High-precision map crowdsourcing collection method, device, platform and medium

Abstract

This application discloses a crowdsourced data collection method, device, platform, and medium for high-precision maps, belonging to the field of high-precision map data collection technology. The method includes: a data acquisition camera collecting road data according to a received data acquisition task instruction; a central processing unit coupled to the camera determining a network load ratio parameter based on the camera's corresponding data acquisition parameters and real-time network transmission speed, and judging the relationship between the network load ratio parameter and a preset threshold. The data acquisition parameters include a preset image compression ratio and / or frame extraction frequency; the central processing unit selects an appropriate data transmission method based on the judgment result to transmit the collected road data. This method selects the corresponding data transmission method based on the relationship between the network load ratio parameter and the preset threshold, avoiding data loss during transmission and ensuring data integrity while performing real-time transmission.

Description

Technical Field

[0001] This application relates to the field of high-precision map data acquisition technology, and in particular to a high-precision map crowdsourcing acquisition method, device, platform and medium. Background Technology

[0002] Currently, the workflow for high-precision map crowdsourcing data collection is as follows: First, the collection area is determined; then, data is collected by driving through the data collection lanes and locations; after the data collection task is completed, the acquired images and vehicle trajectory data are copied to a portable hard drive for subsequent data processing. Existing collection solutions require copying images and trajectory data to a portable hard drive after the data collection task is completed. The massive amount of data necessitates the purchase of numerous portable hard drives, and the manpower and material costs for copying data are high. Furthermore, reading the data from the portable hard drive to the corresponding processing device after copying takes time, resulting in a significant data lag and hindering timely data processing, gradually failing to meet the daily map update freshness requirements. The problem with online data transmission is that the mapping algorithm has high image frame rate requirements, and the upper limits of communication methods such as 4G bandwidth cannot meet the transmission needs. When all collected data is transmitted back, valid data may be washed away, resulting in the loss of important data information, ultimately leading to incomplete map data and affecting subsequent high-precision map production or updates. Summary of the Invention

[0003] In view of the problems in the prior art, high-precision map crowdsourcing data collection devices cannot transmit data in real time after collection, and the data loss is easily caused by the limitations of communication means during real-time transmission. This application proposes a high-precision map crowdsourcing data collection method, device and platform.

[0004] One technical solution of this application provides a high-precision map crowdsourcing data collection method, comprising: a data collection camera collecting road data according to a received data collection task instruction; a central processing unit coupled to the data collection camera determining a network load pressure ratio parameter based on the data collection parameters corresponding to the data collection camera and the real-time network transmission speed, and judging the relationship between the network load pressure ratio parameter and a preset threshold, wherein the data collection parameters include a preset image compression ratio and / or frame extraction frequency; and the central processing unit selecting an appropriate data feedback method to transmit the collected road data based on the judgment result.

[0005] In another technical solution of this application, a high-precision map crowdsourcing data collection device is provided, comprising: a navigation unit, which acquires trajectory information from the vehicle in real time and transmits the trajectory information to a data platform so that the data platform can issue data collection task instructions based on the trajectory information; the navigation unit includes a GNSS system and an inertial navigation system; a central processing unit, which controls the data collection process according to the data collection task instructions issued by the data platform; and a data acquisition camera, which performs road data collection according to the control instructions received from the central processing unit; wherein the central processing unit is coupled to the data acquisition camera and determines the network load pressure ratio parameter based on the data acquisition parameters corresponding to the data acquisition camera and the real-time network transmission speed, and judges the relationship between the network load pressure ratio parameter and a preset threshold, and selects an appropriate data return method to transmit the collected road data based on the judgment result; the data acquisition parameters include a preset image compression ratio and / or frame extraction frequency.

[0006] In another technical solution of this application, a high-precision map crowdsourcing data collection platform is provided, comprising: a data platform and a data collection device. The data platform receives trajectory information from a vehicle and issues data collection task instructions based on the relationship between the coverage rate of the trajectory information and a preset coverage threshold. The vehicle-side data collection device performs road data collection and data transmission according to the data collection task instructions. The data collection device includes: a navigation unit, which acquires the trajectory information from the vehicle in real time and transmits the trajectory information to the data platform so that the data platform can issue data collection task instructions based on the trajectory information; the navigation unit includes a GNSS system and an inertial navigation system; a central processing unit, which controls the data collection process according to the data collection task instructions issued by the data platform; and a data collection camera, which performs road data collection according to the control instructions received from the central processing unit. The central processing unit is coupled to the data collection camera and determines a network load pressure ratio parameter based on the data collection parameters corresponding to the data collection camera and the real-time network transmission speed. It also determines the relationship between the network load pressure ratio parameter and a preset threshold and selects an appropriate data transmission method to transmit the collected road data based on the determination result. The data collection parameters include a preset image compression ratio and / or frame extraction frequency.

[0007] In another technical solution of this application, a computer-readable storage medium is provided, which stores computer instructions, wherein the computer instructions are operated to execute the high-precision map crowdsourcing data collection method in Solution 1.

[0008] The beneficial effects of this application are: by setting the network load pressure ratio parameter, this application selects the corresponding data return method according to the relationship between the network load pressure ratio parameter and the preset threshold, thereby avoiding data loss during transmission, ensuring data integrity, and ensuring real-time data transmission. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This paper illustrates one implementation of the high-precision map crowdsourcing data collection method of this application;

[0011] Figure 2 An example of the execution process of the high-precision map crowdsourcing data collection method of this application is shown;

[0012] Figure 3 An example of the high-precision map crowdsourcing data collection method of this application is shown;

[0013] Figure 4 One embodiment of the high-precision map crowdsourcing data collection device of this application is shown;

[0014] Figure 5 An example of the high-precision map crowdsourcing data collection device of this application is shown;

[0015] Figure 6 An example of the central processing unit of the high-precision map crowdsourcing data collection device of this application is shown;

[0016] Figure 7 An implementation of the high-precision map crowdsourcing data collection platform of this application is shown.

