Method and apparatus for judging reliability of visual recognition

Abstract

The present application relates to a kind of judging methods of visual recognition reliability, the method includes: receiving lane line parameter of visual recognition;By comparing historical lane line parameter with the received lane line parameter, the historical shrinkage degree of lane line is calculated;And whether visual recognition has error is determined based on the historical shrinkage degree.The present application also relates to a kind of judging equipment of visual recognition reliability, computer storage medium, autonomous driving system and vehicle.

Description

Technical Field

[0001] This invention relates to the field of autonomous / assisted driving control, and more specifically, to a method and device for judging the reliability of visual recognition, a computer storage medium, an autonomous driving system, and a vehicle. Background Technology

[0002] Many environmental factors can affect the accuracy of vehicle visual detection, including: sudden strong sunlight; wet ground; and lane markings that have not been completely erased. These factors can affect the quality of lane markings detected by vision, potentially leading to dangerous behaviors such as lane departure and erratic steering while maintaining lane position, thus causing safety issues. Summary of the Invention

[0003] According to one aspect of the present invention, a method for determining the reliability of visual recognition is provided, the method comprising: receiving lane line parameters for visual recognition; calculating the historical contraction degree of lane lines by comparing historical lane line parameters with the received lane line parameters; and determining whether there is an error in visual recognition based on the historical contraction degree.

[0004] As a supplement or replacement to the above scheme, in the above method, the lane line parameters received by visual recognition include: the left lane line and the right lane line received by visual recognition in this frame.

[0005] As a supplement or replacement to the above scheme, in the above method, calculating the historical contraction degree of the lane line by comparing the historical lane line parameters with the received lane line parameters includes: calculating the inward contraction value of the lane line in two consecutive frames; determining whether the inward contraction value reaches a first threshold; and when the inward contraction value is greater than or equal to the first threshold, accumulating the inward contraction value to the historical contraction degree of the lane line.

[0006] As a supplement or replacement to the above scheme, in the above method, calculating the historical contraction degree of the lane line by comparing the historical lane line parameters with the received lane line parameters further includes: when the inward contraction value is less than the first threshold, the historical contraction degree of the lane line is reset to zero.

[0007] As a supplement or replacement to the above scheme, in the above method, the first threshold is determined based on the difference in lateral distance between the beginning and end of the left and right lane lines in the current frame.

[0008] As a supplement or replacement to the above scheme, in the above method, calculating the inward contraction value of the lane line between two consecutive frames includes: updating the recorded lane line parameters of the previous frame according to the vehicle's motion parameters; comparing the updated lane line of the previous frame with the corresponding lane line of the current frame; and calculating the relative deflection trend and magnitude of the lane line ends.

[0009] As a supplement or replacement to the above scheme, in the above method, the relative deflection trend and magnitude of the lane line end is the angle of inward rotation of the lane line in the previous frame compared to the lane line corresponding to the current frame.

[0010] As a supplement or replacement to the above scheme, in the above method, determining whether there is an error in visual recognition based on the historical shrinkage degree includes: comparing the historical shrinkage degree with a second threshold; and determining that there is an error in visual recognition when the historical shrinkage degree is greater than or equal to the second threshold.

[0011] As a supplement or replacement to the above scheme, in the above method, determining whether there is an error in visual recognition based on the historical shrinkage degree further includes: when the historical shrinkage degree is less than the second threshold, determining that there is no visual recognition error, and recording the historical shrinkage degree.

[0012] As a supplement or replacement to the above scheme, the above method also includes: when it is determined that there is an error in visual recognition, outputting a flag bit to remind other modules.

[0013] According to another aspect of the present invention, a device for judging the reliability of visual recognition is provided, the device comprising: a receiving device for receiving lane line parameters of visual recognition; a calculating device for calculating the historical shrinkage degree of lane lines by comparing historical lane line parameters with the received lane line parameters; and a determining device for determining whether there is an error in visual recognition based on the historical shrinkage degree.

[0014] As a supplement or replacement to the above solution, in the above device, the receiving device is configured to receive the left lane line and right lane line of the current frame visual recognition.

