A positioning method and device of a vehicle, a vehicle, and a storage medium

Abstract

The application relates to the technical field of automatic driving, in particular to a positioning method and device of a vehicle, a vehicle and a storage medium. The method is applied to a vehicle terminal; the method comprises the following steps: acquiring a semantic image of a road environment image collected by a vehicle terminal at a target track point; extracting a plurality of compressed information from the semantic image; wherein each compressed information comprises part of image information in the semantic image; respectively matching each compressed information in the plurality of compressed information with a semantic map of a target region to obtain a plurality of predicted positioning information corresponding to the plurality of compressed information in the semantic map; wherein the target region is a region comprising the target track point; and determining positioning information of the target track point from the plurality of predicted positioning information. Therefore, the accuracy of vehicle positioning can be improved, and the cost of vehicle positioning can be reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving, in particular to the field of initial positioning of an automatic driving vehicle based on semantic map information, and specifically relates to a positioning method and device of a vehicle, a vehicle and a storage medium. BACKGROUND

[0002] With the development of high intelligence and integration of automatic driving technology, users have higher and higher requirements for the initial positioning and positioning tracking of an automatic driving vehicle. However, in actual application, the accuracy of positioning of the automatic driving vehicle is often not high, which brings a poor user experience.

[0003] To solve the above problems, the related technology proposes a method of combining a global navigation satellite system (GNSS) with a high-precision map obtained in advance to position the automatic driving vehicle. Although this method improves the accuracy of positioning of the automatic driving vehicle, it has a high dependence on high-precision maps and high-precision real-time kinematic (RTK) devices, resulting in a high cost of the automatic driving vehicle. SUMMARY

[0004] The present application provides a positioning method and device of a vehicle, a vehicle and a storage medium to at least solve the problem of high cost of a vehicle when performing accurate positioning. The technical solutions of the present application are as follows:

[0005] According to a first aspect of the present application, a positioning method of a vehicle is provided, applied to a vehicle terminal. The method comprises: obtaining a semantic image of a road environment image collected by the vehicle terminal at a target trajectory point; extracting a plurality of compressed information from the semantic image; wherein each compressed information includes part of the image information in the semantic image; respectively matching each compressed information in the plurality of compressed information with a semantic map of a target region to obtain a plurality of predicted positioning information corresponding to the plurality of compressed information in the semantic map; the target region is a region including the target trajectory point; and determining the positioning information of the target trajectory point from the plurality of predicted positioning information.

[0006] According to the technical means, compared with the technical solution of positioning the vehicle by relying on high-precision maps and high-precision RTK devices in the related art, the application provides a method that does not rely on high-precision maps and high-precision RTK devices. Specifically, the method can extract a plurality of compressed information (each compressed information includes part of the image information in the semantic image) from the semantic image of the road environment image collected by the vehicle terminal at the target trajectory point. In this way, the data consumption in the transmission process is reduced, the transmission efficiency is improved, and the power consumption of the central processing unit (CPU) of the vehicle is reduced. At the same time, the method matches each compressed information with the semantic map of the target area (including the area of the target trajectory point) to obtain a plurality of predicted positioning information. Then, the positioning information of the target trajectory point is determined from the plurality of predicted positioning information, which can improve the accuracy of vehicle positioning while reducing the cost of the vehicle.

[0007] In a possible implementation, the extracting the plurality of compressed information from the semantic image comprises: determining a compression coefficient of the semantic image according to sparsity of the semantic image; and determining a compression amount of the semantic image according to at least the compression coefficient, wherein the compression coefficient is directly proportional to the compression amount, and the compression amount is used to indicate a number of pixels that can be retained in the semantic image in a single compression process; and the random sampling method is used to extract image information satisfying the compression amount from the semantic image multiple times to generate the plurality of compressed information.

[0008] According to the technical means, the method provided by the application can use the random sampling method to extract image information satisfying the compression amount from a semantic image multiple times to generate a plurality of compressed information, which can ensure the independence and completeness of the image information satisfying the compression amount extracted from the semantic image each time, and facilitate subsequent optimization of positioning of the vehicle using the plurality of compressed information. At the same time, since each compressed information only includes part of the image information in the semantic image, the amount of data transmitted in the positioning process can be reduced, and the speed of positioning of the vehicle can be improved. At the same time, since the amount of data transmitted in the positioning process is small, the method provided by the application can reduce the power consumption of the CPU of the vehicle, so that the CPU of the vehicle can allocate more resources to other modules in the vehicle that consume more power.

[0009] In a possible implementation, the method further comprises: adjusting the compression coefficient according to a preset compression time, wherein the compression coefficient is positively correlated with the preset compression time; and determining the compression amount of the semantic image according to at least the adjusted compression coefficient.

[0010] According to the above technical means, it can be understood that the smaller the compression coefficient is, the smaller the compression amount of the semantic image is, and the smaller the amount of data transmitted in the positioning process is. The method provided by the application can flexibly adjust the compression coefficient according to the preset compression time, can improve the time required for the vehicle to be positioned, and can improve the use experience of the user.

[0011] In a possible implementation, the plurality of compressed information includes target compressed information; the target compressed information is any one of the plurality of compressed information; and the matching of each of the plurality of compressed information with the semantic map of the target area to obtain the plurality of predicted positioning information corresponding to the plurality of compressed information in the semantic map includes: decompressing the target compressed information to obtain target image information; matching the target image information with the semantic images of the plurality of position points in the semantic map of the target area to obtain a plurality of matching results; the matching result reflects the matching degree of the target image information and the semantic image of the position point in the semantic map; and the positioning information of the position point with the highest matching degree in the plurality of matching results is taken as the predicted positioning information of the target trajectory point.

