Vehicle positioning method and device integrating semantic map and road information

Abstract

The present invention discloses a vehicle positioning method and device integrating semantic map and road information, the method steps include: S01. Acquire a road environment image, segment the road area and project it into a grid map under the vehicle body coordinate system; S02. Find the corresponding positioning point on the semantic map according to the initial positioning information currently acquired by the vehicle, and search for the nearest road semantic area to obtain the semantic map area; S03. Take the road area as the center and obtain the target area from the grid map to obtain the grid map area; S04. Match the semantic map area with the grid map area, and when the match is successful, obtain the final posture output of the vehicle according to the current initial positioning information of the vehicle, the position of the positioning point and the positional relationship between the road area and the center of the vehicle body in the grid map. The present invention has the advantages of simple implementation method, low cost, high efficiency and precision, strong environmental adaptability and robustness.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle positioning, and in particular to a vehicle positioning method and device that integrates semantic maps and road information. Background Art

[0002] The vehicle needs to obtain the vehicle's positioning information in real time during driving, especially for autonomous driving vehicles, which must rely on accurate positioning information to locate the position in the semantic map for reliable autonomous driving control. For vehicle positioning, the existing technology mainly adopts the following methods:

[0003] 1. GPS+IMU positioning method. This is a method that integrates GPS (Global Positioning System) and IMU (Inertial Measurement Unit). GPS is used to obtain positioning information in real time, and IMU is used to compensate for GPS positioning information. This type of positioning method that integrates GPS and IMU is not only costly to implement, but also highly dependent on GPS signals. Positioning cannot be performed in situations where the GPS signal is weak or cannot be obtained, and the accuracy and reliability of positioning cannot be guaranteed in situations where the obtained GPS signal is unreliable.

[0004] 2. Laser point cloud data + high-precision map matching positioning method. That is, the self-driving car obtains point cloud data of the surrounding scene through laser radar, matches it with the high-precision map data collected and produced in advance, and obtains an estimate of the vehicle's current position. This type of method needs to rely on high-precision map data, but the production process of high-precision maps is complicated and the update cycle is long. The corresponding matching method (such as the ICP algorithm) is slow, resulting in complex and inefficient actual positioning, and it is difficult to ensure real-time positioning accuracy.

[0005] 3. Positioning method based on semantic map feature recognition. That is, by identifying specific semantic feature points in the semantic map, such as roads and other semantic and easily recognizable features, and then using the identified semantic feature points to correct the vehicle's positioning information in the indoor semantic map. This type of method relies on the information of feature points for feature matching, and the extraction of feature points is only applicable to structured scenes, that is, scenes with specific structures. For unstructured scenes without specific structures, it is actually difficult to effectively extract feature points. Therefore, the above-mentioned positioning method based on semantic map feature recognition is not applicable to unstructured scenes.

[0006] For example, Chinese patent application CN111521186A discloses a vehicle positioning method and device, a vehicle, and a storage medium. The solution is to identify the site entrance; obtain the indoor semantic map of the site according to the site entrance, and the indoor semantic map includes the semantic feature points of the site; and correct the positioning information of the vehicle in the indoor semantic map according to the semantic feature points obtained by the vehicle during movement. This solution can be applied to environments with rich information of highly structured feature points such as parking lots or other indoor sites. However, when there are certain changes in the environment, such as the entry and exit of vehicles in the parking lot, or the change of the parking position of vehicles in the parking lot, the above method cannot achieve positioning. For unstructured scenes, since it is difficult to obtain structured feature point information, the above method is also difficult to obtain accurate positioning effects.

[0007] In summary, the vehicle positioning methods in the existing technologies are either complex to implement, costly, and inefficient, or can only be applied to structured scenes and have poor environmental adaptability, and cannot take into account the efficiency, accuracy, and environmental adaptability of positioning. Summary of the invention

[0008] The technical problem to be solved by the present invention is: in response to the technical problems existing in the prior art, the present invention provides a vehicle positioning method and device that has a simple implementation method, low cost, high efficiency and accuracy, and strong environmental adaptability, and can be applied to unstructured scenarios and integrates semantic maps and road information.