[0017] The accompanying drawings have illustrated specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a product or device comprising a series of steps or units is not necessarily limited to those units explicitly listed, but may include other units not explicitly listed or inherent to such products or devices.

[0020] In existing technologies, after high-precision map crowdsourcing data collection devices collect data, images and trajectory data are copied using external hard drives. Due to the massive amount of data collected, this necessitates purchasing numerous external hard drives, and copying the data requires significant manpower and resources. Furthermore, reading the data from the external hard drive to the corresponding processing device after copying takes time, resulting in a long waiting time that hinders timely data processing and gradually fails to meet the freshness requirements of map updates. Regarding online data transmission, current online transmission communication methods often suffer from communication limitations, leading to the potential loss of valuable data during transmission, resulting in incomplete map data and impacting subsequent high-precision map updates or production processes. Therefore, addressing the problems of significant manpower and resources required for offline data transmission and the limitations imposed by communication methods on online transmission, which lead to the loss of valuable data, this application proposes a high-precision map crowdsourcing data collection method, device, and platform.

[0021] This application adjusts the data acquisition task and the acquired data according to the communication limit requirements of the online transmission method, removes redundant data from the acquired data, reduces the amount of data transmission, and thus achieves real-time transmission of acquired data while ensuring the integrity of effective data in the data transmission and avoiding data loss.

[0022] Figure 1 An implementation of the crowdsourced data collection method for high-precision maps of this application is shown.

[0023] exist Figure 1In the embodiment shown, the high-precision map crowdsourcing data collection method of this application includes: process S101, where the data collection camera collects road data according to the received data collection task instruction; process S102, where the central processing unit coupled to the data collection camera determines the network load pressure ratio parameter according to the data collection parameters corresponding to the data collection camera and the real-time network transmission speed, and judges the relationship between the network load pressure ratio parameter and a preset threshold, wherein the data collection parameters include a preset image compression ratio and / or frame extraction frequency; process S103, where the central processing unit selects an appropriate data feedback method to transmit the collected road data according to the judgment result.

[0024] In this embodiment, the high-precision map crowdsourcing data collection method of this application, based on the relationship between the network load pressure ratio parameter and a preset threshold, appropriately adjusts the data collection task and the collected data. When the network transmission pressure is high, the collection task is stopped or the amount of data transmitted is reduced to avoid repeated execution of the data collection task, reduce the data collection pressure and data return pressure of the data collection device, and achieve real-time data return. Simultaneously, the central processing unit calculates the network load pressure ratio parameter based on the data transmission parameters corresponding to the acquisition camera, and determines the current data transmission status based on the relationship between the network load pressure ratio parameter and the preset threshold. Depending on different situations, the data collection parameters are adjusted to control the amount of data transmitted, and settings are set for full return, partial return, or rejection of data collected by the data acquisition camera. By selectively returning the collected data, real-time data return is achieved, redundant data in the collected data is filtered out, and effective data loss is avoided during data transmission, thus preventing data integrity issues. Compared to offline transmission via mobile hard drives, the technical solution of this application can effectively reduce data transmission costs, achieve real-time data processing, and ensure the integrity of effective data during data transmission, avoiding data loss.

[0025] exist Figure 1 In the specific implementation shown, the high-precision map crowdsourcing data collection method of this application includes process S101, in which the data collection camera performs road data collection according to the received data collection task instruction.

[0026] In one embodiment of this application, the acquisition camera acquires images of the road according to the received data acquisition task instruction, wherein the acquisition camera includes a monocular camera.

[0027] In one example of this application, after receiving a data acquisition task, the acquisition camera collects road data for a road segment. This road data collection utilizes a high dynamic range (HDR) monocular camera. When using a binocular camera for data acquisition, the two lenses of the device are ideally identical, and the distance between them is fixed. However, the hand-polished lenses inevitably introduce differences between the two lenses. Furthermore, the thermal expansion and contraction caused by temperature changes within the vehicle affects the distance between the lenses, ultimately limiting the accuracy of the binocular device and making measurement errors unavoidable. This application, however, uses a monocular camera for road data acquisition, avoiding the bottleneck of difficulty in improving the accuracy of binocular cameras. Furthermore, using a monocular device reduces costs, truly realizing a consumer-grade crowdsourcing device.

[0028] In one embodiment of this application, the acquisition camera performs road data acquisition according to the received data acquisition task instruction, including: acquiring the vehicle's trajectory information in real time through the navigation unit on the vehicle, and transmitting the trajectory information to the data platform so that the data platform can issue a data acquisition task instruction based on the trajectory information. The navigation unit includes a GNSS system and an inertial navigation system. The central processing unit controls the acquisition camera to perform road data acquisition according to the data acquisition task instruction issued by the data platform.

[0029] In this embodiment, the navigation unit on the vehicle, such as a GNSS system or an inertial navigation system, records and acquires the vehicle's driving trajectory information. The vehicle refers to the vehicle where the acquisition camera is located. After the vehicle sends the acquired trajectory information to the data platform, the data platform issues a data acquisition task instruction based on the trajectory information. The central processing unit of the vehicle receives the data acquisition task instruction and then controls the acquisition camera to collect the corresponding road data.

[0030] When collecting crowdsourced data for high-precision maps, the navigation unit acquires real-time trajectory information from the vehicle-mounted cameras. For example, in reality, data collection vehicles may travel along roads, with multiple vehicles potentially traversing the same road segment, leading to duplicate data collection and wasting both collection and data transmission resources. Therefore, based on the vehicles' trajectories, a data collection task is issued when a vehicle is on an uncollected road segment. Conversely, when a vehicle is on a popular road segment frequented by multiple vehicles, no further data collection task instructions are issued for segments where data has already been collected. This staggered allocation of data collection vehicles ensures efficient use of resources and avoids waste. After issuing the data collection task instruction, the cameras collect road data accordingly.

