A method and system for extracting wheel track elevation, and an electronic device

Abstract

The present application discloses a method, a system and an electronic device for extracting wheel trajectory elevation. Among them, the method includes: based on the V-disparity map and vehicle driving information, obtaining a road surface fitting equation and a wheel trajectory V-map; based on the valid points in the wheel trajectory V-map, using a preset reward function to generate a reward map corresponding to the wheel trajectory V-map; according to a preset path rule, calculating the cumulative reward values of each point within the valid interval in the reward map A to generate a cumulative reward map M; in the cumulative reward map, searching row by row for the maximum cumulative reward value, and based on the maximum cumulative reward value, generating a reward path; calculating the difference between the reward path and the road surface fitting equation to obtain wheel trajectory elevation data. Through the technical solution in the present application, the spatial characteristics of the road surface are fully utilized to generate a dynamic and reasonable reward path, and then the wheel trajectory elevation data at the corresponding position in the wheel driving area is obtained, improving the accuracy of the elevation data.

Description

Technical Field

[0001] The present application relates to the technical field of autonomous driving. Specifically, it relates to a method for extracting wheel trajectory elevation, a system for extracting wheel trajectory elevation, and an electronic device. Background Art

[0002] In the existing autonomous driving technology, in order to improve the driving comfort of the vehicle, the suspension of the vehicle can be adjusted according to the elevation of the road surface (such as potholes), so that the driver can still drive smoothly when passing through potholes and bumpy roads. Therefore, it is necessary to accurately detect the road surface elevation.

[0003] In the existing technology, it can usually be achieved through the following two methods:

[0004] Implementation method one: Calculate the road surface elevation according to the parallax of a single pixel, extract the road surface elevation, and then use the time series data to perform weighted fusion on multiple frames of elevation to obtain the final road surface elevation.

[0005] Implementation method two: Perform (weighted) statistics on a small area to obtain the representative parallax of this area, and then calculate the road surface elevation.

[0006] However, these implementation methods are all line-of-sight methods proposed based on single-point parallax. Therefore, when the quality of the obtained parallax is poor, the output elevation data will be inaccurate and fluctuate greatly, and moreover, the spatial information of the road surface is not considered. Summary of the Invention

[0007] The purpose of the present application is: how to improve the accuracy of obtaining the road surface elevation data in front of the vehicle.

[0008] The technical solution of the first aspect of the present application is: providing a method for extracting wheel trajectory elevation, the method includes: Step 1: Based on the V parallax map and the vehicle driving information, obtain the road surface fitting equation and the wheel trajectory V map; Step 2: Based on the valid points in the wheel trajectory V map, use a preset reward function to generate a reward map A corresponding to the wheel trajectory V map; Step 3: According to the preset path rule, calculate the cumulative reward value of each point in the valid interval in the reward map A to generate a cumulative reward map M; where the valid interval is determined by the valid points in the wheel trajectory V map; Step 4: In the cumulative reward map M, search for the maximum cumulative reward value row by row, and generate a reward path based on the maximum cumulative reward value; Step 5: Calculate the difference between the reward path and the road surface fitting equation to obtain the wheel trajectory elevation data.

[0009] In any of the above technical solutions, further, the method further includes: performing an average weighted operation on the obtained wheel trajectory elevation data corresponding to multiple frames of V parallax maps to generate a final wheel trajectory elevation curve.

[0010] In any of the above technical solutions, further, in step 2, it specifically includes: Step 21: Based on the valid points in the wheel trajectory V diagram, using a preset reward function, calculate the reward values row by row and column by column to obtain a row reward value map and a column reward value map respectively; Step 22: Use the method of matrix addition to calculate the sum value between the row reward value and the column reward value, denoted as reward map A.

[0011] In any of the above technical solutions, further, in step 21, the calculation methods of the row reward value map and the column reward value map are the same. Among them, the calculation process of the row reward value map specifically includes: Step 211: Traverse the valid points in each row of the wheel trajectory V diagram, determine the horizontal valid width, and determine the median of the horizontal width; Step 212: Calculate the parallax centroid position of the current row according to the parallax value of the valid points in each row of the wheel trajectory V diagram; Step 213: Calculate the row reward basis based on the parallax centroid position and the median of the horizontal width using Gaussian operation; Step 214: Based on the row reward basis, use a preset reward function to calculate the reward value row by row to obtain the row reward value map.