[0009] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0010] A vehicle positioning method integrating semantic map and road information, comprising the following steps:

[0011] S01. Obtain the road environment image collected by the vehicle-mounted image acquisition device, segment the road area and project it into a grid map under the vehicle coordinate system to obtain a grid map under the road projection;

[0012] S02. According to the initial positioning information currently obtained by the vehicle, the corresponding positioning point is found on the semantic map, and the nearest neighbor road semantic area on the semantic map is searched according to the positioning point to obtain the semantic map area;

[0013] S03. Taking the road area as the center, obtaining the target area from the grid map under the road projection to obtain the grid map area;

[0014] S04. Match the semantic map area with the grid map area. When the match is successful, the final posture output of the vehicle in the semantic map is obtained based on the current initial positioning information of the vehicle, the position of the positioning point, and the positional relationship between the road area and the center of the vehicle body in the grid map under the road projection.

[0015] Furthermore, in step S01, a deep convolutional semantic segmentation model is used to segment the road area from the road environment image.

[0016] Furthermore, the steps of step S02 include:

[0017] S201. The new positioning point is obtained by searching for a point on the line of the road area closest to the center of the circle within an area with a specified radius and the center of the circle being the found positioning point;

[0018] S202. Acquire a designated map area on the semantic map according to the new positioning point to obtain the semantic map area.

[0019] Furthermore, after step S201 and before step S202, the method further includes calculating the angle between the centerline of the current road area and the X-axis obtained by using the new positioning point, and rotating the semantic map according to the angle so that the centerline of the road area is perpendicular to the X-axis; in step S202, taking the new positioning point as a reference point, obtaining the map area to the left, right and forward of the reference point on the rotated semantic map at a specified distance, to obtain the semantic map area.

[0020] Furthermore, the steps of step S03 include:

[0021] S301. According to the grid map under the road projection, calculate the distance from the center of the vehicle body to the left and right boundaries of the road area to obtain the lateral position of the vehicle on the road;

[0022] S302. According to the result calculated in step S301, the road area in the grid map under the road projection is translated to the middle of the grid map, and the direction of displacement is toward the side where the distance between the center of the vehicle body and the left and right boundaries of the road is smaller, to obtain a translated grid map;

[0023] S303. Orient an area at a specified distance to the left and right and a specified distance forward in the translated grid map to obtain the grid map area.

[0024] Furthermore, in the step S04, the iou value of the road area between the semantic map area and the grid map area is calculated, and whether the match is successful is determined based on the iou value. When the match is determined to be unsuccessful, it is determined that the vehicle has deviated from the road area.

[0025] Furthermore, in step S04, the longitudinal coordinate of the positioning point determined in step S02 in the map is used as the final longitudinal positioning coordinate, and the lateral position information of the center of the vehicle body in the grid map under the road projection is combined with the lateral coordinate of the positioning point in the semantic map area to obtain the final lateral positioning coordinate of the vehicle.

[0026] Furthermore, the lateral coordinate of the positioning point in the semantic map area is specifically moved in the direction where the distance between the center of the vehicle body and the left and right boundaries of the road is smaller, and the moving distance is half of the difference between the center of the vehicle body and the left and right boundaries of the road, so as to obtain the final lateral positioning coordinate of the vehicle.

[0027] A vehicle positioning device integrating semantic map and road information, comprising:

[0028] An image segmentation module is used to obtain a road environment image acquired by a vehicle-mounted image acquisition device, segment the road area and project it into a grid map in a vehicle body coordinate system to obtain a grid map under the road projection;

[0029] A semantic map acquisition module, used to find a corresponding positioning point on the semantic map according to the initial positioning information currently obtained by the vehicle, and search for the nearest neighbor road semantic area on the semantic map according to the positioning point to obtain a semantic map area;

[0030] A grid map acquisition module, used to acquire a target area from the grid map under the road projection with the road area as the center, to obtain a grid map area;

[0031] The posture output module is used to match the semantic map area with the grid map area. When the match is successful, the final posture output of the vehicle in the semantic map is obtained according to the current initial positioning information of the vehicle, the position of the positioning point and the positional relationship between the road area and the center of the vehicle body in the grid map under the road projection.

[0032] A computer device comprises a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program. The processor is characterized in that the processor is used to execute the computer program to perform the above method.