[0012] According to the above technical means, compared with the related art, the use of a plurality of candidate semantic images for matching with the semantic map requires the use of a high-precision map and a high-precision RTK device to obtain a plurality of semantic images. The positioning method of the vehicle provided by the application only needs to obtain a plurality of compressed information generated from one semantic image and match each compressed information with the semantic map. The method can simplify the positioning process of the vehicle, reduce the dependence on the high-precision map and the high-precision RTK device, improve the positioning accuracy, and reduce the cost of vehicle positioning.

[0013] In a possible implementation, the determination of the positioning information of the target trajectory point from the plurality of predicted positioning information includes: determining an evaluation score of the plurality of predicted positioning information; the evaluation score is used to represent the reliability of each predicted positioning information; and the predicted positioning information with the highest evaluation score is taken as the positioning information of the target trajectory point.

[0014] According to the above technical means, the positioning method of the vehicle provided by the application can determine the positioning information of the target trajectory point from the plurality of predicted positioning information, without the aid of a high-precision map and a high-precision RTK device. The method can reduce the cost of vehicle positioning, improve the accuracy of vehicle positioning, and improve the use experience of the user.

[0015] In a possible implementation, the vehicle terminal comprises a positioning device; before matching each compressed information in the plurality of compressed information with the semantic map of the target region to obtain a plurality of predicted positioning information corresponding to the plurality of compressed information in the semantic map, the method further comprises: obtaining initial positioning information of the target trajectory point output by the positioning device; determining the target region according to the initial positioning information of the target trajectory point; and obtaining the semantic map of the target region.

[0016] According to the above technical means, the method provided by the application obtains the semantic map of the target region by the initial positioning information output by the positioning device, so as to facilitate matching the plurality of compressed information with the semantic map to position the vehicle on the basis of the semantic information.

[0017] According to a second aspect of the application, a positioning device of a vehicle is provided, and applied to a vehicle terminal; the positioning device comprises: an obtaining module, configured to obtain a semantic image of a road environment image collected by the vehicle terminal at a target trajectory point; an extracting module, configured to extract a plurality of compressed information from the semantic image; each compressed information comprises part of image information in the semantic image; a matching module, configured to match each compressed information in the plurality of compressed information with a semantic map of a target region respectively, to obtain a plurality of predicted positioning information corresponding to the plurality of compressed information in the semantic map; the target region is a region comprising the target trajectory point; and a determining module, configured to determine positioning information of the target trajectory point from the plurality of predicted positioning information.

[0018] In a possible implementation, the extracting module is specifically configured to determine a compression coefficient of the semantic image according to sparsity of the semantic image; determine a compression amount of the semantic image according to at least the compression coefficient; the compression coefficient is directly proportional to the compression amount; the compression amount is used to indicate a number of pixels that can be retained in the semantic image in a single compression process; and the extracting module is configured to generate the plurality of compressed information by extracting image information satisfying the compression amount from the semantic image multiple times by using a random sampling method.

[0019] In a possible implementation, the device further comprises an adjusting module; the adjusting module is configured to adjust the compression coefficient according to a preset compression time; the compression coefficient is positively correlated with the preset compression time; and the extracting module is specifically configured to determine the compression amount of the semantic image according to the adjusted compression coefficient.

[0020] In a possible implementation, the plurality of compressed information includes target compressed information; the target compressed information is any one of the plurality of compressed information; the matching module is specifically configured to decompress the target compressed information to obtain target image information; match the target image information with semantic images of a plurality of position points in the semantic map of the target region to obtain a plurality of matching results; the matching result reflects a matching degree of the target image information and the semantic image of the position point in the semantic map; and positioning information of a position point with the highest matching degree in the plurality of matching results is taken as the predicted positioning information of the target trajectory point.

[0021] In a possible implementation, the determining module is specifically configured to determine an evaluation score of the plurality of predicted positioning information; the evaluation score is used to represent a reliability degree of each predicted positioning information; and the predicted positioning information with the highest evaluation score is taken as the positioning information of the target trajectory point.

[0022] In a possible implementation, the vehicle terminal includes a positioning device; the obtaining module is further configured to obtain initial positioning information of the target trajectory point output by the positioning device; determine the target region according to the initial positioning information of the target trajectory point; and obtain the semantic map of the target region.

[0023] According to a third aspect of the present application, a vehicle is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executes the program to implement the method in the first aspect and any possible implementation thereof.

[0024] According to a fourth aspect of the present application, a computer readable storage medium is provided, when instructions in the computer readable storage medium are executed by a processor of an electronic device, the electronic device can execute the method in the first aspect and any possible implementation thereof.

[0025] Therefore, the above technical features of the present application have the following beneficial effects:

[0026] (1) Compared with the technical solution in the related art that relies on high-precision maps and high-precision RTK devices to position the vehicle, the method provided in the present application does not rely on high-precision maps and high-precision RTK devices, specifically, the method can extract a plurality of compressed information (each compressed information includes part of the image information in the semantic image) from the semantic image of the road environment image collected by the vehicle terminal at the target trajectory point, in this way of data compression, the data consumption in the transmission process is reduced, the transmission efficiency is improved, and then the power consumption of the central processing unit (CPU) of the vehicle can be reduced; at the same time, the method matches each compressed information with the semantic map of the target area (including the area of the target trajectory point) to obtain a plurality of predicted positioning information; and then determines the positioning information of the target trajectory point from the plurality of predicted positioning information, which can improve the accuracy of vehicle positioning while reducing the cost of the vehicle.