[0031] In one embodiment of this application, the data platform issues a data collection task instruction based on trajectory information, including: the data platform issues a data collection task instruction based on the coverage of the trajectory information relative to the geofence, wherein when the coverage does not exceed a preset coverage threshold, a data collection task instruction is issued; when the coverage exceeds the preset coverage threshold, no data collection task instruction is issued.

[0032] In this embodiment, when a vehicle is in motion, its trajectory information is sent to the data platform. Upon receiving this information, the data platform issues data collection tasks based on the coverage of the trajectory image relative to the geofence over a given period, thus achieving staggered matching of popular geofences and collection vehicles. When the coverage rate does not exceed a preset threshold, it indicates that data collection at that location is incomplete, and a data collection task instruction needs to be issued to collect road data. When the coverage rate exceeds the preset threshold, it indicates that road data for that area has been completely collected, and repeated collection is unnecessary; therefore, no data collection task instruction is issued.

[0033] In one example of this application, the data platform receives trajectory information sent by the vehicle. When the geofence coverage of the vehicle trajectory information is not greater than a preset coverage threshold, the data platform normally issues a data collection task instruction to collect data; when it is greater than the preset coverage threshold, the data platform no longer issues a data collection task instruction. The preset coverage threshold can be determined appropriately based on empirical values ​​of the coverage of the geofence by the vehicle's driving trajectory. When the data platform publishes a data collection task, for example, when a frequently used vehicle passes through a popular road segment, the data platform will no longer trigger a data collection task; when the same task is published multiple times and data is collected multiple times, the publication of the task will be terminated, and no more data collection will be performed.

[0034] Figure 2 An example of the execution process of the high-precision map crowdsourcing data collection method of this application is shown.

[0035] like Figure 2 The figure shows an example of the high-precision map crowdsourcing data collection method of this application in a practical application. Specifically, it involves the application of OEM software (Original Equipment Manufacture), which includes a task triggering software module DA-SDK (Software Development Kit). This software is housed in the central processing unit. The practical application of the method of this application will be described below with reference to the accompanying drawings.

[0036] First, the software is launched to initialize the software and SDK modules. Simultaneously, the data acquisition device is activated, and the vehicle-mounted navigation system collects the vehicle's driving trajectory. The vehicle's trajectory is then packaged and reported to the data acquisition task platform and data platform. The data acquisition task platform issues data acquisition task orders based on the vehicle's trajectory information and the coverage of the geofence. The SDK module receives these orders and transmits the corresponding instructions to the acquisition camera, controlling it to collect road data. The road data is in image format. During actual road data acquisition, video recording can be performed by the acquisition camera, and frame extraction can be used to visualize the road data. The collected road images are then processed by the SDK module and reported to the data acquisition task platform. The order is also fed back, thus completing the entire road data acquisition process. After the road data acquisition is complete, the software is shut down, and the SDK module's processing stops. It should be noted that the above is only an example of the application of the method of this application. This application does not limit the application conditions such as the type of software used in the specific application, and all of them are within the protection scope of this application.

[0037] exist Figure 1 In the specific implementation shown, the high-precision map crowdsourcing data acquisition method of this application includes process S102, in which a central processing unit coupled to the acquisition camera determines a network load pressure ratio parameter based on the data acquisition parameters corresponding to the acquisition camera and the real-time network transmission speed, and determines the relationship between the network load pressure ratio parameter and a preset threshold. The data acquisition parameters include a preset image compression ratio and / or frame extraction frequency.

[0038] In this implementation, to avoid limitations imposed by communication methods during online data transmission, the central processing unit (CPU) determines network load parameters based on the data transmission parameters corresponding to the actual acquisition camera, thereby adjusting the amount of data transmitted back. These data transmission parameters include the real-time speed of the acquisition vehicle, the size of the acquired images, the image compression rate, and the corresponding real-time network transmission speed.

[0039] In one embodiment of this application, a central processing unit (CPU) coupled to a data acquisition camera determines a network load pressure ratio parameter based on the data acquisition parameters corresponding to the data acquisition camera and the real-time network transmission speed. This includes: the CPU determining the real-time driving speed of the vehicle based on the coupled navigation unit; the CPU acquiring the data acquisition parameters and the real-time network transmission speed; and the CPU calculating the real-time driving speed, the real-time network transmission speed, and the data acquisition parameters according to a preset network load pressure ratio calculation formula to determine the network load pressure ratio parameter. The network load pressure ratio parameter is positively correlated with the real-time driving speed, negatively correlated with the network transmission speed, and positively correlated with the data acquisition parameters.

[0040] In this embodiment, the real-time driving speed of the vehicle is acquired by a navigation system mounted on the vehicle. Simultaneously, the central processing unit acquires the corresponding real-time network transmission speed, as well as the preset image compression ratio and preset frame extraction frequency in the collected data, and then calculates the network load pressure ratio parameter. Specifically, the higher the real-time driving speed of the vehicle, the higher the corresponding network load pressure ratio parameter; the lower the network transmission speed, the higher the corresponding network load pressure ratio parameter; the higher the compression ratio in the data acquisition parameters, the higher the corresponding network load pressure ratio parameter; and the lower the frame extraction frequency, the higher the corresponding network load pressure ratio parameter.

[0041] In one example of this application, taking a data collection vehicle as an example, the process of calculating the network load pressure ratio parameter based on the data collection parameters is explained.