[0033] Compared with the prior art, the advantages of the present invention are:

[0034] 1. The present invention integrates semantic maps and road information, makes full use of high-precision semantic segmentation information, and performs simple image information matching on the semantic map in combination with the road information of the semantic map to achieve vehicle positioning. There is no need to use complex positioning equipment, and the implementation cost is low. The entire positioning process is simple to operate, and the amount of information required is also small. There is no need to rely on the information of feature points. Not only is the positioning efficiency and accuracy high, but the environmental adaptability and robustness are strong. Since positioning is based on the overall road area segmentation information, it can be applied not only to structured scenes, but also to various unstructured scenes.

[0035] 2. The present invention realizes positioning judgment by calculating the IOU value between the grid map and the semantic map, without performing a complex iterative matching process. It can quickly and efficiently determine whether the vehicle is in a normal road area, avoiding the problems of being too time-consuming and prone to falling into local optimality when traditional ICP and other iterative methods are required for matching, thereby realizing fast and accurate vehicle positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the implementation process of the vehicle positioning method that integrates semantic maps and road information in this embodiment.

[0037] Figure 2 It is a schematic diagram of the implementation process of the present invention in a specific application embodiment of integrating semantic maps and road information to realize vehicle positioning.

[0038] Figure 3 It is a schematic diagram of a grid map obtained in a specific application embodiment of the present invention. DETAILED DESCRIPTION

[0039] The present invention is further described below in conjunction with the accompanying drawings and specific preferred embodiments, but the protection scope of the present invention is not limited thereby.

[0040] like Figure 1 As shown, the vehicle positioning method of this embodiment that integrates semantic map and road information includes the following steps:

[0041] S01. Image segmentation: Obtain a road environment image acquired by a vehicle-mounted image acquisition device, segment the road area, and project it into a grid map under the vehicle body coordinate system to obtain a grid map under the road projection.

[0042] The vehicle-mounted image acquisition device may specifically be a vehicle-mounted camera or other device. The vehicle-mounted camera collects road environment information around the vehicle in real time during the driving process. After obtaining the road environment image collected by the vehicle-mounted camera, the deep convolution semantic segmentation model is used to segment the road area from the road environment image. The deep convolution semantic segmentation model can be trained in advance using images containing various types of roads as training data. The road environment image collected in real time after training is input into the trained deep convolution semantic segmentation model, and the road area in the image can be segmented to obtain the segmented road area.

[0043] After segmenting the road area, the road area is projected into a grid map under the vehicle body coordinate system, thus obtaining a grid map under road projection.

[0044] S02. Semantic map acquisition: According to the current positioning information of the vehicle, the corresponding positioning point is found on the semantic map, and the nearest neighbor road semantic area on the semantic map is searched according to the positioning point to obtain the semantic map area.

[0045] The vehicle's positioning information can be obtained using GPS, IMU, etc. as initial positioning information. Based on the initial positioning information, the corresponding positioning point is found on the semantic map, and then the nearest road semantic area on the semantic map is searched based on the positioning point. The GPS positioning information can be used to integrate the road area and the semantic map to jointly realize vehicle positioning. The above initial positioning information can also be obtained using other positioning systems other than GPS, such as the Beidou system or other types of combined navigation systems.

[0046] The nearest road semantic area is the road area closest to the positioning point. In this embodiment, the specific steps of the above step S02 are:

[0047] S201. A new positioning point is obtained by searching for a point on the midline of the road area closest to the center of the circle within a region with a specified radius and the center of the circle being the positioning point found, wherein the midline of the road area is a line with an equal distance from the road boundary;

[0048] S202. Acquire a designated map area on the semantic map according to the new positioning point to obtain a semantic map area.

[0049] In a specific application embodiment, when searching for the nearest road semantic area on the semantic map according to the positioning point, firstly, in an area with a radius of a specified size (such as a radius of 4 to 6 meters) with the found positioning point as the center, search for the point on the line of the road area closest to the center of the circle to obtain a new positioning point; then, the new positioning point is used as a reference point to obtain the specified area in the semantic map, such as obtaining a semantic map area to the left and right and forward a certain distance on the map according to the new positioning point, and the required semantic map area can be obtained. The above distance size and the area size of the semantic map area can be determined according to actual needs.

[0050] This embodiment uses the initial positioning point determined by GPS on the semantic map to search for the nearest road semantic area, so as to quickly locate and obtain the road area information in the semantic map.