[0012] In any of the above technical solutions, further, the method further includes: when it is determined that the number of valid points in any row of the wheel trajectory V diagram is less than or equal to the first quantity threshold, record the reward value of the corresponding row in the row reward value map as 0; and / or when it is determined that the number of valid points in any column of the wheel trajectory V diagram is less than or equal to the second quantity threshold, record the reward value of the corresponding column in the column reward value map as 0.

[0013] In any of the above technical solutions, further, the method further includes: statistically count the number of valid points in the wheel trajectory V diagram column by column to generate a column histogram; select the weighted evaluation value in the column histogram; when it is determined that the weighted evaluation value is less than or equal to the weighted threshold, perform a weighted operation on the reward value of the corresponding row in the column reward value map according to the number of valid points in any row of the wheel trajectory V diagram to update the column reward value map.

[0014] In any of the above technical solutions, further, step 3 specifically includes: Step 31: Starting from the lower right corner of the reward map A, in the order from right to left and from bottom to top, sequentially confirm the cumulative reward value of each point in any valid interval in the reward map A, where the calculation formula for the cumulative reward value is:

[0015]

[0016] In the formula, (d i , y j ) is the point coordinate of the i-th row and j-th column in the reward map A, and A(d i , y j ) is the point coordinate (d i , yj ) The corresponding reward value, M(d i+1 , y j ), M(d i , y j+1 ), M(d i+1 , y j+1 ) are the cumulative reward values corresponding to the point coordinates (d i+1 , y j ), (d i , y j+1 ), (d i+1 , y j+1 ), respectively; Step 32: Based on the cumulative reward value, generate a cumulative reward map M.

[0017] The technical solution of the second aspect of the present application is: to provide a wheel trajectory elevation extraction system, the system includes: a fitting unit configured to obtain a road surface fitting equation and a wheel trajectory V map based on the V disparity map and vehicle driving information; a reward map unit configured to generate a reward map A corresponding to the wheel trajectory V map by using a preset reward function based on the valid points in the wheel trajectory V map; a cumulative reward map unit configured to calculate the cumulative reward values of each point in the valid interval in the reward map A according to a preset path rule and generate a cumulative reward map M; where the valid interval is determined by the valid points in the wheel trajectory V map; a reward path unit configured to search for the maximum cumulative reward value row by row in the cumulative reward map M and generate a reward path based on the maximum cumulative reward value; an elevation extraction unit configured to calculate the difference between the reward path and the road surface fitting equation to obtain wheel trajectory elevation data.

[0018] In any of the above technical solutions, further, the elevation extraction unit is further configured to: perform an average weighted operation on the wheel trajectory elevation data corresponding to multiple frames of V disparity maps to generate a final wheel trajectory elevation curve.

[0019] The technical solution of the third aspect of the present application is: to provide an electronic device, the electronic device includes: a processor;

[0020] a memory for storing processor-executable instructions;

[0021] wherein, when the processor is configured to execute the executable instructions, it implements the wheel trajectory elevation extraction method described in any one of the technical solutions of the first aspect above.

[0022] The beneficial effects of the present application are:

[0023] The technical solution in this application, based on the wheel trajectory V map and the corresponding reward mechanism, makes full use of the spatial characteristics of the road surface shown in the V map to obtain the corresponding cumulative reward map M, and then obtains the final reward path and the corresponding elevation data, reducing the influence of single-point parallax on the extraction of elevation data and helping to improve the accuracy of wheel trajectory elevation data. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and / or additional aspects of the present application will become apparent and be readily understood in the description of the embodiments in conjunction with the following drawings, where:

[0025] Figure 1 is a schematic flowchart of a method for extracting wheel trajectory elevation according to an embodiment of the present application;

[0026] Figure 2 is a schematic diagram of a wheel driving area according to an embodiment of the present application;

[0027] Figure 3 is a schematic diagram of a left wheel trajectory V map according to an embodiment of the present application;

[0028] Figure 4 is a schematic diagram of the reward value of a left wheel trajectory according to an embodiment of the present application;

[0029] Figure 5 is a schematic diagram of the cumulative reward value map corresponding to a left wheel trajectory according to an embodiment of the present application;

[0030] Figure 6 is a schematic diagram of a weighted wheel trajectory elevation curve according to an embodiment of the present application;

[0031] Figure 7 is a schematic block diagram of a wheel trajectory elevation extraction system according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] In order to more clearly understand the above objects, features and advantages of the present application, the present application will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments may be combined with each other.