[0042] When the data acquisition vehicle collects road data, its real-time speed is obtained through a navigation unit mounted on the vehicle, such as a GNSS system and an inertial measurement system. This real-time speed is denoted as v (km / h). The frame extraction frequency, denoted as n (n frames / s), is preset in the data acquisition parameters. The data is collected using a monocular camera, and the data is represented as images. The frame extraction frequency indicates the number of images extracted from the original image data per unit time. The real-time network transmission speed during online transmission is denoted as a (KB / s). The memory size of a single image is denoted as b (KB), and the preset image compression ratio is denoted as c. Based on the above data, the shooting interval distance d (m) between two acquired images can be calculated as follows:

[0043] d=3.6v / n(1)

[0044] In the high-precision map crowdsourcing data collection method of this application, a network load pressure ratio parameter A is set, wherein the network load pressure ratio parameter A is expressed as:

[0045] A = nbc / a(2)

[0046] Combining the two formulas above, we get:

[0047] A = 3.6vbc / da(3)

[0048] As shown in Formula 3, the higher the vehicle's speed, the larger the memory size of a single image, the higher the image compression ratio, and the larger the corresponding network load pressure ratio A. Conversely, the greater the interval between two consecutive captured images (i.e., the lower the frame rate), and the higher the real-time network speed, the smaller the corresponding network load pressure ratio A. The network load pressure ratio A represents the network data transmission pressure of the current data acquisition vehicle. A large network load pressure ratio A indicates high current network transmission pressure, potentially leading to data loss during transmission; a small network load pressure ratio A indicates low current network transmission pressure, allowing for effective data transmission.

[0049] The high-precision map crowdsourcing data collection method of this application uses a network load pressure ratio parameter to represent the network transmission pressure during the current data transmission. Based on the current network transmission pressure, the data transmission parameters are reasonably adjusted to achieve real-time data transmission with low data transmission volume, while avoiding the loss of effective data in road data due to high network transmission pressure during data transmission.

[0050] exist Figure 1 In the specific implementation shown, the high-precision map crowdsourcing data collection method of this application includes process S103, in which the central processing unit selects the appropriate data transmission method based on the judgment result and transmits the collected road data.

[0051] In this implementation, a preset threshold is set based on the correlation between the network load pressure ratio parameter and the actual network transmission pressure. The corresponding data backhaul method is selected based on the relationship between the network load pressure ratio parameter and the preset threshold to backhaul the collected data. Specifically, when the network load pressure ratio parameter is not greater than the preset threshold, it indicates that the network transmission pressure is low, and no adjustment to the data transmission parameters is needed; the collected data is then transmitted. When the network load pressure ratio parameter is greater than the preset threshold, it indicates that the network transmission pressure is high, and there is a possibility of data loss during transmission. In this case, the data transmission parameters need to be adjusted, for example, by reducing the compression ratio of image data or the frame skipping frequency, to reduce the amount of data transmitted, thereby reducing the network transmission pressure and enabling real-time data backhaul while avoiding the loss of valid data.

[0052] In one embodiment of this application, the central processing unit selects an appropriate data transmission method based on the judgment result to transmit the collected road data, including: if the first network load pressure ratio parameter is not greater than a first threshold, the collected data is directly transmitted back; if the first network load pressure ratio parameter is greater than the first threshold, the data acquisition parameters are adjusted, and the adjusted second network load pressure ratio is calculated; if the second network load pressure ratio parameter is not greater than the second threshold, the collected data is transmitted back; if the second network load pressure ratio parameter is greater than the second threshold, the data acquisition parameters are further adjusted, and the adjusted third network load pressure ratio parameter is calculated; if the third network load pressure ratio parameter is not greater than the second threshold, the collected data is transmitted back; if the third network load pressure ratio parameter is still greater than the second threshold, the current data acquisition task is abandoned.

[0053] In this embodiment, the transmission pressure of the current network data transmission is determined by comparing the network load pressure ratio parameter with the preset first threshold and second threshold. By appropriately adjusting parameters such as compression ratio and frame extraction frequency, the amount of data transmitted is reduced, thereby reducing the transmission pressure of network data transmission, realizing real-time transmission, and avoiding the loss of effective data.

[0054] In one embodiment of this application, the adjustment range of the data acquisition parameters does not exceed a preset adjustment range.

[0055] In this embodiment, when adjusting the data acquisition parameters, including the compression ratio and frame rate, each data parameter is adjusted within a certain range to ensure the effectiveness of the acquired data.

[0056] In one example of this application, when the network load pressure ratio parameter is not greater than 1, it means that the network bandwidth is obviously sufficient to meet the requirements of image transmission, and there is no need to discard images or acquisition tasks. The value of 1 is used as the first threshold of the network load pressure ratio parameter.

[0057] In one example of this application, according to domestic speed limit information, the minimum speed limit for urban roads without a center line is 30 km / h. Substituting this speed value into the above formula (1), we can obtain the actual distance d between the two images taken at this speed:

[0058] d = 30 × 3.6 × 1 / n meters = 108 / n meters

[0059] Based on the distribution of elements such as streetlights, signs, and lane lines in urban road conditions, when the frame rate is 20 frames per second, the shooting distance between every two photos is 5.4 meters. According to surveying experience, this distance can meet the requirements for high-precision map collection.

[0060] The data collection vehicle in this application can use a 4G network card for real-time data transmission. Assuming the network transmission speed is a = 200KB / s and the memory size of each photo is b = 300KB, when the network load pressure parameter A is 1, substituting into the above formula (3), the compression ratio c can be obtained as:

[0061] c=(5.4×200) / (3.6×30×300)=1 / 10

[0062] Therefore, when the vehicle speed v is 30km / h, the network transmission speed a is 200KB / s, and the memory size of each photo b is 300KB, the compression ratio needed to balance the network transmission load is 1 / 10. This compression ratio allows for lossless image compression and decompression, which is a commonly used compression ratio in the industry, and the image data is usable.