[0015] As a supplement or replacement to the above solution, in the above device, the computing device includes: a computing unit for calculating the inward contraction value of the lane line in two consecutive frames; a judging unit for judging whether the inward contraction value reaches a first threshold; and an accumulation unit for accumulating the inward contraction value to the historical contraction degree of the lane line when the inward contraction value is greater than or equal to the first threshold.

[0016] As a supplement or replacement to the above solution, the computing device in the above device further includes: a zeroing device, used to zero out the historical contraction degree of the lane line when the inward contraction value is less than the first threshold.

[0017] As a supplement or replacement to the above scheme, in the above device, the first threshold is determined based on the difference in lateral distance between the beginning and end of the left and right lane lines in this frame.

[0018] As a supplement or replacement to the above solution, in the above device, the calculation unit is configured to: update the recorded lane line parameters of the previous frame according to the vehicle motion parameters; compare the updated lane line of the previous frame with the lane line corresponding to the current frame; and calculate the relative deflection trend and magnitude of the lane line ends.

[0019] As a supplement or replacement to the above scheme, in the above device, the relative deflection trend and magnitude of the lane line end is the angle of inward rotation of the lane line in the previous frame compared to the lane line corresponding to the current frame.

[0020] As a supplement or alternative to the above solution, in the above device, the determining device is configured to compare the historical shrinkage degree with a second threshold; and when the historical shrinkage degree is greater than or equal to the second threshold, to determine that there is an error in visual recognition.

[0021] As a supplement or replacement to the above solution, in the above device, the determining device is further configured to determine that there is no visual recognition error when the historical shrinkage degree is less than the second threshold, and to record the historical shrinkage degree.

[0022] As a supplement or replacement to the above solution, the above device also includes: an output device for outputting a flag bit to alert other modules when it is determined that there is an error in visual recognition.

[0023] According to another aspect of the present invention, a computer storage medium is provided, the medium including instructions that, when executed, perform the method described above.

[0024] According to another aspect of the present invention, an autonomous driving system is provided, the system comprising the devices as described above.

[0025] According to another aspect of the invention, a vehicle is provided, the vehicle including the autonomous driving system as described above.

[0026] The visual recognition reliability judgment scheme of the present invention judges the reliability of visual recognition lane lines by comparing historical lane line parameters with received lane line parameters (e.g., tracking the change state of lane lines in multiple frames before and after) and calculating the historical shrinkage degree of lane lines, thereby solving the problem of visual recognition errors caused by external environmental influences. Attached Figure Description

[0027] The above and other objects and advantages of the present invention will become more fully clear from the following detailed description taken in conjunction with the accompanying drawings, wherein the same or similar elements are indicated by the same reference numerals.

[0028] Figure 1A flowchart illustrating a method for determining the reliability of visual recognition according to an embodiment of the present invention is shown.

[0029] Figure 2 A schematic diagram of a visual recognition reliability determination device according to an embodiment of the present invention is shown; and

[0030] Figure 3 A flowchart of lane quality tracking according to an embodiment of the present invention is shown. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it.

[0032] It should also be noted that, for ease of description, the accompanying drawings show only the parts relevant to the invention and not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. The process may correspond to a method, function, procedure, subroutine, subprogram, etc.

[0033] Although exemplary embodiments are described as using multiple units to perform exemplary processes, it should be understood that these exemplary processes may also be performed by one or more modules.

[0034] Furthermore, the control logic of the present invention can be included as executable program instructions on a computer-readable medium, which are implemented by a processor or the like. Examples of computer-readable media include, but are not limited to, ROM, RAM, optical discs, magnetic tapes, floppy disks, flash drives, smart cards, and optical data storage devices. The computer-readable recording medium can also be distributed across a networked computer system, enabling distributed storage and implementation of the computer-readable medium, for example, via in-vehicle telecommunications services or a Controller Area Network (CAN).

[0035] Unless specifically mentioned or obvious from the context, the term “approximately” as used herein shall be understood as being within the range of normal tolerances in the art, such as within 2 standard deviations of the mean.

