Map construction method and device, vehicle and storage medium

Abstract

The invention discloses a map construction method and device, a vehicle and a readable storage medium, and the method comprises the steps: determining a first region in an initial grid map based on a planned path of a vehicle driving in the initial grid map of a target region; the first area at least comprises a planned path; constructing a grid map for the first area according to the first resolution to obtain a first grid map; constructing a grid map for the second area according to the second resolution to obtain a second grid map; the second area is an area except the first area in the initial grid map; the first resolution is higher than the second resolution; and obtaining a target grid map of the target area based on the first grid map and the second grid map. According to the method provided by the invention, the accuracy of the determined second grid map is relatively high.

Description

Technical Field

[0001] The present application relates to the technical field of vehicles, and more specifically, to a map construction method, apparatus, vehicle, and computer-readable storage medium. Background Art

[0002] A grid map is a map form widely used in the field of autonomous driving to indicate the environmental information around the vehicle during the autonomous driving process of the vehicle; each grid in grid mapping represents a discrete area in the environment, and each grid is marked as passable or not, so as to determine the passable area of the vehicle or the area where obstacles around the vehicle are located through the grid map.

[0003] In the related art, there is a situation where the accuracy of the constructed grid map is relatively low. Summary of the Invention

[0004] The present application provides a map construction method, apparatus, vehicle, and computer-readable storage medium to improve the accuracy of the grid map.

[0005] In a first aspect, an embodiment of the present application provides a map construction method, and the method includes:

[0006] Determine a first area in the initial grid map based on the planned path of the vehicle itself driving in the initial grid map of the target area; the first area includes at least the planned path;

[0007] Construct a grid map for the first area according to a first resolution to obtain a first grid map;

[0008] Construct a grid map for a second area according to a second resolution to obtain a second grid map; the second area is the area in the initial grid map except the first area; the first resolution is higher than the second resolution;

[0009] Obtain a target grid map of the target area based on the first grid map and the second grid map.

[0010] In a second aspect, an embodiment of the present application further provides a map construction apparatus, and the apparatus includes:

[0011] A first determination module, configured to determine a first area in the initial grid map based on the planned path of the vehicle itself driving in the initial grid map of the target area; the first area includes at least the planned path;

[0012] A first construction module, configured to construct a grid map for the first area according to a first resolution to obtain a first grid map;

[0013] A second construction module, configured to construct a grid map for a second area at a second resolution to obtain a second grid map; the second area is the area in the initial grid map except for the first area; the first resolution is higher than the second resolution;

[0014] A second determination module, configured to obtain a target grid map of a target area based on the first grid map and the second grid map.

[0015] In a third aspect, an embodiment of the present application further provides a vehicle, which includes: one or more processors; a memory; one or more applications, where one or more applications are stored in the memory and configured to be executed by one or more processors, and one or more programs are configured to execute the above method.

[0016] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores program code executable by a processor. When the program code is executed by the processor, the processor executes the above method.

[0017] A map construction method, device, vehicle and computer-readable storage medium provided by the present application. In the present application, first, based on a planned path of the vehicle itself traveling in an initial grid map of a target area, a first area is determined in the initial grid map. Then, a grid map is constructed for the first area at a higher first resolution to obtain a first grid map, and a grid map is constructed for the second area at a lower second resolution to obtain a second grid map. Finally, based on the first grid map and the second grid map, a target grid map of the target area is obtained. Thus, a map is constructed for the key area (i.e., the first area) of the vehicle itself at a higher resolution, so that the accuracy of the grid map in the first area is relatively high, thereby improving the accuracy of the constructed target grid map. At the same time, only the first area is constructed with a grid map at a larger resolution, and the second area is not constructed with a grid map at a larger resolution, thereby reducing the data processing volume of the second area, improving the construction efficiency of the second grid map corresponding to the second area, and further improving the construction efficiency of the entire target grid map.

[0018] Other features and advantages of the embodiments of the present application will be described in the subsequent description, and part of them will become obvious from the description, or will be understood by implementing the embodiments of the present application. The objectives and other advantages of the embodiments of the present application can be achieved and obtained through the structures specifically pointed out in the written description, claims, and drawings. Description of the Drawings

[0019] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those skilled in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0020] Figure 1 Fig. shows a schematic diagram of a vehicle hardware environment applicable to the embodiments of the present application.

[0021] Figure 2 Fig. shows a flowchart of a map construction method proposed according to an embodiment of the present application.

[0022] Figure 3 Fig. shows a schematic diagram of a first area in the embodiments of the present application.

[0023] Figure 4 Fig. shows a schematic diagram of the effective area of the sensor of a camera in the embodiments of the present application.

[0024] Figure 5 Fig. shows a schematic diagram of a target observation area in the embodiments of the present application.

[0025] Figure 6 Fig. shows a schematic diagram of the quadtree data format in a first grid map in the embodiments of the present application.

[0026] Figure 7 shows Figure 2 The flowchart of step S120 in the corresponding embodiment in one embodiment.

[0027] Figure 8 Fig. shows a schematic diagram of compensating obstacle points in the embodiments of the present application.

[0028] Figure 9 Fig. shows a schematic diagram of a map construction process in the embodiments of the present application.