[0063] In summary, the preset frame rate can be set to 20 frames per second, and the preset image compression ratio can be set to 1 / 10.

[0064] In one example of this application, based on map surveying experience, to avoid the collected photos completely covering various elements such as streetlights and signs on the road, the maximum shooting interval d between two adjacent photos cannot exceed 15m; at the same time, to ensure that the photos can be used after decompression, the compression ratio cannot exceed 1 / 18. Based on the maximum speed limit of 120km / h stipulated in my country's highway regulations, substituting into the above formula (3), the network load pressure ratio parameter A can be obtained as follows:

[0065] A=(3.6×120×300)÷(15×18×200)=2.4

[0066] The network load pressure ratio parameter A being 2.4 indicates that the image data volume is 2.4 times the network load. Based on experience in network data transmission, it is clear that the network pressure is high and the transmission delay is long at this time. To verify the frame rate at this time, the network load pressure ratio parameter A being 2.4 is substituted into formula (2), and the frame rate n at this time is obtained as follows:

[0067] n = 18 × 2 × 2.4 ÷ 3 = 28.8

[0068] At this point, the frame rate is approximately 29 frames per second, which is relatively high.

[0069] To avoid excessive network load and long transmission times, vehicle speed needs to be limited, and data acquisition tasks at higher vehicle speeds can be abandoned. Based on driving experience, vehicles travel at speeds below 90 km / h for 95% of the time; therefore, images should not be acquired when the speed exceeds 90 km / h. In this case, A is:

[0070] A=(3.6×90×300)÷(15×18×200)km / h=1.8

[0071] At this time, the vehicle speed is close to the maximum speed limit stipulated by my country's highway regulations. Since the vehicle speed is relatively slow on urban roads, it can fully meet the data collection requirements for urban roads. When the vehicle is traveling at high speed, it is also relatively easy to meet this speed requirement. That is, A=1.8 can be used as an indicator to evaluate whether to carry out the data collection task. In other words, the value of 1.8 is used as the second threshold of the network load pressure ratio parameter A.

[0072] In summary, the real-time variables input from the outside are the vehicle speed v and the network transmission speed a; the variables with a specified input range are the shooting interval distance d between two photos or the frame rate n and the network load pressure ratio parameter A; the variables that can be controlled by the algorithm are the frame rate n and the compression ratio c; ultimately, more efficient image transmission is achieved.

[0073] The network load pressure ratio parameter A describes the network transmission pressure. A higher frame extraction frequency, faster transmission speed, and larger compression ratio result in a larger network load pressure ratio A and greater network transmission pressure. In summary:

[0074] When A≤1, the network bandwidth fully meets the transmission requirements, and all images are fully transmitted back.

[0075] When 1 < A ≤ 1.8, the network transmission pressure is high. Reduce the frame skipping frequency or reduce the image compression ratio, and discard the return of some images. When adjusting the frame skipping frequency n, the adjustment range of n must be greater than zero. The frame skipping range of the compression ratio c is between 1 / 10 and 1 / 18.

[0076] When A > 1.8, the network transmission pressure is too high, and the return transmission task for that part is abandoned.

[0077] The high-precision map crowdsourcing data collection method of this application selects the corresponding data return method by setting the network load pressure ratio parameter and the relationship between the network load pressure ratio parameter and the preset threshold, so as to avoid data loss during transmission, ensure data integrity, and ensure real-time data transmission.

[0078] Figure 3 An example of the crowdsourced data collection method for high-precision maps presented in this application is shown.

[0079] like Figure 3 As shown, combined with Figure 3 The example further illustrates the calculation process of the network load pressure ratio and the adjustment process of data acquisition parameters in the central processing unit of this application.

[0080] First, the central processing unit obtains the vehicle speed V through the navigation unit; the memory size b of each image captured by the acquisition camera (e.g., a monocular camera); and the network speed a of the communication module (e.g., a 4G communication module) during road data transmission. When making a judgment, the frame sampling frequency n is loaded, and the distance interval d between two adjacent frames is calculated based on the preset frame sampling frequency n = 20 frames / second. This is done using the formula (1) above, and it is determined whether the value of distance d is lower than the preset distance threshold. Based on map surveying experience, to avoid photos completely covering various elements such as streetlights and signs on the road, the preset distance threshold for the interval d between two photos can be set to 15m. If the calculated value of d is greater than 15m, it indicates that the current data acquisition task does not meet the data acquisition requirements, and the acquisition task is directly abandoned; if the calculated value of d is not greater than 15m, subsequent judgments are made.

[0081] Load the preset image compression ratio c, where the initial default image compression ratio c is 1 / 10. Calculate the value of the network load pressure ratio parameter A using the above formula (3), and determine the relationship between the value of A and the preset threshold. First, determine the relationship between the network load pressure ratio parameter A and the first threshold 1. If the network load pressure ratio parameter A is not greater than 1, it indicates that the current network status fully meets the current data transmission requirements, and the collected data is then transmitted back. If the network load pressure ratio parameter A is greater than 1, adjust the compression ratio c of the collected images and the frame sampling frequency n. By reducing the amount of data transmitted, the network transmission pressure is reduced, thereby reducing the network load pressure ratio parameter A to adapt to the network transmission capacity and avoid the loss of effective data during data transmission.