[0027] (2) The method provided in the present application can use a random sampling method to extract image information satisfying the compression amount from a semantic image multiple times to generate a plurality of compressed information, which can ensure the independence and completeness of the image information satisfying the compression amount extracted from the semantic image each time, facilitating subsequent optimization of the positioning of the vehicle using the plurality of compressed information. At the same time, since each compressed information only includes part of the image information in the semantic image, the amount of data transmitted in the positioning process can be reduced, and the speed of positioning of the vehicle can be improved; at the same time, since the amount of data transmitted in the positioning process is small, the method provided in the present application can reduce the power consumption of the CPU of the vehicle, so that the CPU of the vehicle can allocate more resources to other modules in the vehicle that consume more energy.

[0028] (3) It can be understood that the smaller the compression coefficient, the smaller the compression amount of the semantic image, and the smaller the amount of data transmitted in the positioning process; the method provided in the present application can flexibly adjust the compression coefficient according to the preset compression time, which can improve the time required for the vehicle to position and improve the user experience.

[0029] (4) Compared with the related art, using a plurality of candidate semantic images to match with the semantic map requires using high-precision maps and high-precision RTK devices to obtain a plurality of semantic images in advance, the positioning method of the vehicle provided in the present application only needs to obtain a plurality of compressed information generated from one semantic image and match each compressed information with the semantic map, which can simplify the process of positioning the vehicle, reduce the dependence on high-precision maps and high-precision RTK devices, improve the positioning accuracy, and reduce the cost of vehicle positioning.

[0030] (5) The positioning method of the vehicle provided in the application determines the positioning information of the target track point from the plurality of predicted positioning information, without the aid of high-precision maps and high-precision RTK devices, so as to reduce the positioning cost of the vehicle, improve the accuracy of the positioning of the vehicle, and improve the user experience.

[0031] (6) The method provided in the application acquires the semantic map of the target area through the initial positioning information output by the positioning device, so as to match the plurality of compressed information with the semantic map on the basis of the semantic information to position the vehicle.

[0032] It should be noted that the technical effects brought by any one of the implementation manners in the second to fourth aspects can refer to the technical effects brought by the corresponding implementation manners in the first aspect, which will not be repeated here.

[0033] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the application. BRIEF DESCRIPTION OF DRAWINGS

[0034] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the application and, together with the specification, serve to explain the principles of the application, and do not constitute an improper limitation on the application.

[0035] Figure 1 is a flowchart of a positioning method of a vehicle according to an exemplary embodiment;

[0036] Figure 2 is a flowchart of another positioning method of a vehicle according to an exemplary embodiment;

[0037] Figure 3 is a flowchart of another positioning method of a vehicle according to an exemplary embodiment;

[0038] Figure 4 is a flowchart of another positioning method of a vehicle according to an exemplary embodiment;

[0039] Figure 5 is a flowchart of another positioning method of a vehicle according to an exemplary embodiment;

[0040] Figure 6 is a flowchart of another positioning method of a vehicle according to an exemplary embodiment;

[0041] Figure 7 is an example diagram of a target compressed image according to an exemplary embodiment;

[0042] Figure 8 is an example diagram of a related peak map according to an exemplary embodiment;

[0043] Figure 9 is a flow chart of another positioning method of a vehicle according to an example embodiment;

[0044] Figure 10 is a flow chart of another positioning method of a vehicle according to an example embodiment;

[0045] Figure 11 is a structure diagram of a positioning device of a vehicle according to an example embodiment;

[0046] Figure 12 is a structure diagram of a vehicle according to an example embodiment.

[0047] Wherein, the positioning device 300 of the vehicle, the acquisition module 301, the extraction module 302, the matching module 303, the determination module 304, the adjustment module 305, the vehicle 400, the processor 401, the memory 402. DETAILED DESCRIPTION

[0048] In order to make the ordinary person skilled in the art better understand the technical solutions of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings.

[0049] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all the embodiments consistent with the present application. Rather, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0050] A positioning method of a vehicle, a positioning device of a vehicle, a vehicle and a storage medium of an embodiment of the present application are described below with reference to the drawings.

[0051] The automatic driving technology has gradually become a research hotspot and focus in the fields of industry and people's livelihood, etc. due to its high intelligence and integration. Among them, the map information is an important reference for the automatic driving vehicle when driving, is the data basis for the operation control and intelligent analysis of the automatic driving vehicle, and is also the data basis for the positioning initialization and repositioning of the automatic driving vehicle. However, in actual application, the accuracy of the positioning of the automatic driving vehicle is often not high, which brings a bad user experience. In view of the above problem, the related technology proposes a method of combining GNSS with a high-precision map obtained in advance to position the automatic driving vehicle. Although this method improves the accuracy of the positioning of the automatic driving vehicle, it depends on high-precision maps and high-precision RTK devices, and has a high cost requirement for the automatic driving vehicle.

[0052] With the importance of semantic maps in the automatic driving technology, the development of various technologies based on semantic maps is becoming increasingly important. Positioning vehicles based on semantic maps is also an important means to reduce the cost of vehicles, improve the popularity of vehicles and market competitiveness. Therefore, in view of the above problem, the present application provides a vehicle positioning method. Compared with the technical solution of positioning vehicles by relying on high-precision maps and high-precision RTK devices in the related art, the present application provides a method that does not rely on high-precision maps and high-precision RTK devices. Specifically, the method can extract a plurality of compressed information (each compressed information includes part of the image information in the semantic image) from the semantic image of the road environment image collected by the vehicle terminal at the target trajectory point. In this way, the data consumption in the transmission process is reduced, and the transmission efficiency is improved. At the same time, the method matches each compressed information with the semantic map of the target area (including the area of the target trajectory point) to obtain a plurality of predicted positioning information. Then, the positioning information of the target trajectory point is determined from the plurality of predicted positioning information, which can improve the accuracy of vehicle positioning while reducing the cost of vehicles.

