Verification method for traffic light marking, method and device for generating driving map

Abstract

The present application relates to the technical field of object annotation, and discloses a verification method for traffic light annotation, a method and device for generating a driving map. The method includes: acquiring and displaying road point cloud data and traffic light point cloud data, wherein the road point cloud data represents the road area, and the traffic light point cloud data represents the traffic light area identified in the road area; acquiring result information of annotating the traffic light area, wherein the result information includes the coordinates of the center point of the traffic light area; if the coordinates of the center point of the traffic light area are located in the traffic light area and are located at the center of the traffic light area, then the coordinates of the center point of the traffic light area are displayed in the traffic light area, thereby determining that there is no abnormality in the annotation of the traffic light area. In the above manner, the accuracy of traffic light annotation can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of object annotation, and in particular to a verification method for traffic light annotation, and a method and device for generating a driving map. Background Art

[0002] In the fields of autonomous driving and high-precision maps, crowdsourcing map collection and construction with high update frequency, small calculation amount and fast transmission is becoming more and more popular. The on-board sensors used for map collection generally include LiDAR, cameras, RTK GPS, etc. LiDAR and cameras can sense the surrounding environment of the vehicle and provide positioning functions. RTK GPS can provide millimeter positioning accuracy when the signal is good. In addition to positioning information, the information of traffic elements such as roads, traffic signs, obstacles, pedestrians, etc. obtained by perception will be used to control the steering and speed of the vehicle. Among traffic signs, traffic lights, such as traffic lights, the timeliness and accuracy of traffic light recognition directly affect the safety and reliability of autonomous driving vehicles.

[0003] Therefore, the three-dimensional information and coordinates of traffic lights need to be accurately marked, and how to verify the accuracy of the marking is the most important step in the marking work. Summary of the invention

[0004] The present application provides a verification method for traffic light labeling, a method and device for generating a driving map, which can improve the accuracy of traffic light labeling.

[0005] To solve the above technical problems, a technical solution adopted in the present application is: to provide a verification method for traffic light labeling, the method comprising: acquiring and displaying road point cloud data and traffic light point cloud data, wherein the road point cloud data represents the road area, and the traffic light point cloud data represents the traffic light area identified in the road area; acquiring result information of labeling the traffic light area, wherein the result information includes the coordinates of the center point of the traffic light area; if the coordinates of the center point of the traffic light area are located in the traffic light area and at the center of the traffic light area, then displaying the coordinates of the center point of the traffic light area in the traffic light area, thereby determining that there is no abnormality in the labeling of the traffic light area.

[0006] Among them, it further includes: obtaining a road stitching image about the road area; comparing the road stitching image with the traffic light point cloud data to determine whether the traffic light area represents a traffic light; when the traffic light area does not represent a traffic light, deleting the center point coordinates of the traffic light area; when the traffic light area represents a traffic light, retaining the center point coordinates of the traffic light area, and displaying the traffic light number represented by the traffic light area in the road area.

[0007] The road point cloud data is displayed in a first color, the traffic light point cloud data is displayed in a second color, and the center point coordinates of the traffic light area are displayed in a third color, wherein the first color, the second color and the third color are different from each other.

[0008] Among them, obtaining a road stitching image about the road area includes: obtaining at least two frames of initial road images about the road area, wherein each of the at least two frames of initial road images includes a traffic light area; and stitching the at least two frames of initial road images to obtain a road stitching image.

[0009] The road area is a straight-ahead area; at least two frames of initial road images are obtained by photographing a photographing device located in the vehicle at a preset frequency, wherein the preset frequency is positively correlated with the driving speed of the vehicle in the straight-ahead area.

[0010] The at least two frames of initial road images include a first initial road image, and the first initial road image is captured when the vehicle has a driving speed of 0 in a straight-ahead area.

[0011] The road area is an intersection area; at least two frames of initial road images are obtained by shooting with a preset frequency by a shooting device located in the vehicle, wherein the preset frequency is positively correlated with the turning speed of the vehicle in the intersection area.

[0012] To solve the above technical problems, another technical solution adopted in the present application is: to provide a method for generating a driving map, the method comprising: executing a traffic light annotation process on the original map to obtain a traffic light annotation result; executing a verification process on the traffic light annotation result to generate a driving map; wherein the verification process includes a verification method for traffic light annotation as provided in the above technical solution.