[0051] After step S201 and before step S202, the above-mentioned step also includes obtaining the angle between the center line of the current road area and the X-axis according to the new positioning point, and rotating the semantic map according to the angle so that the center line of the road area is perpendicular to the X-axis; in step S202, specifically taking the new positioning point as the reference point, obtaining the map area to the left, right and forward of the reference point on the rotated semantic map at a specified distance, and obtaining the semantic map area.

[0052] S03. Grid map acquisition: Taking the road area as the center, the target area is acquired from the grid map under the road projection to obtain the grid map area.

[0053] According to step S01, the road area is projected to obtain a grid map, and a region centered on the road area is further acquired from the grid map to obtain a grid map region.

[0054] The above step S03 includes:

[0055] S301. According to the grid map under the road projection, calculate the distance from the center of the vehicle to the left and right boundaries of the road area to obtain the lateral position of the vehicle on the road;

[0056] S302. According to the result calculated in step S301, the road area in the grid map under the road projection is translated to the middle of the grid map, and the direction of displacement is toward the side where the distance between the center of the vehicle body and the left and right boundaries of the road is smaller, to obtain a translated grid map;

[0057] S303. Orient an area at a specified distance to the left and right and a specified distance forward from the translated grid map to obtain a grid map area.

[0058] In this embodiment, it is specifically assumed that the center of the vehicle body is at the center of the grid map. Then, the distance from the center of the vehicle body to the left and right boundaries of the road area is calculated to obtain the lateral position of the vehicle in the road. Then, according to the calculated distance from the center of the vehicle body to the left and right boundaries of the road and the special pixel value of the road area in the grid map (for example, if the pixel value is 0, it corresponds to the road area, and if the pixel value is other values, it corresponds to the non-road area), the road area in the grid map is horizontally translated to the middle of the grid map. The displacement distance is specifically half of the difference between the distance from the center of the vehicle body to the left and right boundaries of the road. The direction of displacement is to move half of the difference between the distance from the center of the vehicle body to the left and right boundaries of the road to the side where the distance from the center of the vehicle body to the left and right boundaries of the road is smaller, that is, to the side where the distance from the center of the vehicle body to the left and right boundaries of the road is smaller. Then, from the grid map area at the specified distance of the translated grid map and the specified distance forward, a grid map with the road area in the middle position can be obtained, so that the grid map contains as much road area information as possible.

[0059] The special pixel values ​​of the above-mentioned road area may be specific pixel values ​​that can represent the characteristics of the road area. Of course, other auxiliary determination methods may also be introduced according to actual needs, such as street lights and trees regularly arranged on both sides of the road.

[0060] S04. Match the semantic map area with the grid map area. When the match is successful, the final position output of the vehicle in the semantic map is obtained based on the current positioning information of the vehicle, the position of the positioning point, and the positional relationship between the road area and the center of the vehicle body in the grid map under road projection.

[0061] After obtaining the semantic map area in step S02 and the grid map area in step S03, the iou value (intersection-over-union ratio) of the road area in the two images is specifically calculated, that is, the ratio of the intersection and union of the semantic map area and the grid map area, and whether the positioning is successful is determined according to the size of the iou value. If the iou value ratio of the road area in the two images of the semantic map area and the grid map area is large, it indicates that the overlapping area of ​​the two images is large, and the vehicle is driving in a normal road area. On the contrary, if the iou value ratio of the road area in the two images of the semantic map area and the grid map area is small, it indicates that the overlapping area of ​​the two images is small, and the vehicle may have deviated from the road area, and further measures such as emergency braking need to be taken.

[0062] This embodiment realizes positioning judgment by calculating the IOU value between the raster map and the semantic map, without performing a complex iterative matching process. It can quickly and efficiently determine whether the vehicle is in a normal road area, avoiding the problems of taking a long time and easily falling into local optimality when using iterative methods such as traditional ICP for matching, thereby achieving fast and accurate vehicle positioning.

[0063] In a specific application embodiment, the positioning is determined to be successful if and only if step S02 can effectively search for a new positioning point and the calculated iou value is greater than a preset threshold (such as 0.7). If other situations occur, it is determined that the vehicle has deviated from the road area, and the vehicle needs to take emergency braking measures to ensure the vehicle's driving safety.

[0064] When the match is determined to be successful, the longitudinal coordinate of the positioning point determined in step S02 in the map is used as the final longitudinal positioning coordinate, and the lateral position information of the center of the vehicle in the grid map under the road projection is combined with the lateral coordinate of the positioning point in the semantic map area to obtain the final lateral positioning coordinate of the vehicle.