[0053] For ease of understanding, the embodiments of the present application are specifically introduced below with reference to the accompanying drawings.

[0054] Figure 1 A vehicle positioning method provided by an embodiment of the present application includes the following steps:

[0055] S101, acquiring a semantic image of a road environment image collected by a vehicle terminal at a target trajectory point.

[0056] In some embodiments, the vehicle terminal collects a road environment image at a target trajectory point, and performs semantic segmentation and mapping on the road environment image to obtain a corresponding semantic image.

[0057] S102, extracting a plurality of compressed information from the semantic image.

[0058] Each compressed information includes a portion of the image information from the semantic image.

[0059] For example, such as Figure 2 As shown, step S102 can be specifically implemented as steps S1021-S1023.

[0060] S1021. Determine the compression coefficient of the semantic image based on its sparsity.

[0061] In some embodiments, the compression coefficient of a semantic image can be determined based on the density of information in the semantic image. Since semantic images are generally sparse, an appropriate compression coefficient can be selected based on the sparsity ratio of the semantic image before and after compression.

[0062] For example, the compression factor r can range from 0.35 to 0.5.

[0063] S1022. Determine the amount of compression of the semantic image based at least on the compression coefficient.

[0064] The compression factor is proportional to the compression amount; the compression amount indicates the number of pixels that the semantic image can retain in a single compression process.

[0065] For example, the smaller the compression coefficient, the smaller the compression amount, and the smaller the number of pixels retained in a single compression process for the semantic image.

[0066] In some embodiments, the compression factor r is multiplied by the number of rows and columns of pixels in the semantic image to obtain the compression amount of the semantic image.

[0067] For example, if the number of rows of pixels in a semantic image is 1000 and the number of columns is 2000, and the selected compression coefficient is 0.5, then the compression amount of the semantic image is: 0.5 * 1000 * 2000 = 1000000, that is, the number of pixels retained in the semantic image during a single compression process is 1000000.

[0068] Understandably, the choice of compression factor affects the time consumed during vehicle positioning and the final positioning accuracy. The smaller the compression factor, the smaller the amount of data transmitted during positioning, the faster the transmission process, and the less time is consumed in positioning. At the same time, the smaller the compression factor, the more times the data is transmitted during positioning in the same amount of time, resulting in a more accurate final positioning result.

[0069] As one possible implementation, the compression coefficient can be adjusted according to the preset compression time and / or preset compression precision, and the amount of compression of the semantic image can be determined according to the adjusted compression coefficient.

[0070] The compression coefficient is positively correlated with the preset compression time; for example, the shorter the preset compression time, the smaller the selected compression coefficient. It is understandable that a shorter preset compression time means less time required for vehicle positioning. Therefore, the compression coefficient can be flexibly adjusted to reduce the time required for vehicle positioning. For example, when there is a time limit for vehicle positioning, the compression coefficient can be decreased proportionally to reduce the computational cost of vehicle positioning and shorten the positioning time.

[0071] The compression coefficient is positively correlated with the preset compression accuracy. For example, the smaller the compression coefficient, the less data is transmitted during the positioning process, and the shorter the transmission time per data transfer. Given the same vehicle positioning time, more data transmissions during the positioning process result in more accurate vehicle positioning. Therefore, the compression coefficient can be flexibly adjusted to improve vehicle positioning accuracy.

[0072] S1023. Using a random sampling method, image information that meets the compression requirements is extracted from the semantic image multiple times to generate multiple compressed information.

[0073] In some embodiments, a random sampling method is used to construct a coefficient distribution matrix for compression, and the semantic image is compressed using the coefficient distribution matrix. Image information that meets the compression amount is extracted from the semantic image, and then compressed information is generated.

[0074] The coefficient distribution matrix is ​​used to characterize the retention status of each pixel in the semantic image. For example, when the element value of the first row and first column of the coefficient distribution matrix is ​​1, it means that the pixel information of the first row and first column of the semantic image is retained; when the element value of the first row and first column of the coefficient distribution matrix is ​​0, it means that the pixel information of the first row and first column of the semantic image is discarded.

[0075] Among them, the random sampling method can be a sampling method based on compressed sensing.

[0076] As a possible implementation, unlike related technologies that use the Nyquist sampling method to sample images, this application employs a non-Nyquist sampling method for random sampling. Furthermore, the random sampling method can select different random sampling strategies based on the sparsity of the semantic image. For example, a relevant random sampling strategy can be applied only to the non-zero pixel locations in the semantic image, or a global random sampling strategy can be used to randomly sample the pixel information of the entire semantic image.

[0077] Meanwhile, due to the randomness of the random sampling method, the coefficient distribution matrix constructed each time is different. Therefore, the random sampling method can be sampled to construct different coefficient distribution matrices multiple times, and the coefficient distribution matrices can be used to extract image information that meets the compression requirements from the semantic image multiple times, thereby generating multiple different compressed information.

[0078] In some embodiments, such as Figure 3 As shown, the above steps S1021-S1023 can be specifically implemented as follows:

[0079] Step a1: Input semantic image.

[0080] Step a2: Determine the compression amount of the semantic image based on the compression coefficient.

[0081] Step a3: Use a random sampling method to construct multiple coefficient distribution matrices for compression.