[0013] In order to solve the above-mentioned technical problems, another technical solution adopted in the present application is: to provide a vehicle-mounted device, which includes a processor and a memory and a display coupled to the processor, the memory is used to store a computer program, and the processor is used to execute the computer program to implement the verification method for traffic light marking as provided in the above-mentioned technical solution or the method for generating a driving map as provided in the above-mentioned technical solution.

[0014] In order to solve the above technical problems, another technical solution adopted in the present application is: providing a non-volatile computer-readable storage medium, which is used to store a computer program. When the computer program is executed by a processor, it is used to implement the verification method for traffic light labeling as provided in the above technical solution or the method for generating a driving map as provided in the above technical solution.

[0015] The beneficial effect of the embodiment of the present application is that, different from the prior art, the verification method for traffic light annotation, the method and device for generating a driving map provided by the present application. The verification method uses the displayed road point cloud data and traffic light point cloud data, and when the center point coordinates of the traffic light area are located in the traffic light area and located at the center of the traffic light area, the center point coordinates of the traffic light area are displayed in the traffic light area, thereby determining that there is no abnormality in the annotation of the traffic light area, realizing the visual verification of the traffic light annotation, and the accuracy of the traffic light annotation can be improved through the visual verification. Further, the accuracy of the driving map can be improved by generating a driving map using the verification method of the traffic light annotation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, 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 application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0017] Figure 1 It is a flow chart of an embodiment of a verification method for traffic light labeling provided by the present application;

[0018] Figure 2 This is a schematic diagram of an application scenario of a verification method for traffic light labeling provided by the present application;

[0019] Figure 3 is a flow chart of another embodiment of a verification method for traffic light labeling provided by the present application;

[0020] Figure 4 It is a flowchart of an embodiment of step 31 provided by the present application;

[0021] Figure 5 It is a flowchart of an embodiment of a method for generating a driving map provided by the present application;

[0022] Figure 6 It is a structural schematic diagram of an embodiment of the vehicle-mounted device provided by the present application;

[0023] Figure 7 It is a structural schematic diagram of an embodiment of a non-volatile computer-readable storage medium provided by the present application. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It will be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only some but not all structures related to the present application are shown in the drawings. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0025] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations 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.

[0026] See also Figure 1 , Figure 1 1 is a flow chart of an embodiment of a verification method for traffic light labeling provided by the present application. The method includes:

[0027] Step 11: Acquire and display road point cloud data and traffic light point cloud data, wherein the road point cloud data represents the road area, and the traffic light point cloud data represents the traffic light area identified in the road area.

[0028] In some embodiments, corresponding technical solutions may be used to determine the road point cloud data and the traffic light point cloud data. For example, the traffic light point cloud data may be determined in the following manner.

[0029] Obtain a road image and an initial point cloud corresponding to the road area. The road image can be collected using a camera, and the initial point cloud can be collected using a radar sensor. For example, the camera and the radar sensor are installed on a movable device. The device can be an automatic mobile device, such as a robot, an autonomous driving vehicle, etc. The camera can be a camera. The radar sensor can be a laser radar sensor, such as a mechanical laser radar.

[0030] In one application scenario, an autonomous driving vehicle is traveling on a road, and a road image is acquired by a shooting device installed on the autonomous driving vehicle, and a corresponding initial point cloud is acquired by a radar sensor.

[0031] Then, the traffic light area in the road image is identified.

[0032] During the driving process of the autonomous vehicle, the acquired road images usually include roads, traffic lights, trees around the road, street lights, buildings, etc. Therefore, the traffic light area in the road image is identified through image recognition technology. For example, the traffic light area in the road image is identified through the trained recognition model.

[0033] After the traffic light area in the road image is identified, the traffic light area may be framed using a rectangular frame.

[0034] Then, a first target point cloud area corresponding to the traffic light area in the initial point cloud is determined.

[0035] In some embodiments, because the initial point cloud has three-dimensional information and the road image is two-dimensional, it is necessary to project the initial point cloud to the plane where the road image is located, and take the point cloud falling within the traffic light area as the first target point cloud, and then take the area where the first target point cloud is located as the first target point cloud area.

[0036] Then, clustering is performed on the first target point cloud region to obtain at least one first point cloud cluster.