[0082] like Figure 3In the example shown, if the network load pressure ratio parameter A is greater than the first threshold 1, the image compression ratio c is reduced. In order to reasonably adjust the image compression ratio c and the frame extraction frequency n and ensure the validity of the image data, the adjustment of the image compression ratio c and the frame extraction frequency n has a certain adjustment range. When reducing the value of the compression ratio c, the value of the compression ratio c is between 1 / 10 and 1 / 14. After reducing the compression ratio c, the value of the network load pressure ratio parameter A at this time is calculated by the above formula (3), and the relationship between the value of A and the second threshold is determined, where the second threshold can be set to 1.8. When the network load pressure ratio parameter A at this time is not greater than the second threshold 1.8, the road data collected at this time is transmitted back. Because the compression ratio c is reduced, the amount of data transmitted at this time is reduced, which is in line with the current network transmission pressure. If the network load pressure ratio A remains greater than 1.8 when the compression ratio c decreases between 1 / 10 and 1 / 14, the frame sampling frequency n needs to be adjusted. Decreasing n lengthens the interval between collected road images, reducing data volume and thus lowering network transmission pressure. The frame sampling frequency n should be greater than zero. If, after adjustment, the network load pressure ratio A is no greater than 1.8, the collected road data is transmitted back. If, after decreasing the frame sampling frequency n, the network load pressure ratio A remains high, the compression ratio of the collected images is reduced between 1 / 14 and 1 / 18. If, after adjustment, the network load pressure ratio A is no greater than 1.8, the collected data is transmitted back. If the network load pressure ratio A remains greater than the second threshold of 1.8, it indicates that the current data collection task is under significant network transmission pressure, and continued data transmission could easily lead to data loss. Therefore, the current data collection task is abandoned. Data collection resumes when the network load pressure ratio A returns to a normal level. To avoid network transmission limitations and reduce the amount of data transmitted, when adjusting the compression ratio and frame rate of captured images, cross-adjustments should be made within a certain range to avoid continuously adjusting a single parameter, which could lead to low quality of the captured data.

[0083] The following are specific values ​​assigned to the above parameters, and the calculation and judgment process is explained as follows: When the vehicle speed is v = 60 km / h, the network transmission speed is a = 150 KB / s, and the memory size of a single image is b = 300 KB, according to formula (1) in the above high-precision map crowdsourcing collection method, when the input frame sampling frequency is n = 20 frames / second, the interval between two consecutive frames is d = (3.6 × 60) ÷ 20 = 10.8 meters < 15 meters. At this time, the next judgment is performed. When the input image compression ratio is c = 1 / 10, according to the above formula (3), the network load pressure ratio parameter A = (3.6 × 60 × 300 × 1 / 10) ÷ (10.8 × 150) = 4 > 1.8. Because it is greater than the preset first threshold, the data collection parameters need to be adjusted to reduce the amount of data transmission. At this point, the compression ratio c is reduced. When c decreases from 1 / 10 to 1 / 14, the network load pressure ratio parameter A = (3.6 × 60 × 300 × 1 / 14) ÷ (10.8 × 150) = 2.86 > 1.8 is calculated using the above formula (3). At this point, the frame dropping frequency n needs to be reduced. When n decreases from 20 to 12, the network load pressure ratio parameter A = (12 × 300 × 1 / 14) ÷ 150 = 1.71 < 1.8 can be calculated using the above formula (2). At this point, the current network meets the requirements for data transmission and will not cause data loss. At this point, the backhaul of road data can begin.

[0084] Figure 4 An embodiment of the high-precision map crowdsourcing data collection device of this application is shown.

[0085] exist Figure 4 In the illustrated embodiment, the high-precision map crowdsourcing data collection device of this application includes: a navigation unit 401, which acquires the trajectory information of the vehicle in real time and transmits the trajectory information to the data platform so that the data platform can issue data collection task instructions based on the trajectory information; the navigation unit includes a GNSS system and an inertial navigation system; a central processing unit 402, which controls the data collection process according to the data collection task instructions issued by the data platform; and a data acquisition camera 403, which performs road data collection according to the control instructions received from the central processing unit; wherein, the central processing unit is coupled to the data acquisition camera, and determines the network load pressure ratio parameter according to the data acquisition parameters corresponding to the data acquisition camera and the real-time network transmission speed, and judges the relationship between the network load pressure ratio parameter and a preset threshold, and selects an appropriate data return method to transmit the collected road data according to the judgment result; the data acquisition parameters include a preset image compression ratio and / or frame extraction frequency.

[0086] In this embodiment, the camera unit employs a high dynamic range (HDR) monocular camera, which reduces costs compared to a binocular camera. It also avoids the bottleneck problem of difficulty in improving the acquisition accuracy of binocular cameras. Furthermore, using a monocular device also reduces costs, truly realizing a consumer-grade crowdsourcing device. The real-time speed of the data acquisition vehicle is obtained through the inertial measurement unit and GNSS measurement unit in the navigation unit.

[0087] In one embodiment of this application, the central processing unit includes a multi-core central processing unit and a neural network processor unit, and the multi-core central processing unit includes a graphics processing unit and a video processor unit.

[0088] In one example of this application, the central processing unit may be a system-on-a-chip (SoC), which includes a central processing unit (CPU) and a neural network processor (NPU). The multi-core CPU includes a graphics processing unit (GPU) and a video processing unit (VPU).

[0089] In one example of this application, the acquisition camera may be equipped with an inertial measurement unit and a GNSS measurement unit to obtain the real-time speed of the acquisition vehicle where the acquisition camera is located. The high-precision map crowdsourcing acquisition device of this application may also include a communication unit connected to a system-on-a-chip (SoC) to transmit road image data wirelessly; and a storage unit connected to the SoC to store the SoC's set parameters and / or the acquired road image data. The storage unit may include an onboard hard disk, an external memory card, and / or a hard disk.

[0090] The execution process of the high-precision map crowdsourcing data collection device in this application is similar in principle to the aforementioned high-precision map crowdsourcing data collection method, and will not be elaborated upon here. The high-precision map crowdsourcing data collection device of this application, by setting the network load pressure ratio parameter, selects the corresponding data return method based on the relationship between the network load pressure ratio parameter and a preset threshold, avoiding data loss during transmission, ensuring data integrity, and guaranteeing real-time data transmission.