[0082] Step a4: Record the row and column numbers corresponding to the non-zero elements in each coefficient distribution matrix.

[0083] Step a5: Output the pixel information of multiple semantic images based on the row and column numbers corresponding to the non-zero elements in each coefficient distribution matrix.

[0084] Step a6: Generate multiple compressed information based on the pixel information of multiple semantic images.

[0085] Understandably, the method provided in this application can employ random sampling to extract image information that meets the compression requirements from a semantic image multiple times, generating multiple compressed information sets. This ensures the independence and completeness of the compressed image information extracted from the semantic image each time, facilitating subsequent optimization of vehicle positioning using multiple compressed information sets. Furthermore, since each compressed information set includes only a portion of the image information from the semantic image, the amount of data transmitted during positioning can be reduced, increasing the speed of vehicle positioning. Simultaneously, due to the smaller amount of data transmitted during positioning, the method provided in this application can reduce the power consumption of the vehicle's CPU, allowing the vehicle's CPU to allocate more resources to other energy-intensive modules within the vehicle.

[0086] S103. Match each of the multiple compressed information with the semantic map of the target area to obtain multiple predicted positioning information corresponding to the multiple compressed information in the semantic map.

[0087] The target area is the region that includes the target trajectory points.

[0088] In some embodiments, such as Figure 4 As shown, before step S103 above, the method also includes steps S201-S203.

[0089] S201. Obtain the initial positioning information of the target trajectory point output by the positioning device.

[0090] Specifically, the initial positioning information of the target trajectory points is obtained based on the Global Positioning System (GPS) signal values.

[0091] S202. Determine the target area based on the initial positioning information of the target trajectory points.

[0092] The target area can be a circular region centered on the target trajectory point and with a preset distance as its radius. The preset distance can be flexibly selected based on actual road conditions and vehicle movement; this embodiment does not limit the specific value of the preset distance. For example, the preset distance can be 10 meters.

[0093] S203. Obtain the semantic map of the target area.

[0094] It is understood that the method provided in this application obtains a semantic map of the target area through the initial positioning information output by the positioning device, which facilitates the matching of multiple compressed information with the semantic map based on the semantic information to locate the vehicle.

[0095] In some embodiments, the plurality of compression information includes target compression information; the target compression information is any one of the plurality of compression information. The method provided in step S103 above will be described below using target compression information as an example. For example, as shown below… Figure 5 As shown, step S103 above can be specifically implemented as follows:

[0096] S1031. Decompress the target compressed information to obtain the target image information.

[0097] In some embodiments, a fast least-squares matching algorithm is used to recover the target compressed information into an image sequence, and the image sequence is rearranged to obtain the target image information.

[0098] Understandably, the significance of the least squares matching algorithm lies in solving the similarity between the corresponding points of the target compressed information and the semantic image, and extracting the corresponding sample values ​​from the relatively sparse target compressed information to obtain the target image information.

[0099] For example, such as Figure 6 As shown, step S1031 above can be specifically implemented as follows:

[0100] Step b1: Obtain the length, width, and compression coefficient of the semantic image.

[0101] Step b2: Use a fast least squares matching algorithm to restore the target compressed information into an image sequence.

[0102] Step b3: Rearrange the image sequence according to the length, width and compression coefficient of the semantic image.

[0103] Step b4: Obtain the target image information based on the rearranged image sequence.

[0104] In some embodiments, such as Figure 7 The image shows multiple compressed target images obtained using the least squares matching algorithm with different compression coefficients. Among them, compression recovery... Figure 1 For the first target compressed image, compression recovery Figure 2 For the second target compressed image, compression recovery Figure 3 The third target compressed image shows that, in addition to the corresponding noise, the target image information also retains certain semantic information.

[0105] Optionally, since the target image information may contain some noise due to the optimization solution using a fast least squares matching algorithm, a suitable random sampling strategy can be selected during the compression process to reduce the noise, or the target image information can be filtered and denoised.

[0106] S1032. Match the target image information with the semantic images of multiple location points in the semantic map of the target area to obtain multiple matching results.

[0107] The matching result reflects the degree of matching between the target image information and the semantic image of the location point in the semantic map.

[0108] In some embodiments, the SSSIG matching algorithm, based on the average gray gradient value of the sub-region, is used to match the target image information with the semantic images of multiple location points in the semantic map of the target region, resulting in multiple matching results.

[0109] Among them, such as Figure 8 As shown, the matching result can be represented as a correlation peak map. In the correlation peak map, the value corresponding to each location point is the correlation coefficient between the target image information and the semantic image of each location point in the semantic map of the target region. The larger the correlation coefficient, the higher the similarity between the target image information and the semantic image of that location point.

[0110] Understandably, compared to related technologies that use multiple candidate semantic images to match semantic maps, which require pre-acquiring multiple semantic images using high-precision maps and high-precision RTK devices, the vehicle localization method provided in this application only needs to obtain multiple compressed information from a single semantic image and match each compressed information with a semantic map. This simplifies the vehicle localization process, reduces reliance on high-precision maps and high-precision RTK devices, improves localization accuracy, and lowers the cost of vehicle localization.

[0111] S1033. Use the location information of the location point with the highest matching degree among multiple matching results as the predicted location information of the target trajectory point.

[0112] In some embodiments, the location point corresponding to the peak in the relevant peak map is the location point with the highest matching degree among multiple matching results, and the positioning information of the location point corresponding to the peak is used as the predicted positioning information of the target trajectory point.

[0113] As one possible implementation, if there are obvious peaks in the correlation peak map obtained by the SSSIG matching algorithm, the location information of the points corresponding to the peaks in the correlation peak map can be directly obtained as the predicted location information of the target trajectory points.