[0037] In some embodiments, due to road environment factors, there is a situation where the traffic light is at least partially blocked. For example, when collecting road images and initial point clouds, leaves block the traffic light, and some points belonging to leaves exist in the point cloud projected to the traffic light area. Therefore, it is easy to misdetect traffic lights. If the points belonging to leaves are regarded as traffic lights, the position of the identified traffic light will not match the actual position of the traffic light, affecting the subsequent map construction based on traffic lights and vehicle driving.

[0038] In some embodiments, the first target point cloud region may be clustered using a Euclidean clustering algorithm, a K-Means clustering algorithm, or the like to obtain at least one first point cloud cluster.

[0039] That is, through the clustering algorithm, the points in the first target point cloud area can be divided and further divided into a first point cloud cluster belonging to a traffic light and a first point cloud cluster belonging to a non-traffic light.

[0040] Finally, a point cloud cluster corresponding to a traffic light in at least one first point cloud cluster is labeled.

[0041] At this time, some point cloud areas in the point cloud data are automatically marked as traffic light point cloud data, and the remaining point clouds are road point cloud data.

[0042] Step 12: Obtain result information of marking the traffic light area, wherein the result information includes the coordinates of the center point of the traffic light area.

[0043] Furthermore, when annotating the traffic light area, the center point coordinates may be determined from the traffic light point cloud data corresponding to the traffic light area, and then the center point coordinates may be synchronously annotated in the annotation information of the traffic light area.

[0044] The method for determining the center point coordinates is to construct a three-dimensional model of the traffic light based on the point cloud data of each traffic light, and then determine the center point coordinates based on the three-dimensional model.

[0045] Therefore, the result information can be expressed in the following form: light-mn, where m represents the traffic light number, n represents the center point coordinates, and m is a non-negative integer.

[0046] It can be understood that the coordinates of the center point may be a data point in the traffic light point cloud data, or may not be a data point in the traffic light point cloud data.

[0047] Step 13: If the center point coordinates of the traffic light area are located in the traffic light area and at the center of the traffic light area, the center point coordinates of the traffic light area are displayed in the traffic light area, thereby determining that there is no abnormality in the marking of the traffic light area.

[0048] In step 12, the center point coordinates of the traffic light area are determined, and it is verified again whether the center point coordinates of the traffic light area are located in the traffic light area and at the center of the traffic light area. If so, the center point coordinates of the traffic light area are displayed in the traffic light area, thereby determining that the marking of the traffic light area is normal. If not, it is determined that the marking of the traffic light area is abnormal.

[0049] The road point cloud data can be displayed in a first color, the traffic light point cloud data can be displayed in a second color, and the center point coordinates of the traffic light area can be displayed in a third color, wherein the first color, the second color, and the third color are different from each other. For example, the road point cloud data is displayed in green, the traffic light point cloud data is displayed in red, and the center point coordinates of the traffic light area are displayed in purple. By coloring the point cloud data according to the road point cloud, traffic light point cloud, and center point coordinates, different point clouds can be more clearly distinguished during display, which facilitates visual verification.

[0050] In some embodiments, after the traffic light area is labeled, there are some errors in the labeling. For example, in the same image, if there is a bus, the taillight of the bus is similar to the traffic light, so the point cloud data corresponding to the taillight is also easily labeled as the traffic light area.

[0051] Therefore, it is necessary to filter these incorrectly labeled traffic light areas.

[0052] For example, by adopting a manual exclusion method, when it is manually determined that the target point cloud area corresponds to a traffic light, a labeling determination instruction is input, and in response to the labeling determination instruction for the target point cloud area in the partial point cloud area, the labeling of the target point cloud area is retained.

[0053] For example, by adopting the manual exclusion method, the road point cloud data and the traffic light point cloud data, as well as the center point coordinates of the traffic light point cloud data are displayed first.

[0054] When it is manually determined that the traffic light point cloud data does not correspond to a traffic light, a label deletion instruction is input. In response to the label deletion instruction for the traffic light point cloud data, the label of the traffic light point cloud data is deleted, and it can be converted into road point cloud data.