[0065] In a specific application embodiment, the lateral coordinate of the positioning point in the semantic map area is moved in the direction where the distance between the center of the vehicle body and the left and right boundaries of the road is smaller. The moving distance is half of the difference between the center of the vehicle body and the left and right boundaries of the road, thus obtaining the final lateral positioning coordinate of the vehicle.

[0066] The present invention realizes vehicle positioning by integrating semantic maps with road information, making full use of high-precision semantic segmentation information, and performing simple image information matching on the semantic map in combination with the road information of the semantic map. There is no need to use complex positioning equipment, and the implementation cost is low. The entire positioning process is simple to operate, and the amount of information required is also small. There is no need to rely on the information of feature points. Not only is the positioning efficiency and accuracy high, but the environmental adaptability and robustness are strong. Since positioning is based on the overall road area segmentation information, it can be applied not only to structured scenes, but also to various unstructured scenes.

[0067] The present invention is further described below by taking the above method of the present invention as an example in a specific application embodiment to realize vehicle positioning.

[0068] like Figure 2 As shown, the detailed steps of integrating semantic map and road information to realize vehicle positioning in this embodiment are as follows:

[0069] Step 1: Use the deep convolutional semantic segmentation model to segment the road area and project the road area into a grid map in the vehicle coordinate system. The grid map obtained in the specific application embodiment is as follows: Figure 3 As shown, the white grid area represents the road area projected into the vehicle body coordinate system.

[0070] Step 2: Get the vehicle's current GPS and initial positioning information given by imu.

[0071] Step 3: According to the positioning information provided by GPS and imu, locate the corresponding positioning point on the semantic map, search for the point on the midline of the road area closest to the center of the circle with a radius of 6 meters, and use it as the new positioning point, where the midline of the road area is the line with the same distance from the road boundary.

[0072] Step 4: Get the point and the center line according to step 3, get the angle between the center line of the current position and the X-axis, and rotate the map according to the angle so that the center line is perpendicular to the X-axis.

[0073] Step 5: Obtain the semantic map area 25 meters to the left and right and 45 meters forward on the map based on the new positioning point.

[0074] Step 6: Based on the grid map obtained by projecting the road area and the assumption that the center of the vehicle is at the center of the grid map, calculate the distance from the center of the vehicle to the left and right boundaries of the road area, which is the lateral position of the vehicle on the road.

[0075] Step 7: Based on the calculated distance from the center of the vehicle to the left and right boundaries of the road and the special pixel value of the road area in the grid map, the road area in the grid map is horizontally translated to the middle of the grid map. The displacement distance is half of the difference between the center of the vehicle and the left and right boundaries of the road. The direction of displacement is to move to the side where the distance between the center of the vehicle and the left and right boundaries of the road is smaller. Then, take the grid map area 25 meters to the left and right of the grid and 45 meters forward.

[0076] Step 8: Using the semantic map area obtained in step 5 and the grid map area obtained in step 7, calculate the iou (intersection over union) of the road area in the two images, that is, the ratio of the intersection and union of the semantic map area and the grid map area.

[0077] Step 9: If and only if step 3 can effectively search for a new positioning point, and the value of iou obtained in step 8 is greater than 0.7, the positioning is determined to be successful. If other situations occur, it means that the vehicle has deviated from the road area and the vehicle needs to be controlled to take emergency braking measures.

[0078] Step 10: Take the longitudinal coordinates of the positioning point on the map road obtained in step 3 as the final longitudinal positioning coordinates, take the lateral position information of the vehicle center in the grid map road area obtained in step 6, and combine it with the lateral coordinates of the positioning point on the semantic map road obtained in step 3 to obtain the final lateral positioning coordinates, that is, move the coordinates of step 3 to the direction where the distance between the vehicle center and the left and right boundaries of the road obtained in step 6 is smaller, and the moving distance is half of the difference between the distances from the vehicle center to the left and right boundaries of the road.

[0079] Step 11: Output the final result, including the final position of the current vehicle in the semantic map and the value of iou.

[0080] Through the above steps, this embodiment can make full use of high-precision semantic segmentation information and combine it with the semantic map to perform IOU algorithm matching. The required information is small but the robustness is strong. The calculation method is simple and the execution efficiency is very high, and the corresponding matching positioning can be achieved quickly and efficiently.