[0114] Another possible implementation is that if there are no obvious peaks or multiple peaks in the correlation peak map obtained by the SSSIG matching algorithm, it may be due to problems such as shooting angle, repetition of semantic map and vehicle pose change. In this case, a second-order matching function can be added to the SSSIG matching algorithm, and the coefficients of scaling and rotation terms should be added to deal with the gray background problem, shooting angle problem and scaling and rotation problem of the corresponding semantic image, so as to ensure the stability of the peaks in the correlation peak map.

[0115] Understandably, second-order matching functions can ignore the deformation in the x and y directions, that is, set the element value to 0 at the corresponding positions.

[0116] It should be noted that for each of the multiple compressed information, the predicted positioning information can be obtained according to the method provided in steps S1031-S1033 above.

[0117] It is understood that the vehicle positioning method provided in this application, by determining the positioning information of the target trajectory point from multiple predicted positioning information, can reduce the cost of vehicle positioning, improve the accuracy of vehicle positioning, and enhance the user experience without relying on high-precision maps and high-precision RTK equipment.

[0118] S104. Determine the positioning information of the target trajectory point from multiple predicted positioning information.

[0119] In some embodiments, such as Figure 9 As shown, the above step S104 can be specifically implemented as the following steps S1041-S1042.

[0120] S1041. Determine the evaluation scores for multiple predicted positioning information.

[0121] The evaluation score is used to characterize the reliability of each predicted location information.

[0122] In some embodiments, the maximum likelihood method is used to evaluate multiple predicted positioning information.

[0123] For example, using a positive and negative 3σ principle, evaluation scores for multiple predicted positioning information are determined based on maximum likelihood estimation. The higher the evaluation score of the predicted positioning information, the greater the probability that the predicted positioning information is the positioning information of the target trajectory point.

[0124] S1042. Use the predicted positioning information with the highest evaluation score as the positioning information of the target trajectory point.

[0125] In some embodiments, the method further includes: repeating steps S101-S104 according to a preset number of iterations to obtain multiple positioning information of the target trajectory point, and then determining the final positioning information of the target trajectory point from the multiple positioning information of the target trajectory point through maximum likelihood estimation.

[0126] The compression coefficient selected is different each time steps S101-S104 are executed.

[0127] The preset number of loop calculations can be 5 to 10.

[0128] Optionally, if there is a time constraint on the vehicle positioning process, maximum likelihood estimation can be omitted. Instead, the average of multiple positioning information of the target trajectory point can be calculated, and the average of these multiple positioning information can be used as the final positioning information of the target trajectory point. If the average value method is used, the same compression coefficient must be used each time steps S101-S104 are executed.

[0129] The above are embodiments of the vehicle positioning method provided in this application. For ease of understanding, the above vehicle positioning method will be further explained below with examples.

[0130] For example, such as Figure 10 As shown, the vehicle positioning method provided in this application mainly includes the following steps:

[0131] Step c1: Perform semantic segmentation on the road environment images collected by the vehicle terminal at the target trajectory point to obtain the corresponding semantic images.

[0132] Step c2: Select the compression factor.

[0133] Step c3: Based on the compressed sensing method, obtain multiple coefficient distribution matrices.

[0134] Step c4: Based on the coefficient distribution matrix, extract multiple compressed information from the semantic image; wherein, the multiple compressed information includes target compressed information; the target compressed information is any one of the multiple compressed information.

[0135] Step c5: Based on the GPS signal, obtain a semantic map of the target area where the target trajectory point is located.

[0136] Step c6: Based on the least squares matching algorithm, obtain the target image information corresponding to the target compression information.

[0137] Step c7: Using the SSSIG matching algorithm, the target image information is matched with the semantic images of multiple location points in the semantic map of the target area to obtain multiple matching results.

[0138] Step c8: Determine whether the relevant peak maps corresponding to the above multiple matching results have a unique peak. If so, proceed to step c9 below; otherwise, proceed to step c10 below.

[0139] Step c9: Use the location information of the location points corresponding to the peak values ​​in the relevant peak map as the predicted location information of the target trajectory points.

[0140] Step c10: Add a second-order matching function to the SSSIG matching algorithm to rematch the target image information with the semantic images of multiple location points in the semantic map of the target region, and obtain multiple matching results.

[0141] Step c11: Determine whether the preset number of iterations has been reached. If yes, proceed to step c12 below; otherwise, proceed to step c3 above.

[0142] Step c12: Based on maximum likelihood estimation, determine the positioning information of the target trajectory point from multiple predicted positioning information.

[0143] The foregoing mainly describes the solutions provided by the embodiments of this application from a methodological perspective. To achieve the above functions, the vehicle positioning device or electronic device includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0144] This application embodiment can, according to the above method, exemplarily divide a vehicle positioning device or electronic device into functional modules. For example, the vehicle positioning device or electronic device may include functional modules corresponding to each functional division, or two or more functions may be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division; in actual implementation, there may be other division methods.

[0145] Figure 11 This is a block diagram illustrating a vehicle positioning device according to an exemplary embodiment, applied to a vehicle terminal. The vehicle positioning device 300 includes: an acquisition module 301, an extraction module 302, a matching module 303, a determination module 304, and an adjustment module 305.

[0146] The acquisition module 301 is used to acquire semantic images of road environment images collected by the vehicle terminal at the target trajectory point.

[0147] Extraction module 302 is used to extract multiple compressed information from the semantic image; wherein each compressed information includes partial image information in the semantic image.