[0029] Figure 10 Fig. shows a structural block diagram of a map construction device proposed according to an embodiment of the present application. Detailed implementation manners

[0030] To enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Generally, the components of the embodiments of this application described and illustrated herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the drawings is not intended to limit the scope of the claimed application, but merely represents the selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative efforts belong to the scope of protection of this application.

[0031] It should be noted that similar reference numerals and letters indicate similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0032] Referring to Figure 1 , Figure 1 , a schematic diagram of a vehicle hardware environment applicable to the embodiments of this application is shown. The vehicle 100 includes a driving system 110. The driving system 110 can be built-in with a variety of autonomous driving functions. The driving system 110 can store an electronic map. The driving system 110 can plan a driving route according to the electronic map stored by itself, and can also control the vehicle to drive autonomously according to the planned driving route.

[0033] The driving system 110 may include a data acquisition device 111, one or more (only one is shown in the figure) processors 112, and a memory 113.

[0034] The data acquisition device 111 is used to detect the pose information of the vehicle and the environmental information around the vehicle. The data acquisition device 111 may include a camera and a radar, etc. Among them, both the camera and the radar can be used to collect the environmental information around the vehicle.

[0035] The processor 112 may be a microcontroller unit (MCU). The microcontroller unit has a built-in memory 113. A program that can execute the content in the following embodiments is stored in the memory 113, and the processor 112 can execute the program stored in the memory 113.

[0036] Among them, the processor 112 may include one or more processors. The processor 112 utilizes various interfaces and lines to splice various parts within the entire vehicle 100, and by running or executing instructions, programs, code sets, or instruction sets stored in the memory 113, as well as calling data stored in the memory 113, it executes various functions of the vehicle 10 and processes data.

[0037] The memory 113 may include a Random Access Memory (RAM), and may also include a Read-Only Memory. The memory 15 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 15 may include a program storage area and a data storage area. Among them, the program storage area can store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the following various method embodiments, etc.

[0038] Please refer to Figure 2 , Figure 2 , which shows a flowchart of a map construction method proposed according to an embodiment of the present application for a vehicle. The method includes:

[0039] S110. Based on the planned path of the vehicle itself driving in the initial grid map of the target area, determine a first area in the initial grid map.

[0040] Among them, the first area includes at least the planned path.

[0041] In the present application, the vehicle can be a bus, a car, or a truck, and the vehicle can also be an electric vehicle, a fuel vehicle, a hybrid vehicle, etc. The vehicle itself refers to the vehicle.

[0042] The initial grid map may include multiple grids, and the resolution of the initial grid map may be the initial resolution. Each grid in the initial grid map may also correspond to an obstacle search result, which is used to indicate whether there is an obstacle in the grid (or whether the vehicle itself can pass. The presence of an obstacle means that the vehicle itself cannot pass, and the absence of an obstacle means that the vehicle itself can pass). The obstacle search result can be in the form of probability. That is, in the initial grid map, the probability of a grid refers to the possibility of an obstacle existing in the real environment area indicated by the grid. Usually, when storing the initial grid map, the obstacle search result corresponding to the grid in the initial grid map is stored as the attribute information of the grid.

[0043] In some embodiments, the vehicle may be equipped with multiple sensors (such as multiple cameras or multiple radars). The vehicle constructs an initial grid map based on the sensor information collected by the multiple sensors according to the initial resolution.

[0044] The target area can be the area where the host vehicle is traveling. For example, when the host vehicle is about to park near a parking lot, the target area is the parking lot where the host vehicle is located. Another example is that when the host vehicle is traveling at an intersection, the target area is the intersection where the host vehicle is traveling.

[0045] Sensor data of the target area can be collected by the sensors of the host vehicle to obtain obstacle points indicating different obstacles. An obstacle point is a point used to indicate the area where an obstacle is located. Generally, an obstacle can be indicated by multiple obstacle points. For example, a water horse is indicated by 100 obstacle points indicating the location of the water horse.

[0046] In some embodiments, the sensors of the host vehicle include a camera and a radar. The camera and the radar are used to collect obstacle points for an obstacle, and then an initial grid map is determined according to the obstacle points collected by the camera and the radar at the initial resolution. At this time, the obstacle points collected by the camera of the host vehicle are visual obstacle points, and the obstacle points collected by the radar are used as radar obstacle points.

[0047] In some embodiments, the target area can be divided into grids to obtain a grid area to be processed. The grid area to be processed includes multiple grids. Then, through the Inverse Sensor Model (ISM), based on the obstacle points observed by the camera and the radar, the obstacle search result of each grid in the grid area to be processed is determined. Then, based on the obstacle search results of each grid, an initial grid map is obtained. This inverse observation model mainly updates the state of each grid (that is, updates the obstacle search result of each grid) by reversely simulating the observation process of the sensor (such as the aforementioned camera or radar).

[0048] First, a grid area (grid area with the initial resolution) is initialized for the target area to obtain a grid area to be processed including multiple grids. One grid in the grid area to be processed corresponds to a part of the target area around the host vehicle. The size of the area indicated by each grid in the target area can be set based on requirements. For example, each grid indicates an area of 0.5m × 0.5m in the target area.

[0049] Then, the inverse observation model calculates the likelihood function based on the observed obstacle points, and then updates the posterior probability of each grid in the grid area to be processed through the Bayesian formula. Finally, each grid is marked according to the posterior probability of each grid to obtain the obstacle search result of each grid. Finally, based on the obstacle search results of each grid, an initial grid map is obtained.