[0081] The vehicle positioning device of this embodiment that integrates semantic map and road information includes:

[0082] An image segmentation module is used to obtain a road environment image acquired by a vehicle-mounted image acquisition device, segment the road area and project it into a grid map in a vehicle body coordinate system to obtain a grid map under the road projection;

[0083] The semantic map acquisition module is used to find the corresponding positioning point on the semantic map according to the initial positioning information currently obtained by the vehicle, and search the nearest neighbor road semantic area on the semantic map according to the positioning point to obtain the semantic map area;

[0084] A grid map acquisition module is used to acquire a target area from a grid map under road projection with the road area as the center to obtain a grid map area;

[0085] The posture output module is used to match the semantic map area with the grid map area. When the match is successful, the final posture output of the vehicle in the semantic map is obtained based on the vehicle's current initial positioning information, the position of the positioning point, and the positional relationship between the road area and the center of the vehicle body in the grid map under road projection.

[0086] In this embodiment, the semantic map acquisition module includes:

[0087] A search unit is used to obtain a new positioning point by searching for a point on the line of the road area closest to the center of the circle within an area with a specified radius and taking the found positioning point as the center of the circle, and obtaining a new positioning point;

[0088] The first map acquisition unit is used to acquire a designated map area on the semantic map according to the new positioning point to obtain a semantic map area.

[0089] In this embodiment, the grid map acquisition module includes:

[0090] A calculation unit, used to calculate the distance from the center of the vehicle body to the left and right boundaries of the road area according to the grid map under the road projection, and obtain the lateral position of the vehicle on the road;

[0091] A translation unit is used to translate the road area in the grid map under the road projection to the middle of the grid map according to the result calculated by the calculation unit, and the direction of displacement is toward the side where the distance between the center of the vehicle body and the left and right boundaries of the road is smaller, so as to obtain a translated grid map;

[0092] The second map acquisition unit is used to obtain a grid map area from an area with a specified distance to the left and right and a specified distance forward in the translated grid map.

[0093] In this embodiment, the posture output module calculates the iou value of the road area between the semantic map area and the grid map area, and determines whether the match is successful according to the iou value. When the match is determined to be unsuccessful, it is determined that the vehicle deviates from the road area. When it is determined that the match is successful, the longitudinal coordinate of the positioning point in the map determined in the semantic map acquisition module is used as the final longitudinal positioning coordinate, and the lateral position information of the center of the vehicle in the grid map under the road projection is combined with the lateral coordinate of the positioning point in the semantic map area to obtain the final lateral positioning coordinate of the vehicle.

[0094] The vehicle positioning device that integrates the semantic map and road information in this embodiment corresponds one to one with the vehicle positioning method that integrates the semantic map and road information described above, and will not be described in detail here.

[0095] This embodiment also provides a computer device, including a processor and a memory, wherein the memory is used to store a computer program and the processor is used to execute the computer program. The feature of the device is that the processor is used to execute the computer program to perform the above method.

[0096] Those skilled in the art will appreciate that the above description of the computer device is merely an example and does not constitute a limitation on the computer device. The computer device may include more or fewer components than described above, or may combine certain components, or may include different components, such as input and output devices, network access devices, buses, etc. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer device, and uses various interfaces and lines to connect the various parts of the entire computer device.

[0097] The memory can be used to store computer programs and / or modules. The processor realizes various functions of the computer device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0098] If the module / unit integrated in the computer device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can be executed by the processor. The steps of the above-mentioned distributed crawler method embodiment based on template annotation can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, electrical signal and software distribution medium, etc.

[0099] The above is only a preferred embodiment of the present invention, and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment, it is not intended to limit the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection of the technical solution of the present invention.