[0036] It should be understood that the term "vehicle" or other similar terms used herein include motor vehicles in general, such as passenger cars (including SUVs, buses, trucks, etc.), various commercial vehicles, etc., and includes hybrid vehicles, electric vehicles, etc. A hybrid vehicle is a vehicle with two or more power sources, such as a gasoline-powered and an electric vehicle.

[0037] In the following, a scheme for determining the reliability of visual recognition according to various exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0038] Figure 1 A flowchart illustrating a method 1000 for determining the reliability of visual recognition according to an embodiment of the present invention is shown. Figure 1 As shown, the method 1000 for judging the reliability of visual recognition includes the following steps:

[0039] In step S110, lane line parameters recognized by vision are received;

[0040] In step S120, the historical lane line contraction degree is calculated by comparing the historical lane line parameters with the received lane line parameters; and

[0041] In step S130, it is determined whether there is an error in visual recognition based on the historical degree of shrinkage.

[0042] In the context of this invention, "visually recognized lane line parameters" refers to lane line information identified by various visual sensors (e.g., cameras, ultrasonic radar, etc.). Visual recognition, as an important source of lane line information, is often used to construct road models.

[0043] In one embodiment, the term "historical lane line parameters" refers to the lane line parameters received in the previous frame. Since visually recognized lane lines are updated in real time every frame, the historical degree of lane line contraction can be calculated by comparing the differences between the lane line parameters received in the previous frame and the lane line parameters in the current frame.

[0044] In the context of this invention, "historical contraction degree" refers to the tendency or extent to which lane lines, as detected by the vehicle's visual sensors, contract inward over a period of time. In one embodiment, this historical contraction degree is equal to the sum of inward contraction values ​​(under certain conditions) across multiple frames. The inward contraction value represents the relative (inward) deflection of the lane lines between consecutive frames. 。 In one or more embodiments, the inward contraction value is characterized by the angle at which the lane line ends deflect inward between two consecutive frames.

[0045] Environmental factors have a significant impact on visual recognition and may pose considerable risks to vehicle movement. Through big data analysis and scenario summarization, the inventors of this application discovered that in scenarios such as sudden strong sunlight, wet ground, and lane markings that have not been completely erased, both left and right lane markings tend to contract inwards. Therefore, by tracking and judging whether lane markings contract inwards across multiple frames (i.e., historical contraction levels), the influence of the external environment on visual recognition can be reflected. If the historical contraction level reaches a certain threshold, it indicates that the confidence level of the visual recognition result for that frame is low (visual recognition error exists). In this case, stronger filtering can be designed in subsequent target selection and guidance to reduce the adverse effects that inaccurate lane marking parameters in that frame may cause.

[0046] In one embodiment, step S110 includes receiving the left lane line and right lane line visually recognized in the current frame. In one embodiment, the visually recognized lane lines are updated in real time every frame. Due to visual distance limitations and the complexity of actual scenarios, if the lane line parameters identified in multiple frames are filtered using filtering or other methods, a large amount of real-time lane line change information will be lost. The filtered lane line parameters will differ significantly from the lane lines detected in real time, thus posing a greater risk. Therefore, it is necessary to update the lane line parameters based on visual recognition every frame. Considering that directly adopting the latest parameter information while losing historical information would be detrimental to handling scenarios with visual recognition errors, the change status of the lane lines in multiple frames is tracked.

[0047] It should be noted that, although in the context of this invention, the left and right lane lines (i.e., two lane lines) identified in the current frame are used as the basis for tracking lane lines and thus measuring the confidence level of visual recognition, in one embodiment, in addition to the left and right lane lines identified in the current frame, the lane lines of adjacent lanes in the current frame can also be received as the basis for tracking lane lines. This further avoids false alarms by understanding the positional relationship between multiple lane lines.

[0048] In one embodiment, step S120 includes: calculating the inward contraction value of the lane line in two consecutive frames; determining whether the inward contraction value reaches a first threshold; and when the inward contraction value is greater than or equal to the first threshold, accumulating the inward contraction value to the historical contraction level of the lane line. When the inward contraction value is less than the first threshold, resetting the historical contraction level of the lane line to zero.