[0050] In some other embodiments, the sensors of the host vehicle include a camera and a radar. The camera can be a fish-eye camera, which is used to detect visual obstacle points in the target area. At the same time, the relative distance between each observed visual obstacle point and the host vehicle is determined by the radar of the host vehicle as the obstacle point distance. Based on the visual obstacle points and the corresponding obstacle point distances, an initial grid map is constructed.

[0051] First, a grid area (grid area with initial resolution) is initialized for the target area, and a to-be-processed grid area including multiple grids is obtained. One grid in the to-be-processed grid area corresponds to a part of the target area around the host vehicle. Then, based on the aforementioned observed visual obstacle points and obstacle point distances, the obstacle search result of each grid can be obtained through an inverse observation model. Based on the obstacle search results of each grid in the to-be-processed grid area, an initial grid map is obtained.

[0052] The inverse observation model calculates a likelihood function based on the initial visual obstacle points and obstacle point distances, then updates the posterior probability of each grid in the to-be-processed grid area through Bayes' formula. Finally, each grid is labeled according to the posterior probability of each grid to obtain the obstacle search result of each grid. Finally, based on the obstacle search results of each grid, an initial grid map is obtained.

[0053] It is worth mentioning that at any time point during the driving of the host vehicle, a to-be-processed grid area can be created, and then based on the obstacle points observed at this time point, the initial grid map for this time point is determined.

[0054] After obtaining the initial grid map, a path for the host vehicle to drive while avoiding obstacles can be searched in the initial grid map through path planning algorithms (including A* search algorithm, Dijkstra algorithm, RRT (Randomized Rapidly-Exploring Random Trees) algorithm, etc.), which serves as the planned path for the host vehicle to drive in the target area.

[0055] That is to say, after obtaining the initial grid map, the area including the planned path is divided from the initial grid map as the first area. At the same time, the area outside the first area in the initial grid map can also be used as the second area.

[0056] In some embodiments, the area formed by the grids covered by the planned path in the initial grid map can be directly obtained as the first area. In some other embodiments, in the initial grid map, inflation can be performed based on a specified distance threshold with the planned path as the center to obtain the first area; and the area in the initial grid map except the first area is obtained as the second area. The distance threshold can be set according to requirements, for example, 1m.

[0057] In this application, the expansion can include expanding along the width direction of the vehicle itself with the planned path as the center. Of course, it can also include expanding along the driving direction of the vehicle itself with the planned path as the center. At this time, it means expanding backward in the driving direction from the starting point of the planned path and expanding forward in the driving direction from the end point of the planned path.

[0058] For example, when the aforementioned expansion is expanding along the width direction of the vehicle itself with the planned path as the center, the obtained first area. As Figure 3 shown, for the area included in the dotted line 31 which is the planned path of the vehicle itself, after expanding the planned path 31, the area 32 covered by the dark grid lines is obtained as the first area.

[0059] It is worth mentioning that as the vehicle itself drives, the planned path of the vehicle itself also changes. For each time point, the determined planned path also changes. Therefore, for each time point, a planned path can be determined, and a first area and a second area can be determined.

[0060] S120. Construct a grid map for the first area according to the first resolution to obtain the first grid map.

[0061] In other words, the first area where the planned path of the vehicle itself is located is used as the key area, and a grid map is constructed with the first resolution to obtain the first grid map.

[0062] In some embodiments, S120 may include: dividing the first area into grids according to the first resolution to obtain a high-resolution grid area; the high-resolution grid area includes a plurality of high-resolution grids; based on the obstacle points observed by the vehicle itself at multiple time points, determine the obstacle search result of each high-resolution grid; based on the obstacle search result of each high-resolution grid, determine the first grid map.

[0063] In this application, the first area can be directly divided into a plurality of high-resolution grids to obtain a high-resolution grid area; for example, the area composed of all 100 grids located in the first area in the initial grid map is re-divided into 150 high-resolution grids to obtain a high-resolution grid area. It is also possible to divide each grid located in the first area in the initial grid map into a plurality of high-resolution grids to obtain a high-resolution grid area. For example, for each grid located in the first area in the initial grid map, it is divided into 3×3 high-resolution grids to obtain a high-resolution grid area.

[0064] As mentioned above, the obstacle points observed by the vehicle itself can include visual obstacle points observed by the vehicle itself through a camera, and the obstacle points observed by the vehicle itself can also include radar obstacle points observed by the vehicle itself through a radar.

[0065] At each time point after determining the initial grid map, the ego vehicle can observe obstacle points, and determine the obstacle search results of each high-resolution grid at each time point based on the obstacle points observed at each time point. The determination process of the obstacle search results of each high-resolution grid refers to the determination process of the obstacle search results of each grid in the aforementioned initial grid map, which will not be elaborated here.

[0066] In some embodiments, S120 may include: dividing the first area into grids according to the first resolution to obtain a high-resolution grid area; the high-resolution grid area includes a plurality of high-resolution grids; determining the obstacle search results of each high-resolution grid based on the obstacle points observed by the ego vehicle at multiple time points and the position where the first sensor for observing the obstacle points is located; and determining the first grid map based on the obstacle search results of each high-resolution grid.