[0055] In one application scenario, see Figure 2 ,exist Figure 2 In the figure, the road point cloud data and the traffic light point cloud data and the center point coordinates of the traffic light point cloud data are displayed, wherein the point cloud area A and the point cloud area B are automatically marked as traffic light areas. In response to the mark determination instruction for the point cloud area A, the traffic light mark for the point cloud area A is retained. In response to the mark deletion instruction for the point cloud area B, the traffic light mark for the point cloud area B is deleted.

[0056] It should be noted that the present application relates to traffic light marking, but the present application is not limited to traffic lights, but can also be other traffic equipment, such as traffic monitoring cameras, etc.

[0057] In this embodiment, road point cloud data and traffic light point cloud data are displayed, and when the center point coordinates of the traffic light area are located in the traffic light area and at the center of the traffic light area, the center point coordinates of the traffic light area are displayed in the traffic light area. Thus, it is determined that there is no abnormality in the marking of the traffic light area, thereby realizing visual verification of the traffic light marking. The accuracy of the traffic light marking can be improved through visual verification.

[0058] See also Figure 3 , Figure 3 FIG. 1 is a flow chart of another embodiment of a verification method for traffic light labeling provided by the present application. The method further comprises:

[0059] Step 31: Obtain a road stitching image of the road area.

[0060] In some embodiments, during the driving process of the autonomous driving vehicle, a road image of the road area can be obtained by using a shooting device installed on the autonomous driving vehicle. Therefore, the image of the corresponding point cloud data can be saved. Because multiple frames of initial road images of the road area are obtained during the driving process of the autonomous driving vehicle, these initial road images need to be selectively saved to reduce the consumption of storage space.

[0061] In some embodiments, see Figure 4 , step 31 may be the following process:

[0062] Step 311: Acquire at least two frames of initial road images regarding a road area, wherein each of the at least two frames of initial road images includes a traffic light area.

[0063] In some embodiments, the photographing device collects multiple frames of initial road images at a preset frequency. Since the collection frequency is fixed, it is necessary to screen these multiple frames of initial road images.

[0064] For example, the image recognition technology is used to determine the initial road image including the traffic light area, and the remaining initial road images not including the traffic light area can be directly deleted.

[0065] In one application scenario, the road area is a straight area; at least two frames of initial road images are captured by a shooting device located in the vehicle at a preset frequency, wherein the preset frequency is positively correlated with the driving speed of the vehicle in the straight area.

[0066] The at least two frames of initial road images include a first initial road image, and the first initial road image is captured when the vehicle has a driving speed of 0 in a straight-ahead area.

[0067] Specifically, when the vehicle's speed in the straight-ahead area is 0, such as when waiting for a traffic light or temporarily stopping, during this period of time, no matter how many frames of initial road images the camera captures, the content in the road image is almost unchanged. That is, the position of the fixed object that appears is unchanged. Therefore, only a single frame of road image needs to be saved.

[0068] In another application scenario, the road area is an intersection area; at least two frames of initial road images are captured by a shooting device located in the vehicle at a preset frequency, wherein the preset frequency is positively correlated with the turning speed of the vehicle in the intersection area.

[0069] Specifically, the faster the turning speed of the vehicle, the shorter the turning time, and the camera needs to capture more initial road images. Therefore, the preset frequency of the camera is set to be positively correlated with the turning speed of the vehicle in the intersection area to capture more initial road images per unit time.

[0070] In an application scenario, the shooting device shoots at a fixed frequency, so the frame number of the initial road image is positively correlated with the driving speed of the vehicle.

[0071] In some embodiments, when the vehicle's speed is zero, such as when waiting for a traffic light or stopping temporarily, during this period of time, no matter how many frames of initial road images are collected by the image acquisition device, the content in the road image is almost unchanged. That is, the position of the fixed object that appears is unchanged. Therefore, only a single frame of road image needs to be saved.

[0072] In some embodiments, when the vehicle's driving speed is not zero, the initial road image needs to be dynamically determined according to the vehicle's driving speed during this time period. For example, when the vehicle's driving speed is a first speed, the initial road image of the first frame number is saved. When the vehicle's driving speed is a second speed, the initial road image of the second frame number is saved. The first speed is greater than the second speed, and the first frame number is greater than the second frame number.

[0073] In some embodiments, a corresponding initial road image is determined from a plurality of frames of initial road images according to a turning speed of the vehicle; wherein the time interval between adjacent frames of initial road images is negatively correlated with the turning speed.