[0049] In the above embodiments, the first threshold is determined based on the difference in lateral distance between the start and end points of the left and right lane lines in the current frame. For example, within this frame, the difference in lateral distance between the start points of the left and right lane lines is d1, the difference in lateral distance between the end points of the left and right lane lines is d2, and the distance from the start point to the end point is s. In one embodiment, the threshold can be... Of course, the above is just an example of a first threshold, and those skilled in the art can adjust the threshold as needed.

[0050] In one embodiment, calculating the inward contraction value of lane lines between two consecutive frames includes: updating the recorded lane line parameters of the previous frame based on the vehicle's motion parameters; comparing the updated lane lines of the previous frame with the corresponding lane lines of the current frame; and calculating the relative deflection trend and magnitude of the lane line ends. Specifically, the lane lines of the previous frame can be updated based on VPE (Vehicle Position Estimate). For example, the lane lines of the previous frame are adaptively updated based on the vehicle's lateral / longitudinal turning angle deviation. After updating the lane lines of the previous frame based on the vehicle's displacement, they can be compared with the corresponding lane lines of the current frame to calculate the relative deflection trend and magnitude of the lane line ends. In one embodiment, the relative deflection trend and magnitude of the lane line ends is the angle of inward rotation of the lane lines of the previous frame compared to the corresponding lane lines of the current frame.

[0051] In one embodiment, step S130 includes: comparing the historical shrinkage degree with a second threshold; and determining that a visual recognition error exists when the historical shrinkage degree is greater than or equal to the second threshold. When the historical shrinkage degree is less than the second threshold, determining that no visual recognition error exists, and recording the historical shrinkage degree.

[0052] In this embodiment, the second threshold can be determined based on statistical judgment. In one embodiment, the design and selection of the second threshold needs to balance the ratio of true positive rate to false positive rate.

[0053] although Figure 1 As not shown in the figure, in one embodiment, the above method 1000 further includes: when it is determined that there is an error in visual recognition, outputting a flag bit to alert other modules.

[0054] The visual recognition reliability judgment method of one or more embodiments of the present invention tracks the change state of lane lines in multiple frames before and after the lane lines, and combines the relative positional relationship of the left and right lane lines in the same frame to judge the reliability of visual recognition of lane lines, thereby solving the problem of visual recognition errors caused by external environmental influences and avoiding false alarms.

[0055] Furthermore, those skilled in the art will readily understand that the visual recognition reliability determination method provided in one or more embodiments of the present invention can be implemented by a computer program. For example, when a computer storage medium (e.g., a USB flash drive) containing the computer program is connected to a computer, running the computer program will execute the visual recognition reliability determination method of one or more embodiments of the present invention.

[0056] refer to Figure 2 , Figure 2 A schematic diagram of a visual recognition reliability determination device 2000 according to an embodiment of the present invention is shown. Figure 2 As shown, the visual recognition reliability assessment device 2000 includes a receiving device 210, a computing device 220, and a determining device 230. The receiving device 210 receives lane line parameters from visual recognition; the computing device 220 calculates the historical contraction degree of the lane lines by comparing historical lane line parameters with the received lane line parameters; and the determining device 230 determines whether there is an error in the visual recognition based on the historical contraction degree.

[0057] In the context of this invention, "visually recognized lane line parameters" refers to lane line information identified by various visual sensors (e.g., cameras, ultrasonic radar, etc.). Visual recognition, as an important source of lane line information, is often used to construct road models.

[0058] In one embodiment, the term "historical lane line parameters" refers to the lane line parameters received in the previous frame. Since visually recognized lane lines are updated in real time every frame, the historical degree of lane line contraction can be calculated by comparing the differences between the lane line parameters received in the previous frame and the lane line parameters in the current frame.

