module is used to determine a target fusion performance evaluation result based on the deviation rate.

9. An electronic device, characterized in that: comprising a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps of the target fusion performance evaluation method for intelligent driving as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the target fusion performance evaluation method for intelligent driving as described in any one of claims 1 to 7 are implemented.

Images