[0074] Specifically, the faster the turning speed of the vehicle is, the shorter the turning time is, the fewer initial road images captured by the photographing device at a fixed frequency are, and the time interval of the determined target initial road images is shorter.

[0075] The slower the turning speed of the vehicle, the longer the turning time, the more initial road images the camera collects at a fixed frequency, and the longer the time interval of the determined target initial road images. The target initial road images are at least two frames of the multiple frames of initial road images.

[0076] For example, when turning, 3 frames of initial road images need to be saved. When the turning speed of the vehicle is at the third speed, 5 frames of initial road images are collected, and 3 frames of target initial road images are determined from the 5 frames.

[0077] When the turning speed of the vehicle is at the fourth speed, 10 frames of initial road images are collected, and 3 frames of target initial road images are determined from the 10 frames. The third speed is greater than the fourth speed.

[0078] Therefore, the determined time interval between the target initial road images of adjacent frames is negatively correlated with the turning speed.

[0079] Step 312: stitching at least two frames of initial road images to obtain a road stitching image.

[0080] Since the vehicle includes multiple shooting devices, at least two frames of initial road images captured by the multiple shooting devices can be stitched together to obtain a 360-degree panoramic road stitching image. Through image stitching, redundant objects in at least two frames of initial road images can be removed.

[0081] Step 32: Compare the road stitching image with the traffic light point cloud data to determine whether the traffic light area represents a traffic light.

[0082] Step 33: When the traffic light area does not represent a traffic light, the center point coordinates of the displayed traffic light area will be deleted.

[0083] For example, if it is found from the road stitching image that the image area corresponding to the traffic light point cloud data in the road stitching image does not represent the traffic light area, it means that the traffic light area cannot represent the traffic light, and the coordinates of the center point of the traffic light area are deleted. Furthermore, the result information of the traffic light point cloud data annotation corresponding to the traffic light area can be changed to the road point cloud data.

[0084] Step 34: When the traffic light area represents a traffic light, the coordinates of the center point of the displayed traffic light area are retained, and the traffic light number represented by the traffic light area is displayed in the road area.

[0085] When it is determined that the traffic light area represents the traffic light, the traffic light area can be re-numbered and displayed. For example, when initially marked, the traffic light area of ​​the traffic light is numbered 2. Since the traffic light area with the traffic light number 1 cannot represent the traffic light, the corresponding marking information has been deleted, and the traffic light area originally with the traffic light number 2 is changed to 1.

[0086] In one application scenario, the road point cloud data is displayed in green, the traffic light point cloud data is displayed in red, and the center point coordinates of the traffic light area are displayed in purple. The road stitching image is compared with the traffic light point cloud data to determine whether the traffic light area represents a traffic light. When the traffic light area does not represent a traffic light, the center point coordinates of the traffic light area will be deleted and the corresponding traffic light point cloud data will be displayed in green.

[0087] In this embodiment, the road stitching image is compared with the traffic light point cloud data to assist in determining whether the traffic light area represents a traffic light, thereby realizing visual verification of the traffic light annotation. The visual verification can improve the accuracy of the traffic light annotation and the verification efficiency.

[0088] In an application scenario, take an autonomous driving vehicle as an example:

[0089] The autonomous driving vehicle is provided with a plurality of camera devices and a plurality of radar sensors.

[0090] Therefore, during the driving process of the autonomous driving vehicle, each shooting device can collect corresponding road images, and multiple radar sensors can collect corresponding point cloud data.

[0091] Then, based on the road image and the point cloud data, the point cloud area corresponding to the traffic light in the point cloud data is determined and automatically labeled. The center point coordinates of each point cloud can be calculated and used as the annotation information.

[0092] Then save all center point coordinates to a file, for example, in a text file format.

[0093] Furthermore, the point cloud data of the traffic intersection corresponding to the traffic lights will be stored. Among them, these files do not have the point cloud data corresponding to the above traffic lights.

[0094] Before saving, the point cloud corresponding to the traffic intersection is colored green, for example, to facilitate the distinction from the traffic light point cloud data during subsequent visualization.

[0095] Furthermore, the traffic light point cloud data is stored, and each traffic light point cloud data is matched with the traffic intersection point cloud data, for example, named as junction-mz, where m is the traffic light number and z is the traffic intersection number.