[0059] In the context of this invention, "historical contraction degree" refers to the tendency or extent to which lane lines, as detected by the vehicle's visual sensors, contract inward over a period of time. In one embodiment, this historical contraction degree is equal to the sum of inward contraction values ​​(under certain conditions) across multiple frames. The inward contraction value represents the relative (inward) deflection of the lane lines between consecutive frames. 。 In one or more embodiments, the inward contraction value is characterized by the angle at which the lane line ends deflect inward between two consecutive frames.

[0060] In one embodiment, the receiving device 210 is configured to receive the left and right lane lines identified in the current visual frame. It should be noted that, although in the context of this invention, the left and right lane lines (i.e., the two lane lines) identified in the current visual frame are used as the basis for tracking lane lines and thus measuring the confidence level of the visual recognition, in one embodiment, in addition to the left and right lane lines identified in the current visual frame, the receiving device 210 is also configured to receive the lane lines of adjacent lanes in the current frame as the basis for tracking lane lines. This further avoids false alarms by understanding the positional relationships between multiple lane lines.

[0061] In one embodiment, the computing device 220 includes: a computing unit for calculating the inward contraction value of the lane line in two consecutive frames; a judging unit for judging whether the inward contraction value reaches a first threshold; and an accumulation unit for accumulating the inward contraction value to the historical contraction degree of the lane line when the inward contraction value is greater than or equal to the first threshold. In one embodiment, the computing device 220 further includes: a zeroing device for zeroing the historical contraction degree of the lane line when the inward contraction value is less than the first threshold.

[0062] In the above embodiments, the first threshold is determined based on the difference in lateral distance between the start and end points of the left and right lane lines in the current frame. For example, within this frame, the difference in lateral distance between the start points of the left and right lane lines is d1, the difference in lateral distance between the end points of the left and right lane lines is d2, and the distance from the start point to the end point is s. In one embodiment, the threshold can be... Of course, the above is just an example of a first threshold, and those skilled in the art can adjust the threshold as needed.

[0063] In one embodiment, the calculation unit is configured to: update the recorded lane line parameters of the previous frame according to the vehicle's motion parameters; compare the updated lane lines of the previous frame with the corresponding lane lines of the current frame; and calculate the relative deflection trend and magnitude of the lane line ends. Specifically, the calculation unit may be configured to: update the lane lines of the previous frame according to VPE (Vehicle Position Estimate). For example, the lane lines of the previous frame are adaptively updated based on the vehicle's lateral / longitudinal turning angle deviation. After updating the lane lines of the previous frame according to the vehicle's displacement, the calculation unit can compare them with the corresponding lane lines of the current frame to calculate the relative deflection trend and magnitude of the lane line ends. In one embodiment, the relative deflection trend and magnitude of the lane line ends is the angle of inward rotation of the lane lines of the previous frame compared to the lane lines of the current frame.

[0064] In one embodiment, the determining device 230 is configured to compare the historical shrinkage level with a second threshold; and to determine that a visual recognition error exists when the historical shrinkage level is greater than or equal to the second threshold. In another embodiment, the determining device 230 is further configured to determine that no visual recognition error exists when the historical shrinkage level is less than the second threshold, and to record the historical shrinkage level. In this embodiment, the second threshold may be determined based on statistical judgment. In one embodiment, the design and selection of the second threshold needs to balance the ratio of true positive rate to false positive rate.

[0065] although Figure 2 As not shown in the diagram, in one embodiment, the device 2000 may further include: an output device for outputting a flag bit to alert other modules when it is determined that an error exists in visual recognition.

[0066] In one or more embodiments, the aforementioned visual recognition reliability assessment device 2000 can be integrated into an autonomous driving system or an advanced driver assistance system (ADAS). An ADAS system, also known as an advanced driver assistance system, utilizes various sensors installed on the vehicle (e.g., millimeter-wave radar, lidar, monocular / dual-lens cameras, and satellite navigation) to continuously sense the surrounding environment while the vehicle is in motion, collect data, identify, detect, and track static and dynamic objects, and combine this data with navigation map data for system calculations and analysis. This allows the driver to anticipate potential dangers, effectively increasing driving comfort and safety. In one embodiment, the advanced driver assistance system includes a navigation and real-time traffic system (TMC), an electronic speed enforcement system (ISA, or intelligent speed advice), vehicular communication systems, adaptive cruise control (ACC), lane departure warning system (LDWS), lane keep assistance system, collision avoidance system (or pre-collision system), night vision system, adaptive light control, pedestrian protection system, automatic parking system, traffic sign recognition, blind spot detection, driver fatigue detection, hill descent control, and electric vehicle warning sounds system.