[0067] For example, if the obstacle points are visual obstacle points collected by a camera, then the camera is the first sensor, and the position where the first sensor is located is the position where the camera is located. Also, if the obstacle points are radar obstacle points collected by a radar, then the radar is the first sensor, and the position where the first sensor is located is the position where the camera is located.

[0068] In this application, the obstacle search results of each high-resolution grid are determined by combining the obstacle points observed by the ego vehicle at multiple time points and the position where the first sensor for observing the obstacle points is located.

[0069] Specifically, the target observation area of each time point can be determined based on the position where the first sensor is located; the first candidate obstacle points located in the corresponding target observation area are obtained from the obstacle points of each time point; and the obstacle search results of each high-resolution grid are determined based on the first candidate obstacle points of multiple time points.

[0070] Generally, there is at least one first sensor, and each first sensor can correspond to a sensor effective area. For each time point, the sensor effective areas of each first sensor on the ego vehicle are obtained as the target observation area of each time point.

[0071] The sensor effective area of the first sensor refers to the area where the observation value of the first sensor is relatively accurate. The sensor effective area of the first sensor can be determined based on requirements and the characteristics of the sensor itself. This application does not make a limit. For example, when the first sensor is a camera, the sensor effective area of the first sensor is as Figure 4 shown, Figure 4 In, after the first sensor C1 is installed on the ego vehicle, the observation range is outside the edge C3 of the ego vehicle (that is Figure 4Above C3 in the figure, the optical axis of the first sensor C1 is C11. Based on the characteristic that the first sensor is a camera, the effective sensor area of the first sensor is the dark filled area C12.

[0072] Correspondingly, when multiple cameras are installed on the host vehicle, the target observation area determined for a time point Tr is as Figure 5 shown. The host vehicle 4 is equipped with four cameras 41, 42, 43, and 44. The four target observation areas determined for this time point Tr are the effective sensor area 411 of 41, the effective sensor area 421 of 42, the effective sensor area 431 of 43, and the effective sensor area 441 of 44, respectively.

[0073] For each time point, the target observation area is determined, and then the obstacle points among the obstacle points observed at this time point that are located in the target observation area are obtained as the first candidate obstacle points. Then, the first candidate obstacle points of each time point are aggregated, and based on the first candidate obstacle points of each time point, the obstacle search result of the high-resolution grid is determined.

[0074] In some embodiments, determining the obstacle search result of each high-resolution grid based on the first candidate obstacle points of multiple time points includes: determining the sampling interval based on the size of the high-resolution grid; sampling the first candidate obstacle points of multiple time points based on the sampling interval to obtain sampled obstacle points; and determining the obstacle search result of each high-resolution grid based on the sampled obstacle points.

[0075] In this application, the determined sampling interval is not greater than the side length of the high-resolution grid. For example, if the size of the high-resolution grid is 0.1m × 0.1m, the sampling interval can be 0.1m or 0.09m, etc. In this way, when obtaining the sampled obstacle points by sampling from all the first candidate obstacle points, each high-resolution grid can be traversed for sampling.

[0076] After sampling to obtain the sampled obstacle points, the obstacle search result of each high-resolution grid can be determined according to the position where the sampled obstacle points are located.

[0077] S130. Construct a grid map for the second area according to the second resolution to obtain the second grid map.

[0078] Among them, the first resolution is higher than the second resolution. That is to say, the first area where the planned path of the host vehicle is located is used as the key area, and a grid map is constructed with a higher first resolution to obtain the first grid map. Similarly, the second area outside the first area is used as the non-critical area, and a grid map is constructed with a lower second resolution to obtain the second grid map.

[0079] In some embodiments, the second resolution may be the aforementioned initial resolution. In this case, it means that for the second region, the grid map is constructed according to the initial resolution corresponding to the initial grid map to obtain the second grid map, without changing the original resolution of the second region.

[0080] To distinguish the grids of the first resolution and the grids of the second resolution, in this application, the low-resolution grids are used as the grids of the second resolution. That is, a grid in the second grid map is a low-resolution grid.

[0081] As described above, at each time point after the initial grid map is determined, the ego vehicle can observe obstacle points, and based on the obstacle points observed at each time point, determine the obstacle search results of each low-resolution grid at each time point. The determination process of the obstacle search results of each low-resolution grid refers to the determination process of the obstacle search results of each grid in the aforementioned initial grid map, which will not be elaborated here.

[0082] S140. Obtain the target grid map of the target region based on the first grid map and the second grid map.

[0083] As can be known from the above, the first grid map is actually the grid map corresponding to a part of the target region, and the second grid map is also the grid map corresponding to another part of the target region. Therefore, after obtaining the first grid map and the second grid map, directly summarize the first grid map and the second grid map into a large grid map to indicate the target region through this large grid map, and this large grid map is the target grid map of the target region.

[0084] As can be known from the above, the storage of the grid map is actually to store the obstacle search results of the grids in the grid map as the attributes of the grids. Thus, in the target grid map, the part within the first region is high-resolution grids, and the attribute of each high-resolution grid includes its respective obstacle search result; the part within the second region in the target grid map is low-resolution grids, and the attribute of each low-resolution grid includes its respective obstacle search result.