[0096] Furthermore, in the actual automatic annotation process, it was found that point cloud processing and visualization also have difficulty in handling some edge events. For example, bridge heads and non-traffic light signs that look like traffic lights are easily perceived and misidentified, and the corresponding traffic light point clouds are easily extracted by the automatic annotation algorithm. Therefore, visual-assisted verification can improve the accuracy of verification.

[0097] Furthermore, in the process of saving road images, it is impossible to save all images due to too many images, which will make the size of the entire saved image data very large. In addition, too many useless road images will also increase the difficulty of selection.

[0098] Therefore, the following scheme is adopted to stitch the road images.

[0099] First, the road image is marked. For example, during the recognition process, the road image in which the traffic light is recognized is marked.

[0100] Then, according to the vehicle's speed, it is selectively saved from the marked road images.

[0101] For example, when the vehicle stops, only one image is saved. For example, when the vehicle stops and waits for the traffic light; according to the speed, the image is stored at a reduced frequency. For example, when the vehicle speed is very slow, it is not necessary to store every image; when the vehicle turns, more images need to be stored. According to the turning speed, the frequency is reduced reasonably. Otherwise, the co-viewing range of the road images is small, which is not conducive to road image stitching.

[0102] The marked road image can be named in the following format: z_tk, where z is the number of the traffic intersection and tk is the timestamp at time k.

[0103] Then, according to the number of the traffic intersection, all images with the same number of the traffic intersection are selected for offline stitching.

[0104] Therefore, we use the point cloud visualization tool cloudcompare for verification, and use the stitched images for visual assistance.

[0105] For example, the traffic intersection point cloud, the traffic light point cloud data, and the center point of the traffic light point cloud data are rendered in different colors. If the center point is determined to be in the middle of the traffic light point cloud data, a label determination instruction is generated. In response to the label determination instruction, the label of the traffic light area is retained. Otherwise, a label deletion instruction is generated, and in response to the label deletion instruction, the label of the traffic light area is deleted.

[0106] In this process, the road stitching image can be combined to assist in verifying the traffic light point cloud. If the traffic light point cloud data is incorrectly labeled, a label deletion instruction is generated. In response to the label deletion instruction, the label of the target point cloud area is deleted.

[0107] See also Figure 5 , Figure 5 1 is a flow chart of an embodiment of a method for generating a driving map provided by the present application. The method comprises:

[0108] Step 51: Execute the traffic light labeling process on the original map to obtain the traffic light labeling result.

[0109] By collecting point cloud data and multiple frames of road images, an original map can be formed. Then, the traffic lights are identified through the traffic light recognition algorithm, and the traffic light annotation process is performed on the original map to obtain the traffic light annotation result.

[0110] Step 52: Perform a verification process on the traffic light annotation results to generate a driving map.

[0111] The verification process includes a verification method for traffic light labeling as provided in any of the above embodiments.

[0112] In this embodiment, the verification method for traffic light labeling provided in any of the above embodiments can improve the accuracy of traffic light labeling, thereby generating a high-precision driving map and subsequent driving of the vehicle.

[0113] See also Figure 6 , Figure 6 6 is a schematic diagram of the structure of an embodiment of the vehicle-mounted device provided by the present application. The vehicle-mounted device 60 includes a processor 61, a memory 62 and a display 63 coupled to the processor 61, the memory 62 is used to store a computer program, and the processor 61 is used to execute the computer program to implement the following method:

[0114] Acquire and display road point cloud data and traffic light point cloud data, wherein the road point cloud data represents the road area, and the traffic light point cloud data represents the traffic light area identified in the road area; obtain result information of marking the traffic light area, wherein the result information includes the coordinates of the center point of the traffic light area; if the coordinates of the center point of the traffic light area are located in the traffic light area and at the center of the traffic light area, then display the coordinates of the center point of the traffic light area in the traffic light area, thereby determining that there is no abnormality in the marking of the traffic light area.

[0115] It can be understood that the processor 61 is also used to execute a computer program to implement the method of any of the above embodiments, which will not be elaborated here.