[0148] The matching module 303 is used to match each of the multiple compressed information with the semantic map of the target area to obtain multiple predicted positioning information corresponding to the multiple compressed information in the semantic map; wherein, the target area is the area including the target trajectory points.

[0149] The determination module 304 is used to determine the positioning information of the target trajectory point from multiple predicted positioning information.

[0150] In one possible implementation, the extraction module 301 is specifically used to determine the compression coefficient of the semantic image based on the sparsity of the semantic image; and at least based on the compression coefficient, determine the compression amount of the semantic image; wherein the compression coefficient is proportional to the compression amount; the compression amount is used to indicate the number of pixels that the semantic image can retain in a single compression process; and using a random sampling method, extracting image information that meets the compression amount from the semantic image multiple times to generate multiple compression information.

[0151] In one possible implementation, the adjustment module 305 is used to adjust the compression coefficient according to a preset compression time; wherein the compression coefficient is positively correlated with the preset compression time. The extraction module 301 is specifically used to determine the compression amount of the semantic image based on the adjusted compression coefficient.

[0152] In one possible implementation, the multiple compressed information includes target compressed information; the target compressed information is any one of the multiple compressed information; the matching module 303 is specifically used to decompress the target compressed information to obtain target image information; match the target image information with the semantic images of multiple location points in the semantic map of the target region to obtain multiple matching results; the matching results reflect the matching degree between the target image information and the semantic images of the location points in the semantic map; and use the location information of the location point with the highest matching degree among the multiple matching results as the predicted location information of the target trajectory point.

[0153] In one possible implementation, the determining module 304 is specifically used to determine the evaluation scores of multiple predicted positioning information; the evaluation scores are used to characterize the reliability of each predicted positioning information; and the predicted positioning information with the highest evaluation score is used as the positioning information of the target trajectory point.

[0154] In one possible implementation, the vehicle terminal includes a positioning device; the acquisition module 301 is further configured to acquire initial positioning information of the target trajectory points output by the positioning device; determine the target area based on the initial positioning information of the target trajectory points; and acquire a semantic map of the target area.

[0155] Based on the aforementioned technical means, compared to related technologies that rely on high-precision maps and high-precision RTK equipment for vehicle positioning, this application provides a method that does not rely on high-precision maps and high-precision RTK equipment. Specifically, this method can extract multiple compressed information (each compressed information includes a portion of the image information in the semantic image) from the semantic image of the road environment image collected by the vehicle terminal at the target trajectory point. This data compression reduces data consumption during transmission, improves transmission efficiency, and thus reduces the power consumption of the vehicle's central processing unit (CPU). Simultaneously, this method obtains multiple predicted positioning information by matching each compressed information with the semantic map of the target area (including the area of ​​the target trajectory point). Then, the positioning information of the target trajectory point is determined from these multiple predicted positioning information, which can improve the accuracy of vehicle positioning while reducing vehicle costs.

[0156] Meanwhile, the method provided in this application can employ random sampling to extract image information that meets the compression requirements from a semantic image multiple times, generating multiple compressed information sets. This ensures the independence and completeness of the compressed image information extracted from the semantic image each time, facilitating subsequent optimization of vehicle positioning using multiple compressed information sets. Furthermore, since each compressed information set includes only a portion of the image information from the semantic image, the amount of data transmitted during positioning can be reduced, increasing the speed of vehicle positioning. Simultaneously, due to the smaller amount of data transmitted during positioning, the method provided in this application can reduce the power consumption of the vehicle's CPU, allowing the vehicle's CPU to allocate more resources to other energy-intensive modules within the vehicle.

[0157] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0158] Figure 12 This is a block diagram illustrating a vehicle according to an exemplary embodiment. Figure 12 As shown, vehicle 400 includes, but is not limited to, processor 401 and memory 402.

[0159] The memory 402 described above is used to store the executable instructions of the processor 401. It is understood that the processor 401 is configured to execute instructions to implement the vehicle positioning method in the above embodiments.

[0160] It should be noted that those skilled in the art will understand that Figure 12 The vehicle structure shown does not constitute a limitation on the vehicle; a vehicle may include, but is not limited to, other types of vehicles. Figure 12 This may indicate more or fewer components, or combinations of certain components, or different component arrangements.

[0161] The processor 401 is the control center of the vehicle, connecting various parts of the vehicle through various interfaces and lines. It performs various vehicle functions and processes data by running or executing software programs and / or modules stored in the memory 402, and by calling data stored in the memory 402, thereby providing overall vehicle monitoring. The processor 401 may include one or more processing units. Optionally, the processor 401 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, while the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 401.

[0162] The memory 402 can be used to store software programs and various data. The memory 402 may primarily include a program storage area and a data storage area. The program storage area may store the operating system, application programs required by at least one functional module (such as a determination unit, processing unit, etc.), etc. Furthermore, the memory 402 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0163] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory 402 including instructions, which can be executed by a processor 401 of a vehicle 400 to implement the vehicle positioning method in the above embodiments.

[0164] In actual implementation, Figure 11 The functions of the acquisition module 301, extraction module 302, matching module 303, determination module 304, and adjustment module 305 can all be provided by... Figure 12 The processor 401 calls the computer program stored in the memory 402 to implement the process. The specific execution process can be found in the description of the vehicle positioning method in the previous embodiment, and will not be repeated here.

[0165] Optionally, the computer-readable storage medium may be a non-transitory computer-readable storage medium, such as a read-only memory (ROM), random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device.

[0166] In an exemplary embodiment, this application also provides a computer program product including one or more instructions, which can be executed by the vehicle's processor 401 to complete the vehicle positioning method in the above embodiments.