[0085] Generally speaking, for different scenarios, an initial grid map can be determined based on the obstacle points observed at the first time point of the scenario. For each time point after the first time point, the obstacle search results of the high-resolution grid and the obstacle search results of the low-resolution grid are determined in the aforementioned manner at each time point. The first grid map is determined based on the obstacle search results of the high-resolution grid, and the second upper map is determined based on the obstacle search results of the low-resolution grid. Then, the first grid map and the second grid map are merged to obtain the target grid map, so as to realize real-time updating of the corresponding target grid map for each time point. Among them, the scenarios in this application can include parking scenarios, driving scenarios, lane-changing scenarios, and so on.

[0086] For example, for the first time point t1 of the parking scenario, an initial grid map is determined. For the fourth time point t4 of the parking scenario, first, the planned path at the time point t4 is determined, and a first area is determined based on the planned path at the time point t4. Then, the first area is divided into high-resolution grids. After that, the first candidate obstacle points are screened according to the obstacle points observed at t1, t2, t3, and t4, and the sampled obstacle points are obtained. According to the sampled obstacle points, the obstacle search results of the high-resolution grid at the time point t4 are determined. At the same time, according to the obstacle points observed at t4, the obstacle search results of the low-resolution grid are determined. Based on the obstacle search results of the high-resolution grid and the obstacle search results of the low-resolution grid at the time point t4, the target grid map for the time point t4 is obtained.

[0087] In this embodiment, first, based on the planned path of the vehicle traveling in the initial grid map of the target area, a first area is determined in the initial grid map. Then, a grid map of the first area is constructed at a higher first resolution to obtain the first grid map, and a grid map of the second area is constructed at a lower second resolution to obtain the second grid map. Finally, based on the first grid map and the second grid map, the target grid map of the target area is obtained. Thus, it is realized to construct a map of the key area (i.e., the first area) of the vehicle at a larger resolution, so that the accuracy of the grid map in the first area is relatively high, thereby improving the accuracy of the constructed target grid map. At the same time, only the first area is constructed with a grid map at a larger resolution, and the second area is not constructed with a grid map at a larger resolution, thereby reducing the data processing volume of the second area, improving the construction efficiency of the second grid map corresponding to the second area, and further improving the construction efficiency of the entire target grid map.

[0088] In addition, among the obstacle points observed at multiple time points in the present application, the first candidate obstacle points located in the key area (i.e., the target observation area) are determined, so as to achieve the goal of combining the obstacle points at multiple time points to determine the obstacle search result of the high-resolution grid. This can avoid the situation where when only the obstacle points at one time point are used to determine the obstacle search result of the high-resolution grid, inaccurate obstacle points caused by false detection lead to inaccurate obstacle search results of the determined high-resolution grid, effectively improving the determined obstacle search result of the high-resolution grid, and further improving the accuracy of the entire target grid map. At the same time, only the first candidate obstacle points located in the target observation area among the obstacle points at multiple time points are searched, reducing the number of obstacle points and improving the efficiency of determining the obstacle search result of the high-resolution grid based on the first candidate obstacle points, and further improving the construction efficiency of the entire target grid map.

[0089] Furthermore, sampling obstacle points from the first candidate obstacle points at multiple time points does not require determining the obstacle search result of the high-resolution grid based on all the first candidate obstacle points, further reducing the number of obstacle points and improving the determination efficiency of the obstacle search result of the high-resolution grid, and further improving the construction efficiency of the entire target grid map.

[0090] In some embodiments, on the premise that the acquisition process of the foregoing high-resolution grid area is "dividing each grid located in the first area in the initial grid map into multiple high-resolution grids to obtain the high-resolution grid area", if for each grid located in the first area in the initial grid map, the number of high-resolution grids obtained after grid division is an integer multiple of 4, and the second resolution of the second grid map is the same as the initial resolution of the initial grid map, which means that one low-resolution grid in the second grid map includes an integer multiple of 4 high-resolution grids. At this time, S120 may further include: for each grid located in the first area in the initial grid map, using the obstacle search results of the multiple high-resolution grids located in the grid as the leaf nodes of the quadtree data model, constructing the quadtree data of the grid as the target search result corresponding to the grid; based on the target search results of each grid located in the first area in the initial grid map, determining the first grid map.

[0091] For example, as Figure 6As shown in the figure, for any four grids ABCD in the initial grid map, where A is a grid in the first area, which is divided into 4×4 high-resolution grids, and BCD are grids in the second area, without division, serving as low-resolution grids. For grid A, the smallest high-resolution grids (such as the four high-resolution grids IJKL) serve as the leaf nodes of the quadtree data model, and are merged into an intermediate grid H. Similarly, the four high-resolution grids obtained after the division of the intermediate grids EFGH respectively serve as the leaf nodes of the quadtree data model, and the intermediate grids EFGH are merged into a grid A.

[0092] That is to say, in the form of a quadtree, grids of different sizes are compatible to store the grid map, so as to store the first area with high precision (high-resolution grids with smaller sizes) and the second area with low precision (grids with larger sizes) at the same time. When the second resolution is the resolution of the initial grid map, it can avoid the excessive consumption of storage resources and the waste of storage performance caused by separately storing the first grid map of the first area and the second grid map of the second area, and improve the storage performance.

[0093] In some embodiments, as Figure 7 shown, S120 further includes:

[0094] S210. Divide the grids in the first area according to the first resolution to obtain a high-resolution grid area; based on the obstacle points observed by the vehicle at multiple time points and the position of the first sensor used to observe the obstacle points, determine the obstacle search result of each high-resolution grid.