[0116] See also Figure 7 , Figure 7 1 is a schematic diagram of a structure of an embodiment of a non-volatile computer-readable storage medium provided by the present application. The computer-readable storage medium 70 is used to store a computer program 71. When the computer program 71 is executed by a processor, it is used to implement the following method:

[0117] Acquire and display road point cloud data and traffic light point cloud data, wherein the road point cloud data represents the road area, and the traffic light point cloud data represents the traffic light area identified in the road area; obtain result information of marking the traffic light area, wherein the result information includes the coordinates of the center point of the traffic light area; if the coordinates of the center point of the traffic light area are located in the traffic light area and at the center of the traffic light area, then display the coordinates of the center point of the traffic light area in the traffic light area, thereby determining that there is no abnormality in the marking of the traffic light area.

[0118] It can be understood that when the computer program 71 is executed by the processor, it is also used to implement the method of any of the above embodiments, which will not be repeated here.

[0119] In the several embodiments provided in this application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device implementation described above is only illustrative, for example, the division of the circuit or unit is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0120] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0121] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0122] The above description is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made according to the description and drawings of the present application, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A verification method for traffic light labeling, characterized in that: include: Acquire and display road point cloud data and traffic light point cloud data, wherein the road point cloud data represents a road area, and the traffic light point cloud data represents a traffic light area identified in the road area; Obtaining result information of labeling the traffic light area, wherein the result information includes the coordinates of the center point of the traffic light area; wherein the method of determining the coordinates of the center point includes: constructing a three-dimensional model of the traffic light based on each of the traffic light point cloud data, and determining the coordinates of the center point based on the three-dimensional model; If the center point coordinates of the traffic light area are located in the traffic light area and at the center of the traffic light area, the center point coordinates of the traffic light area are displayed in the traffic light area, thereby determining that there is no abnormality in the marking of the traffic light area.

2. The method according to claim 1, characterized in that Further including: Acquire a road stitching image about the road area; Comparing the road stitching image with the traffic light point cloud data to determine whether the traffic light area represents a traffic light; When the traffic light area does not represent a traffic light, the coordinates of the center point of the traffic light area will be deleted; When the traffic light area represents a traffic light, the center point coordinates of the traffic light area are kept displayed, and the traffic light number represented by the traffic light area is displayed in the road area.

3. The method according to claim 1 or 2, characterized in that: The road point cloud data is displayed in a first color, the traffic light point cloud data is displayed in a second color, and the center point coordinates of the traffic light area are displayed in a third color, wherein the first color, the second color and the third color are different from each other.

4. The method according to claim 2, characterized in that: The acquiring of the road stitching image about the road area comprises: Acquire at least two frames of initial road images about the road area, wherein each of the at least two frames of initial road images includes the traffic light area; The at least two frames of initial road images are spliced ​​to obtain the road spliced ​​image.

5. The method according to claim 4, characterized in that The road area is a straight-ahead area; The at least two frames of initial road images are obtained by photographing a photographing device located in the vehicle at a preset frequency, wherein the preset frequency is positively correlated with the driving speed of the vehicle in the straight-ahead area.

6. The method according to claim 5, characterized in that The at least two frames of initial road images include a first initial road image, and the first initial road image is captured when the driving speed of the vehicle in the straight-ahead area is 0.

7. The method according to claim 4, characterized in that The road area is a junction area; The at least two frames of initial road images are obtained by photographing a photographing device located in the vehicle at a preset frequency, wherein the preset frequency is positively correlated with a turning speed of the vehicle at the intersection area.

8. A method for generating a driving map, characterized in that: include: Execute the traffic light labeling process on the original map to obtain the traffic light labeling result; Performing a verification process on the traffic light annotation result to generate the driving map; Wherein, the verification process includes the verification method for traffic light marking as described in any one of claims 1-7.

9. A vehicle-mounted device, characterized in that: The vehicle-mounted device includes a processor and a memory and a display coupled to the processor, the memory is used to store a computer program, and the processor is used to execute the computer program to implement the verification method for traffic light labeling as described in any one of claims 1 to 7 or the method for generating a driving map as described in claim 8.

10. A non-volatile computer-readable storage medium, characterized in that: The non-volatile computer-readable storage medium is used to store a computer program, which, when executed by a processor, is used to implement the verification method for traffic light labeling as described in any one of claims 1 to 7 or the method for generating a driving map as described in claim 8.

Images