[0167] It should be noted that when one or more instructions in the computer-readable storage medium or computer program product are executed by the vehicle's processor, they implement the various processes of the above-described vehicle positioning method embodiments and achieve the same technical effects as the above-described vehicle positioning method. To avoid repetition, these will not be elaborated here.

[0168] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0169] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or 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 apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0170] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the classified units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0171] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0172] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, essentially, or the part that contributes to the prior art, or a complete or partial classification of the technical solution, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0173] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for locating a vehicle, characterized in that, Applied to a vehicle terminal; the method includes: Obtain the semantic image of the road environment image collected by the vehicle terminal at the target trajectory point; The compression coefficient of the semantic image is determined based on its sparsity. The compression amount of the semantic image is determined at least based on the compression coefficient; wherein the compression coefficient is proportional to the compression amount; the compression amount is used to indicate the number of pixels that the semantic image can retain in a single compression process; A random sampling method is used to extract image information that meets the compression amount from the semantic image multiple times, generating multiple compressed information; wherein each compressed information includes a portion of the image information in the semantic image; Each of the multiple compressed information pieces is matched with a semantic map of the target region to obtain multiple predicted positioning information pieces corresponding to the multiple compressed information pieces in the semantic map; wherein, the target region is the region including the target trajectory point; The positioning information of the target trajectory point is determined from the plurality of predicted positioning information.

2. The method according to claim 1, characterized in that, The method further includes: The compression coefficient is adjusted according to a preset compression time; wherein the compression coefficient is positively correlated with the preset compression time. Determining the compression amount of the semantic image based at least on the compression coefficient includes: The amount of compression of the semantic image is determined based on the adjusted compression coefficient.

3. The method according to claim 1, characterized in that, The plurality of compressed information includes target compressed information; the target compressed information is any one of the plurality of compressed information; the step of matching each of the plurality of compressed information with a semantic map of the target region to obtain plurality of predicted localization information corresponding to the plurality of compressed information in the semantic map includes: The target compressed information is decompressed to obtain the target image information; The target image information is matched with the semantic images of multiple location points in the semantic map of the target region to obtain multiple matching results; wherein, the matching results reflect the degree of matching between the target image information and the semantic images of location points in the semantic map; The location information of the position point with the highest matching degree among the multiple matching results is used as the predicted location information of the target trajectory point.

4. The method according to claim 1, characterized in that, Determining the positioning information of the target trajectory point from the plurality of predicted positioning information includes: Determine an evaluation score for the plurality of predicted location information; the evaluation score is used to characterize the reliability of each of the predicted location information. The predicted positioning information with the highest evaluation score is used as the positioning information of the target trajectory point.

5. The method according to claim 1, characterized in that, The vehicle terminal includes a positioning device; before matching each of the plurality of compressed information with a semantic map of the target area to obtain the plurality of predicted positioning information corresponding to the plurality of compressed information in the semantic map, the method further includes: Obtain the initial positioning information of the target trajectory point output by the positioning device; The target area is determined based on the initial positioning information of the target trajectory points; Obtain the semantic map of the target region.

6. A vehicle positioning device, characterized in that, Applied to vehicle terminals; including: The acquisition module is used to acquire semantic images of road environment images collected by the vehicle terminal at the target trajectory point; An extraction module is configured to determine the compression coefficient of the semantic image based on its sparsity; and to determine the compression amount of the semantic image based at least on the compression coefficient; wherein the compression coefficient is proportional to the compression amount; the compression amount indicates the number of pixels that the semantic image can retain in a single compression process; and to extract image information satisfying the compression amount multiple times from the semantic image using a random sampling method to generate multiple compression information; wherein each compression information includes a portion of the image information in the semantic image. The matching module is used to match each of the multiple compressed information with the semantic map of the target region to obtain multiple predicted positioning information corresponding to the multiple compressed information in the semantic map; wherein, the target region is the region including the target trajectory point; The determination module is used to determine the positioning information of the target trajectory point from the plurality of predicted positioning information.

7. The apparatus according to claim 6, characterized in that, The device also includes an adjustment module; The adjustment module is used to adjust the compression coefficient according to a preset compression time; wherein the compression coefficient is positively correlated with the preset compression time. The extraction module is specifically used to determine the compression amount of the semantic image based on the adjusted compression coefficient.

8. The apparatus according to claim 6, characterized in that, The plurality of compressed information includes target compressed information; the target compressed information is any one of the plurality of compressed information. The matching module is specifically used to decompress the target compressed information to obtain target image information; match the target image information with the semantic images of multiple location points in the semantic map of the target region to obtain multiple matching results; the matching results reflect the degree of matching between the target image information and the semantic images of location points in the semantic map; The location information of the position point with the highest matching degree among the multiple matching results is used as the predicted location information of the target trajectory point.

9. The apparatus according to claim 6, characterized in that, The determining module is specifically used to determine the evaluation scores of the plurality of predicted positioning information; the evaluation scores are used to characterize the reliability of each of the predicted positioning information; and the predicted positioning information with the highest evaluation score is used as the positioning information of the target trajectory point.

10. The apparatus according to claim 6, characterized in that, The vehicle terminal includes a positioning device; The acquisition module is further configured to acquire the initial positioning information of the target trajectory point output by the positioning device; determine the target area based on the initial positioning information of the target trajectory point; and acquire the semantic map of the target area.

11. A vehicle, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the vehicle positioning method as described in any one of claims 1 to 5.

12. A computer-readable storage medium, characterized in that, When the computer-executable instructions stored in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device is capable of performing the vehicle positioning method as described in any one of claims 1 to 5.

Images