[0095] Among them, the description of S210 refers to the description of S120 in the foregoing embodiments, and will not be elaborated here.

[0096] S220. During the process of the vehicle driving along the planned path, obtain the target time period when the attitude of the vehicle meets the target attitude.

[0097] The target attitude may refer to the attitude of the vehicle when it needs to trace the edges of some surrounding objects (such as a parking space or a road edge, etc.); for example, in the automatic parking scenario, the target attitude may refer to the attitude when the vehicle is basically straightened during the parking process; when the vehicle is within the parking space range and the included angle between the central axis of the vehicle and the central axis of the parking space is less than a preset angle (the preset angle is, for example, 10° etc.), it is determined that the vehicle is basically straightened.

[0098] Another example is that in the driving scenario, the target attitude is for the vehicle to drive along the side or park the vehicle, etc. Driving along the side means that the distance between the vehicle and the lane edge is less than a distance threshold (the distance threshold is, for example, 40 cm).

[0099] The period during which the attitude of the host vehicle satisfies the target attitude is used as the target period, and the target period may include at least one time point.

[0100] S230. Obtain the obstacle points collected by the second sensor of the host vehicle during the target period as the second candidate obstacle points.

[0101] Among them, the second sensor is different from the first sensor. The second sensor may refer to the radar on the host vehicle, and the obstacle points collected by the radar during the target period are used as the second candidate obstacle points.

[0102] S240. Determine the supplementary obstacle search results of each high-resolution grid according to the second candidate obstacle points.

[0103] After obtaining the second candidate obstacle points, the obstacle search results of each high-resolution grid can be determined according to the second obstacle points as the supplementary obstacle search results of each high-resolution grid respectively.

[0104] In some embodiments, S230 may further include: determining the compensation parameter of the second sensor; the compensation parameter is used to indicate the observation error of the second sensor; performing compensation processing on the second candidate obstacle points through the compensation parameter to obtain the compensated obstacle points corresponding to the second candidate obstacle points; determining the supplementary obstacle search results of each high-resolution grid based on the compensated obstacle points.

[0105] The compensation parameter may refer to the gap between the true value and the actual value observed by the second sensor. For example, the distance between the host vehicle observed by the radar of the host vehicle and the obstacle point a1 is 100 cm, and the true distance between the host vehicle and the obstacle point a1 is 110 cm, then the compensation parameter is 10 cm.

[0106] Generally, before the host vehicle observes using the second sensor, the second sensor can be calibrated to determine the compensation parameter.

[0107] Exemplarily, measure the measured values of multiple specified distances through the second sensor, and determine the compensation parameter based on the average value of the differences between the measured values of the multiple specified distances and the true values. Generally, the compensation parameter may include positive and negative values. A positive compensation parameter means that the measured value of the second sensor is less than the true value, and a negative compensation parameter means that the measured value of the second sensor is greater than the true value.

[0108] For each second candidate obstacle point, perform compensation processing on the second candidate obstacle point according to the compensation parameter to obtain the result after compensating the second candidate obstacle point, which is used as the compensated obstacle point corresponding to the second candidate obstacle point. Generally speaking, the second candidate obstacle point can be directly moved by a specified distance to obtain the compensated obstacle point. For example, if the measurement value of the second sensor is less than the true value, move the second candidate obstacle point in the direction away from the second sensor of the host vehicle by the distance value indicated by the compensation parameter to obtain the compensated obstacle point corresponding to the second candidate obstacle point. Another example is that if the measurement value of the second sensor is greater than the true value, move the second candidate obstacle point in the direction close to the second sensor of the host vehicle by the distance value indicated by the compensation parameter to obtain the compensated obstacle point corresponding to the second candidate obstacle point.

[0109] After obtaining the compensated obstacle points, the supplementary obstacle search results corresponding to each high-resolution grid mark can be determined based on the compensated obstacle points, that is, whether there is an obstacle point detected by the second sensor in the high-resolution grid.

[0110] Of course, the number of determined second candidate obstacle points may be relatively large. Therefore, an obstacle point can be sampled at a specified interval (for example, the specified interval is 0.1 m) to sample the third candidate obstacle points from the second candidate obstacle points to obtain evenly distributed and fewer third candidate obstacle points. Then, based on the compensation parameter, perform compensation on the third candidate obstacle points to obtain the compensated obstacle points corresponding to the third candidate obstacle points. Finally, based on the compensated obstacle points, determine the supplementary obstacle search results corresponding to each high-resolution grid mark, that is, whether there is an obstacle point detected by the second sensor in the high-resolution grid.

[0111] As Figure 8 shown, the compensated obstacle points determined for the host vehicle 8 are Figure 8 the hollow circles in, and the compensated obstacle points are evenly distributed on the side of the obstacles 81 and 82 close to the host vehicle 8.

[0112] S250. Determine the first grid map according to the obstacle search results and supplementary obstacle search results of each high-resolution grid.

[0113] Obtain the obstacle search results and supplementary obstacle search results of each high-resolution grid, and store the obstacle search results and supplementary obstacle search results of each high-resolution grid as different-dimensional attribute data of each high-resolution grid (the obstacle search result is the attribute data of one dimension, and the supplementary obstacle search result is the attribute data of another dimension) in each high-resolution grid, so as to obtain the stored high-resolution grid, which is used as the first grid map.