[0091] Figure 5 An example of the high-precision map crowdsourcing data collection device of this application is shown, such as... Figure 5As shown, the HDR monocular camera (High-Dynamic Range, or HDR) connects to the system-on-a-chip (SoC) via a MIPI connector. The navigation unit includes an Inertial Measurement Unit (IMU) and a GPS unit, connected to the SoC via a serial port. A 4G module serves as the communication module, connecting to the SoC to provide a network for data transmission. The storage unit includes an onboard eMMC (Embedded Multi Media Card) located on the SoC and an external TF card (TransFlash), connected to the SoC via a corresponding interface for data storage. A portable hard drive connects to the SoC via a USB 3.0 interface for offline data transfer. Additionally, the SoC connects to LPDDR4 (Low Power Double Data Rate 4) memory, widely used in mobile devices as "working memory," for data storage. In specific settings, the central processing unit (CPU) can employ a 6-core CPU and a neural network processing unit (NPU). The six-core CPU includes a graphics processing unit (GPU) and a video processing unit (VPU).

[0092] Figure 6 An example of the central processing unit of the high-precision map crowdsourcing data collection device of this application is shown.

[0093] like Figure 6 As shown, the central processing unit of the high-precision map crowdsourcing data collection device of this application mainly includes a data collection module, a decision module, and a feedback module. The data collection module primarily receives data collection task instructions from the data platform and controls the coupled-connected data collection cameras to collect road data. The decision module calculates the network load pressure ratio parameter based on data collection parameters during the road data collection process, real-time network transmission speed, and real-time vehicle speed, and controls the data collection process of the data collection module and the data transmission process of the feedback module based on the calculation results. The specific judgment process of the decision module and the calculation process of the network load pressure ratio parameter are the same as those in the high-precision map crowdsourcing data collection method of this application, and will not be repeated here.

[0094] This application's high-precision map crowdsourcing data collection device employs a multi-sensor fusion scheme integrating a monocular camera, an inertial vehicle unit (IMU), and a GNSS unit GPS satellite positioning system. This solves the problems of high cost and inconvenient installation associated with laser equipment, and the limited accuracy of binocular equipment, reducing the device's dependence on hardware and significantly lowering costs. Furthermore, the task triggering software module (SDK) in this application enables data collection task triggering and real-time data transmission, saving storage and data traffic costs, meeting the needs of high-precision map crowdsourcing data collection, improving the freshness of high-precision maps, and saving on high-precision map collection costs.

[0095] Figure 7 An implementation of the high-precision map crowdsourcing data collection platform of this application is shown.

[0096] like Figure 7 As shown, the high-precision map crowdsourcing data collection platform of this application includes: a data platform and collection equipment. The data platform receives trajectory information from vehicles and issues data collection task instructions based on the relationship between the coverage rate of the trajectory information and a preset coverage threshold. The collection equipment on the vehicle performs road data collection and data transmission according to the data collection task instructions. The collection equipment includes: a navigation unit, which acquires the trajectory information from the vehicle in real time and transmits the trajectory information to the data platform so that the data platform can issue data collection task instructions based on the trajectory information; the navigation unit includes a GNSS system and an inertial navigation system; a central processing unit, which controls the data collection process according to the data collection task instructions issued by the data platform; and a data collection camera, which performs road data collection according to the control instructions received from the central processing unit. The central processing unit is coupled to the data collection camera and determines the network load pressure ratio parameter based on the data collection parameters corresponding to the data collection camera and the real-time network transmission speed. It also judges the relationship between the network load pressure ratio parameter and a preset threshold, and selects an appropriate data transmission method to transmit the collected road data based on the judgment result. The data collection parameters include a preset image compression ratio and / or frame extraction frequency.

[0097] In this embodiment, the high-precision map crowdsourcing data collection platform of this application includes a data platform and collection devices. The collection devices are located on the vehicle-mounted end, collecting road data, while the data platform is a server-side component for data integration and processing. The data platform analyzes the vehicle-mounted trajectory information transmitted by the collection devices to determine whether to issue a data collection task instruction. If the data collection conditions are met, the collection devices on the vehicle-mounted end issue the data collection task instruction, and then proceed with subsequent data collection according to the instruction.

[0098] The vehicle-mounted data acquisition equipment includes a navigation unit, a central processing unit (CPU), and acquisition cameras. The navigation unit acquires the vehicle's trajectory information, providing data for the data platform to determine whether to collect data. Upon receiving a data acquisition task instruction, the CPU receives the instruction and controls the acquisition cameras to collect data. The CPU determines whether and how to transmit the collected road data back to the system based on data acquisition parameters and network transmission conditions. This ensures the integrity of the transmitted road data and prevents the loss of valuable data.

[0099] This high-precision map crowdsourcing data collection platform solution achieves load balancing in image transmission by adjusting the frame extraction frequency and image compression ratio. This avoids the loss of effective image data and also prevents redundant image data from consuming network channel load during backhaul, thereby enabling efficient real-time backhaul of large amounts of data.

[0100] In one specific embodiment of this application, a computer-readable storage medium stores computer instructions, wherein the computer instructions are operated to perform the high-precision map crowdsourcing data collection method described in any embodiment. The storage medium may be located directly in hardware, in a software module executed by a processor, or in a combination of both.

[0101] Software modules may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in this art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium.

[0102] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof. A general-purpose processor can be a microprocessor, but alternatively, it can be any conventional processor, controller, microcontroller, or state machine. The processor can also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors incorporating a DSP core, or any other such configuration. Alternatively, the storage medium can be integrated with the processor. The processor and storage medium can reside in an ASIC. The ASIC can reside in the user terminal. Alternatively, the processor and storage medium can reside as discrete components in the user terminal.

[0103] In one specific embodiment of this application, a computer device includes a processor and a memory, the memory storing computer instructions, wherein the processor operates the computer instructions to execute the high-precision map crowdsourcing data collection method described in any embodiment.