Claims

1. A vehicle positioning method integrating semantic map and road information, characterized in that the steps include: S01. Obtain the road environment image collected by the vehicle-mounted image acquisition device, segment the road area and project it into a grid map under the vehicle coordinate system to obtain a grid map under the road projection; S02. According to the initial positioning information currently obtained by the vehicle, the corresponding positioning point is found on the semantic map, and the nearest neighbor road semantic area on the semantic map is searched according to the positioning point to obtain the semantic map area; S03. Taking the road area as the center, obtaining the target area from the grid map under the road projection to obtain the grid map area; S04. Matching the semantic map area with the grid map area, when the match is successful, obtaining the final position output of the vehicle in the semantic map according to the current initial positioning information of the vehicle, the position of the positioning point, and the position relationship between the road area and the center of the vehicle body in the grid map under the road projection; In the step S04, the iou value of the road area between the semantic map area and the grid map area is calculated, and whether the match is successful is determined according to the iou value. When the match is determined to be unsuccessful, it is determined that the vehicle has deviated from the road area. When the match is determined to be successful, the longitudinal coordinate of the positioning point determined in step S02 in the map is used as the final longitudinal positioning coordinate, and the lateral position information of the center of the vehicle in the grid map under the road projection is combined with the lateral coordinate of the positioning point in the semantic map area to obtain the final lateral positioning coordinate of the vehicle.

2. The vehicle positioning method integrating semantic map and road information according to claim 1, characterized in that: In the step S01, a deep convolutional semantic segmentation model is used to segment the road area from the road environment image.

3. The vehicle positioning method integrating semantic map and road information according to claim 1, characterized in that: The step S02 comprises: S201. A new positioning point is obtained by searching for a point on the line of the road area closest to the center of the circle within an area with a specified radius and the center of the circle being the positioning point found; S202. Acquire a designated map area on the semantic map according to the new positioning point to obtain the semantic map area.

4. The vehicle positioning method integrating semantic map and road information according to claim 3 is characterized in that: After step S201 and before step S202, the method further includes obtaining the angle between the center line of the current road area and the X-axis according to the new positioning point, and rotating the semantic map according to the angle so that the center line of the road area is perpendicular to the X-axis; in step S202, taking the new positioning point as a reference point, obtaining the map area to the left, right and forward of the reference point on the rotated semantic map at a specified distance, and obtaining the semantic map area.

5. The vehicle positioning method integrating semantic map and road information according to claim 1, characterized in that: The step S03 comprises: S301. According to the grid map under the road projection, calculate the distance from the center of the vehicle body to the left and right boundaries of the road area to obtain the lateral position of the vehicle on the road; S302. According to the result calculated in step S301, the road area in the grid map under the road projection is translated to the middle of the grid map, and the direction of displacement is toward the side where the distance between the center of the vehicle body and the left and right boundaries of the road is smaller, to obtain a translated grid map; S303. Locate an area with a specified distance to the left, a specified distance to the right, and a specified distance forward from the translated grid map to obtain the grid map area.

6. The vehicle positioning method integrating semantic map and road information according to claim 1, characterized in that: Specifically, the lateral coordinate of the positioning point in the semantic map area is moved in the direction where the distance between the center of the vehicle body and the left and right boundaries of the road is smaller. The moving distance is half of the difference between the center of the vehicle body and the left and right boundaries of the road, so as to obtain the final lateral positioning coordinate of the vehicle.

7. A vehicle positioning device integrating semantic map and road information, characterized in that: include: An image segmentation module is used to obtain a road environment image acquired by a vehicle-mounted image acquisition device, segment the road area and project it into a grid map in a vehicle body coordinate system to obtain a grid map under the road projection; A semantic map acquisition module, used to find a corresponding positioning point on the semantic map according to the initial positioning information currently obtained by the vehicle, and search for the nearest neighbor road semantic area on the semantic map according to the positioning point to obtain a semantic map area; A grid map acquisition module, used to acquire a target area from the grid map under the road projection with the road area as the center, to obtain a grid map area; A posture output module, used for matching the semantic map area with the grid map area, and when the matching is successful, obtaining the final posture output of the vehicle in the semantic map according to the current initial positioning information of the vehicle, the position of the positioning point, and the positional relationship between the road area and the center of the vehicle body in the grid map under the road projection; In the posture output module, the iou value of the road area between the semantic map area and the grid map area is calculated, and whether the match is successful is determined according to the iou value. When the match is determined to be unsuccessful, it is determined that the vehicle has deviated from the road area. When it is determined that the match is successful, the longitudinal coordinate of the positioning point determined in the semantic map acquisition module in the map is used as the final longitudinal positioning coordinate, and the lateral position information of the center of the vehicle in the grid map under the road projection is combined with the lateral coordinate of the positioning point in the semantic map area to obtain the final lateral positioning coordinate of the vehicle.

8. A computer device comprising a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program, wherein: The processor is configured to execute the computer program to perform the method according to any one of claims 1 to 6.

Images