[0114] Specifically, the supplementary obstacle search results of each high-resolution grid can be stored as the attribute data of the high-resolution grid in the form of a structure of high obstacle points.

[0115] In this embodiment, when the host vehicle is in the target pose, the second candidate obstacle points collected by the second sensor of the host vehicle are also introduced to determine the supplementary obstacle search results of the high-resolution grid according to the second candidate obstacle points, so that there are two types of obstacle search results corresponding to the high-resolution grid, namely the corresponding obstacle search results and the supplementary obstacle search results, making the high-resolution grid include more information, the information in the first area where the high-resolution grid is located is more complete and accurate, and the accuracy of the first area is relatively high, thereby improving the accuracy of the second grid map.

[0116] In some embodiments, the map construction process is as Figure 9 shown. First, an initial grid map with low precision (that is, a grid map with a larger grid size) is established. After that, the planned path of the host vehicle is determined in the initial grid map, and the first area and the second area are divided based on the planned path of the host vehicle. The first area is a high-precision requirement area including the planned path, and the second area is a low-precision requirement area. Then, low-precision mapping is performed on the second area - based on the obstacle points observed by the host vehicle, the obstacle search results of each grid in the second area are determined, that is, the grids in the second area are not divided and directly used as low-resolution grids.

[0117] Each grid in the first area is divided into high-resolution grids, and then point taking and sampling are performed to obtain sampled obstacle points: the obstacle points at multiple time points are summarized, and the first candidate obstacle points in the target observation area at each time point are obtained, and then the sampled obstacle points are obtained by sampling among the first candidate obstacle points.

[0118] Meanwhile, within the target time period when the host vehicle meets the target pose, based on the radar obstacle points observed by the radar, the compensation obstacle points for the radar are determined.

[0119] After that, high-precision mapping is performed: based on the sampled obstacle points and the compensation obstacle points, the obstacle search results and the supplementary obstacle search results of each high-resolution grid are determined.

[0120] Finally, based on the obstacle search results of each low-resolution grid in the second area, the obstacle search results and the supplementary obstacle search results of each high-resolution grid, quadtree mapping merging is performed to obtain the target grid map.

[0121] Refer to the appendix Figure 10 , Figure 10 which shows the structural block diagram of a map construction device proposed in an embodiment of the present application. For the host vehicle, device 1000 includes:

[0122] The first determination module 1010 is configured to determine a first area in the initial grid map based on the planned path of the vehicle traveling in the target area in the initial grid map; the first area includes at least the planned path;

[0123] The first construction module 1020 is configured to construct a grid map for the first area according to a first resolution to obtain a first grid map;

[0124] The second construction module 1030 is configured to construct a grid map for a second area according to a second resolution to obtain a second grid map; the second area is the area in the initial grid map except the first area; the first resolution is higher than the second resolution;

[0125] The second determination module 1040 is configured to obtain a target grid map of the target area based on the first grid map and the second grid map.

[0126] Optionally, the first construction module 1020 is further configured to divide the grid of the first area according to the first resolution to obtain a high-resolution grid area; the high-resolution grid area includes a plurality of high-resolution grids; based on the obstacle points observed by the vehicle at multiple time points and the position of the first sensor for observing the obstacle points, determine the obstacle search result of each high-resolution grid; based on the obstacle search result of each high-resolution grid, determine the first grid map.

[0127] Optionally, the first construction module 1020 is further configured to determine the respective target observation areas at each time point based on the position of the first sensor; obtain the first candidate obstacle points located in the corresponding target observation areas from the obstacle points at each time point; based on the first candidate obstacle points at multiple time points, determine the obstacle search result of each high-resolution grid.

[0128] Optionally, the first construction module 1020 is further configured to determine a sampling interval based on the size of the high-resolution grid; sample the first candidate obstacle points at multiple time points based on the sampling interval to obtain sampled obstacle points; based on the sampled obstacle points, determine the obstacle search result of each high-resolution grid.

[0129] Optionally, the first construction module 1020 is further configured to, during the process of the vehicle traveling along the planned path, obtain a target time period when the attitude of the vehicle meets the target attitude; obtain the obstacle points collected by the second sensor of the vehicle during the target time period as second candidate obstacle points; the second sensor is different from the first sensor; determine the supplementary obstacle search result of each high-resolution grid according to the second candidate obstacle points; determine the first grid map according to the obstacle search result and the supplementary obstacle search result of each high-resolution grid.

[0130] Optionally, the first construction module 1020 is further configured to determine compensation parameters for the second sensor; the compensation parameters are used to indicate the observation error of the second sensor; the second candidate obstacle points are compensated by the compensation parameters to obtain compensated obstacle points corresponding to the second candidate obstacle points; based on the compensated obstacle points, the supplementary obstacle search results of each high-resolution grid are determined.

[0131] Optionally, the first construction module 1020 is further configured to divide each grid located in the first area in the initial grid map into a plurality of high-resolution grids to obtain a high-resolution grid area.

[0132] Optionally, for each grid located in the first area in the initial grid map, the number of high-resolution grids obtained after grid division is an integer multiple of 4; the first construction module 1020 is further configured to, for each grid located in the first area in the initial grid map, use the obstacle search results of the multiple high-resolution grids located within the grid as leaf nodes of a quadtree data model, construct the quadtree data of the grid as the target search result corresponding to the grid; based on the target search results of each grid located in the first area in the initial grid map, the first grid map is determined.