[0067] Further reference Figure 3 , Figure 3 A lane quality tracking flowchart 3000 according to an embodiment of the present invention is shown. Figure 3As shown, in step 310, the left and right lane lines of the current frame are obtained through visual sensor recognition. After obtaining the original visual recognition signal, the lane line parameters of the previous frame recorded need to be updated according to the vehicle's motion parameters. Then, compared with the corresponding lane line in the current frame, the relative deflection trend and magnitude of the lane line ends are calculated as the increment of the historical contraction degree of this lane line. In step 320, the inward contraction value of the lane lines in the preceding and following frames is calculated. Next, in step 330, it is determined whether the relative contraction of the left and right lane lines reaches a threshold, which can be obtained by comparing the difference in lateral distance between the start and end of the left and right lane lines. Based on the determination in step 330, if the threshold is not reached, the inward contraction value of the lane lines is cleared to zero in step 340. If the threshold is reached, the lane line contraction degree is accumulated in step 350. In step 360, it is further determined whether the contraction of the preceding and following frames reaches another threshold. If the historical contraction degree reaches the specified threshold, it is determined in step 370 that there is an error in the visual recognition. If the historical shrinkage level does not reach the specified threshold, then in step 380, the shrinkage value is accumulated to the next frame, and the process returns to step 310 to begin a new round of visual recognition. Additionally, after determining that there is an error in the visual recognition, in step 390, a flag is output indicating that the confidence level of the visually recognized lane lines is relatively low. This allows for stronger filtering in subsequent target selection and guidance to mitigate the potential impact of inaccurate lane line parameters in this frame.

[0068] The above embodiments track the lane line performance across multiple frames to determine if there are situations with low confidence in the lane line, and distinguish them from scenarios where the actual lane line changes, thus avoiding false alarms.

[0069] In summary, the visual recognition reliability judgment scheme of the embodiments of the present invention judges the reliability of visual recognition lane lines by comparing historical lane line parameters with received lane line parameters (for example, tracking the change state of lane lines in multiple frames before and after) and calculating the historical contraction degree of lane lines, thereby solving the problem of visual recognition errors caused by external environmental influences.

[0070] Although the foregoing specification describes only some embodiments of the invention, those skilled in the art will understand that the invention can be implemented in many other forms without departing from its spirit and scope. Therefore, the examples and embodiments shown are to be considered illustrative rather than restrictive, and the invention may encompass various modifications and substitutions without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. A method for judging the reliability of visual recognition, characterized in that, The method includes: Receive lane line parameters from visual recognition; The historical lane line contraction degree is calculated by comparing historical lane line parameters with received lane line parameters. This historical contraction degree is equal to the sum of inward contraction values ​​across multiple frames, where the inward contraction value is characterized by the inward deflection angle of the lane line ends between consecutive frames. The presence of errors in visual recognition is determined based on the historical degree of shrinkage, wherein the reliability of visual recognition is low when the historical degree of shrinkage reaches a certain threshold.

2. The method as described in claim 1, wherein, Lane line parameters received from visual recognition include: Receive the left lane line and right lane line from the visual recognition of this frame.

3. The method as described in claim 1, wherein, The historical lane line contraction is calculated by comparing historical lane line parameters with the received lane line parameters, including: Calculate the inward contraction value of the lane lines in the two consecutive frames; Determine whether the inward contraction value reaches a first threshold; and When the inward contraction value is greater than or equal to the first threshold, the inward contraction value is accumulated to the historical contraction level of the lane line.

4. The method of claim 3, wherein, Calculating the historical lane line contraction by comparing historical lane line parameters with the received lane line parameters also includes: When the inward contraction value is less than the first threshold, the historical contraction degree of the lane line is reset to zero.