[0104] In the embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0105] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0106] The above are merely embodiments of this application and do not limit the scope of this patent application. Any equivalent structural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of this application.

Claims

1. A high-precision map crowd-sourcing collection method, characterized in that, The application relates to a data collection system for road data collection, comprising: a collection camera which collects road data according to a received data collection task instruction; a central processor which is coupled to the collection camera, determines a network load pressure ratio parameter according to data collection parameters corresponding to the collection camera and a real-time network transmission speed, judges a relationship between the network load pressure ratio parameter and a preset threshold, and selects a corresponding data return mode according to a result of the judgment. The central processor transmits the collected road data according to the result of the judgment.

2. The high-precision map crowd-sourcing collection method according to claim 1, characterized in that, The collection camera collects road data according to a received data collection task instruction, comprising: a navigation unit which is carried on a vehicle and acquires real-time trajectory information of the vehicle, and transmits the trajectory information to a data platform so that the data platform issues the data collection task instruction according to the trajectory information, wherein the navigation unit comprises a GNSS system and an inertial navigation system; the central processor controls the collection camera to collect road data according to the data collection task instruction issued by the data platform.

3. The high-precision map crowd-sourcing collection method of claim 1, wherein, The central processor which is coupled to the collection camera, determines a network load pressure ratio parameter according to data collection parameters corresponding to the collection camera and a real-time network transmission speed, comprising: the central processor determines a real-time driving speed of the vehicle according to a coupled navigation unit; the central processor acquires the data collection parameters and the real-time network transmission speed; the central processor calculates the real-time driving speed, the real-time network transmission speed and the data collection parameters according to a preset network load pressure ratio calculation formula to determine the network load pressure ratio parameter, wherein the network load pressure ratio parameter is positively correlated with the real-time driving speed, negatively correlated with the network transmission speed, positively correlated with a picture compression ratio and negatively correlated with a frame extraction frequency.

4. The high-precision map crowd-sourcing collection method of claim 1, wherein, The central processor selects a corresponding data return mode according to a result of the judgment to transmit the collected road data, comprising: if a first network load pressure ratio parameter is not greater than a first threshold, the collected road data is returned; if the first network load pressure ratio parameter is greater than the first threshold, the data collection parameters are adjusted, and a second network load pressure ratio parameter after adjustment is calculated; if the second network load pressure ratio parameter is not greater than a second threshold, the collected road data is returned; if the second network load pressure ratio parameter is greater than the second threshold, the data collection parameters are continuously adjusted, and a third network load pressure ratio parameter after adjustment is calculated; if the third network load pressure ratio parameter is not greater than the second threshold, the collected road data is returned, and if the third network load pressure ratio parameter is still greater than the second threshold, the data collection task is abandoned.

5. The high-precision map crowd-sourcing collection method according to claim 2, characterized in that, The data platform issues the data collection task instruction according to the trajectory information, comprising: the data platform issues the data collection task instruction according to a coverage rate of a relative geographic fence with respect to the trajectory information, wherein the data platform issues the data collection task instruction according to a coverage rate of a relative geographic fence with respect to the trajectory information, wherein publish the data collection task instruction when the coverage rate does not exceed the preset coverage rate threshold; publish the data collection task instruction when the coverage rate exceeds the preset coverage rate threshold.

6. The high-precision map crowd-sourcing collection method of claim 1, wherein, The adjustment range of the data collection parameter does not exceed a preset adjustment range.

7. A high-precision map crowd-sourcing collection device, characterized in that, Comprise: a navigation unit that acquires trajectory information of a vehicle in real time and transmits the trajectory information to a data platform, so that the data platform issues a data collection task instruction according to the trajectory information, the navigation unit comprising a GNSS system and an inertial navigation system; a central processing unit that controls a data collection process according to the data collection task instruction issued by the data platform; a collection camera that collects road data according to a control instruction received from the central processing unit; The central processing unit is coupled to the collection camera, determines a network load pressure ratio parameter according to a data collection parameter corresponding to the collection camera and a real-time network transmission speed, judges a relationship between the network load pressure ratio parameter and a preset threshold, and selects a corresponding data return mode for transmitting the collected road data according to a result of the judgment, the data collection parameter comprising a preset picture compression ratio and a frame extraction frequency.

8. The high-precision map crowdsourcing collection device of claim 7, wherein The central processing unit comprises a multi-core central processing unit and a neural network processing unit, and the multi-core central processing unit comprises a graphics processing unit and a video processing unit.

9. A high-precision map crowd-sourcing collection platform, characterized in that, Comprise: a data platform and a collection device, wherein The data platform receives trajectory information of a vehicle and issues a data collection task instruction according to a relationship between a coverage rate of the trajectory information and a preset coverage rate threshold; The collection device of the vehicle collects road data and returns data according to the data collection task instruction, wherein the collection device comprises a navigation unit that acquires trajectory information of a vehicle in real time and transmits the trajectory information to a data platform, so that the data platform issues a data collection task instruction according to the trajectory information, the navigation unit comprising a GNSS system and an inertial navigation system; a central processing unit that controls a data collection process according to the data collection task instruction issued by the data platform; a collection camera that collects road data according to a control instruction received from the central processing unit; The central processing unit is coupled to the collection camera, determines a network load pressure ratio parameter according to a data collection parameter corresponding to the collection camera and a real-time network transmission speed, judges a relationship between the network load pressure ratio parameter and a preset threshold, and selects a corresponding data return mode for transmitting the collected road data according to a result of the judgment, the data collection parameter comprising a preset picture compression ratio and a frame extraction frequency.

10. A computer-readable storage medium storing computer instructions, wherein the computer instructions are operated to perform the high-precision map crowdsourcing collection method of any one of claims 1-6.

Images