[0133] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described devices and modules can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0134] In addition, in each embodiment of the present application, the various functions may be integrated in one processing module, or each module may exist physically alone, or two or more modules may be integrated in one module. The above-integrated modules may be implemented in the form of hardware or in the form of software function modules.

[0135] On the other hand, the present application further provides a computer-readable storage medium, in which program code is stored, and the program code can be called by a processor to execute the method described in the foregoing method embodiments.

[0136] A computer-readable storage medium may be an electronic memory such as a flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, a hard disk, or a ROM. Optionally, the computer-readable storage medium includes a non-transitory computer-readable storage medium. The computer-readable storage medium has a storage space for program code that executes any of the method steps in the above-described methods. These program codes may be read from or written to one or more computer program products. The program codes may be compressed in a suitable form, for example.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A map construction method, characterized in that: The method comprises: Based on a planned path of the vehicle traveling in an initial grid map of a target area, determining a first area in the initial grid map; the first area at least includes the planned path; Constructing a grid map for the first area according to a first resolution to obtain the first grid map; Constructing a grid map for a second area according to a second resolution to obtain a second grid map; the second area is an area in the initial grid map excluding the first area; the first resolution is higher than the second resolution; A target grid map of the target area is obtained based on the first grid map and the second grid map.

2. The method according to claim 1, characterized in that The step of constructing a grid map for the first area according to the first resolution to obtain the first grid map includes: Dividing the first region into grids according to the first resolution to obtain a high-resolution grid region; the high-resolution grid region includes a plurality of high-resolution grids; Determine obstacle search results for each of the high-resolution grids based on obstacle points observed by the vehicle at multiple time points and a location of a first sensor for observing the obstacle points; A first grid map is determined based on the obstacle search results of each of the high-resolution grids.

3. The method according to claim 2, characterized in that The determining of the obstacle search results of each high-resolution grid based on the obstacle points observed by the vehicle at multiple time points and the position of the first sensor used to observe the obstacle points comprises: Determine a target observation area at each of the time points based on the location of the first sensor; Acquire a first candidate obstacle point located in the corresponding target observation area from the obstacle points at each of the time points; Based on the first candidate obstacle points at the multiple time points, an obstacle search result of each of the high-resolution grids is determined.

4. The method according to claim 3, characterized in that: The determining, based on the first candidate obstacle points at the plurality of time points, an obstacle search result of each high-resolution grid comprises: Determining a sampling interval based on the size of the high-resolution grid; Sampling the first candidate obstacle points at the plurality of time points based on the sampling interval to obtain sampled obstacle points; Based on the sampled obstacle points, an obstacle search result of each of the high-resolution grids is determined.

5. The method according to claim 2, characterized in that: The step of determining a first grid map based on the obstacle search results of each high-resolution grid comprises: During the driving of the vehicle according to the planned path, obtaining a target time period during which the posture of the vehicle meets the target posture; Obtaining an obstacle point collected by a second sensor of the vehicle during the target period as a second candidate obstacle point; the second sensor is different from the first sensor; Determine, according to the second candidate obstacle point, a supplementary obstacle search result for each of the high-resolution grids; A first grid map is determined according to the obstacle search results and the supplementary obstacle search results of each of the high-resolution grids.

6. The method according to claim 5, characterized in that The step of determining, according to the second candidate obstacle point, a supplementary obstacle search result for each of the high-resolution grids includes: Determining a compensation parameter of the second sensor; the compensation parameter is used to indicate an observation error of the second sensor; Performing compensation processing on the second candidate obstacle point by using the compensation parameter to obtain a compensated obstacle point corresponding to the second candidate obstacle point; Based on the compensated obstacle points, a supplementary obstacle search result for each of the high-resolution grids is determined.

7. The method according to claim 2, characterized in that The step of dividing the first area into grids according to the first resolution to obtain a high-resolution grid area includes: Each grid in the first area of ​​the initial grid map is divided into a plurality of high-resolution grids to obtain the high-resolution grid area.

8. The method according to claim 7, characterized in that For each grid in the first area in the initial grid map, the number of high-resolution grids obtained after dividing the grid is an integer multiple of 4; The step of determining a first grid map based on the obstacle search results of each high-resolution grid comprises: For each grid in the first area in the initial grid map, using obstacle search results of multiple high-resolution grids in the grid as leaf nodes of a quadtree data model, and constructing quadtree data of the grid as target search results corresponding to the grid; A first grid map is determined based on the target search result of each grid in the initial grid map that is located within the first area.

9. A map construction device, characterized in that: The device comprises: A first determination module, configured to determine a first area in an initial grid map based on a planned path of the vehicle traveling in the initial grid map of the target area; the first area at least includes the planned path; A first construction module, configured to construct a grid map for the first area according to a first resolution to obtain the first grid map; A second construction module is used to construct a grid map for a second area according to a second resolution to obtain a second grid map; the second area is an area in the initial grid map excluding the first area; the first resolution is higher than the second resolution; The second determining module is used to obtain a target grid map of the target area based on the first grid map and the second grid map.

10. A vehicle, characterized in that: include: one or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more applications are configured to execute the method according to any one of claims 1-8.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program code executable by a processor, and when the program code is executed by the processor, the processor executes the method according to any one of claims 1 to 8.

Images