5. The method as described in claim 3 or 4, wherein, The first threshold is determined based on the difference in lateral distance between the beginning and end of the left and right lane lines in this frame.

6. The method of claim 3, wherein, The calculation of the lane line inward contraction value between two consecutive frames includes: Update the lane line parameters from the previous frame based on the vehicle's motion parameters; Compare the updated lane lines from the previous frame with the corresponding lane lines in the current frame; and Calculate the relative deflection trend and magnitude at the end of the lane line.

7. The method of claim 6, wherein, The relative deflection trend and magnitude of the lane line ends are the inward rotation angle of the lane line in the previous frame compared to the corresponding lane line in the current frame.

8. The method of claim 1, wherein, Determining whether visual recognition is flawed based on the historical degree of shrinkage includes: Compare the historical degree of contraction with a second threshold; and When the historical shrinkage degree is greater than or equal to the second threshold, it is determined that there is an error in visual recognition.

9. The method of claim 8, wherein, Determining whether visual recognition is flawed based on the historical degree of shrinkage also includes: When the historical shrinkage level is less than the second threshold, it is determined that there is no visual recognition error, and the historical shrinkage level is recorded.

10. The method of claim 1, further comprising: When an error is detected in the visual recognition, a flag is output to alert other modules.

11. A device for judging the reliability of visual recognition, characterized in that, The device includes: A receiving device for receiving lane line parameters for visual recognition; A computing device is configured to calculate the historical contraction degree of lane lines by comparing historical lane line parameters with received lane line parameters, wherein the historical contraction degree is equal to the sum of inward contraction values ​​over multiple frames, wherein the inward contraction value is characterized by the angle of inward deflection of the lane line ends between consecutive frames; and A determining device is used to determine whether there is an error in visual recognition based on the historical shrinkage degree, wherein the reliability of visual recognition is low when the historical shrinkage degree reaches a certain threshold.

12. The device as claimed in claim 11, wherein, The receiving device is configured to receive the left lane line and right lane line visually recognized in this frame.

13. The device as claimed in claim 11, wherein, The computing device includes: The calculation unit is used to calculate the inward contraction value of the lane lines in two consecutive frames; The judgment unit is used to determine whether the inward contraction value reaches a first threshold; and The accumulation unit is used to accumulate the inward contraction value to the historical contraction level of the lane line when the inward contraction value is greater than or equal to the first threshold.

14. The device as claimed in claim 13, wherein, The computing device further includes: A zeroing device is used to zero out the historical contraction degree of the lane line when the inward contraction value is less than the first threshold.

15. The device as claimed in claim 13 or 14, wherein, The first threshold is determined based on the difference in lateral distance between the beginning and end of the left and right lane lines in this frame.

16. The device as claimed in claim 13, wherein, The calculation unit is configured to: update the lane line parameters of the previous frame based on the vehicle's motion parameters; compare the updated lane line of the previous frame with the lane line of the current frame; and calculate the relative deflection trend and magnitude of the lane line ends.

17. The device as claimed in claim 16, wherein, The relative deflection trend and magnitude of the lane line ends are the inward rotation angle of the lane line in the previous frame compared to the corresponding lane line in the current frame.

18. The device as claimed in claim 11, wherein, The determining device is configured to compare the historical degree of shrinkage with a second threshold; and to determine that there is an error in visual recognition when the historical degree of shrinkage is greater than or equal to the second threshold.

19. The device as claimed in claim 18, wherein, The determining device is further configured to determine that there is no visual recognition error when the historical shrinkage degree is less than the second threshold, and to record the historical shrinkage degree.

20. The apparatus of claim 11, further comprising: An output device is used to output a flag to alert other modules when an error is determined to exist in visual recognition.

21. A computer storage medium, characterized in that, The medium includes instructions that, when executed, perform the method as described in any one of claims 1 to 10.

22. An autonomous driving system, characterized in that, The system includes the device as described in any one of claims 11 to 20.

23. A vehicle, characterized in that, The vehicle includes the autonomous driving system as described in claim 22.

Images