[0033] In the following description, many specific details are set forth in order to fully understand the present application. However, the present application may be implemented in other ways different from those described herein. Therefore, the protection scope of the present application is not limited by the specific embodiments disclosed below.

[0034] Embodiment 1

[0035] After obtaining the left and right images, the binocular camera can obtain the corresponding disparity map based on the principle of disparity calculation, and then can calculate the distance between the front target and the binocular camera. Therefore, after the binocular camera is installed on the vehicle, the elevation of the front road surface can be calculated according to the disparity value.

[0036] As Figure 1 shown, this embodiment provides a method for extracting the elevation of the wheel trajectory, and the method includes:

[0037] Step 1: Based on the V disparity map and the vehicle driving information, obtain the road surface fitting equation and the wheel trajectory V map;

[0038] Step 2: Based on the valid points in the wheel trajectory V map, use a preset reward function to generate a reward map A corresponding to the wheel trajectory V map; where the valid points are the points with non-zero disparity values.

[0039] Step 3: According to the preset path rule, calculate the cumulative reward values of the points in the valid interval of the reward map A to generate a cumulative reward map M; where the valid interval is determined by the valid points in the wheel trajectory V map; that is, for any row in the wheel trajectory V map, if there are invalid points between several valid points in this row, the area corresponding to these valid points and the invalid points in the middle is used as the valid interval of this row corresponding to the reward map A.

[0040] Step 4: In the cumulative reward map M, search for the maximum cumulative reward value row by row, and generate a reward path based on the maximum cumulative reward value;

[0041] Step 5: Calculate the difference between the reward path and the road surface fitting equation to obtain the wheel trajectory elevation data.

[0042] Specifically, the disparity map can be converted by projection to obtain the V disparity map, and the specific process will not be elaborated here. Among them, the abscissa in the V disparity map is the disparity value, and the ordinate is the row of the image.

[0043] In this embodiment, a plane equation is selected to fit the road surface, and the corresponding calculation formula is:

[0044] x*A + y*B + d*C + D = 0

[0045] In the formula, A, B, C, and D are fitting parameters, x is the row coordinate in the image, y is the column coordinate in the image, and d is the disparity value corresponding to the pixel point (x, y).

[0046] Considering the inevitable error when the disparity map is projected onto the V-disparity map, which makes the represented road surface have a certain thickness. Therefore, in this embodiment, the lowest point in each column of the V-disparity map is selected as the fitting data point, that is, the lowest point in the image for any disparity value. After selecting the fitting data points, the equation of the plane where the road surface is located can be obtained by fitting (such as the least squares method), denoted as the road surface fitting equation.

[0047] It should be noted that the road surface fitting equation can be simplified to a straight-line model, that is, A = 0. At this time, it has nothing to do with the row coordinates in the image (left image or right image), that is, the image has no left or right tilt.

[0048] As Figure 2 shown, after this embodiment obtains the vehicle driving information (including at least the wheel rotation angle and vehicle speed), based on the kinematic model of the vehicle, the turning radius 201 of the vehicle can be calculated. Furthermore, based on information such as the fitting road surface equation, the distance between the left and right wheels, and the wheel width, the driving area 202 of the wheels can be obtained.

[0049] Since the image (such as the left image) obtained by the binocular camera corresponds to the disparity map, therefore, based on the above driving area 202, the corresponding area of the wheels in the disparity map can be obtained, and then the wheel trajectory V map can be obtained, where the wheel trajectory V map includes the left wheel trajectory V map and the right wheel trajectory V map.

[0050] It should be noted that the processing methods for the left and right wheel trajectory V maps in this embodiment are the same. Among them, the left wheel trajectory V map is as Figure 3 shown, where the abscissa is the disparity value d (the value increases sequentially from left to right), the ordinate is the number of rows of the image (the value increases sequentially from top to bottom), and the corresponding boundaries are [DMIN, DMAX] and [YMIN, YMAX] respectively.

[0051] In this embodiment, a reward value model can be established based on historical data as a preset reward function. Using the coordinates (d, y) of the disparity point in the wheel trajectory V map, the corresponding reward value can be calculated, and then the corresponding reward value, such as A(d, y), can be obtained. Furthermore, the reward map A corresponding to the wheel trajectory V map can be obtained.

[0052] Those skilled in the art can understand that in the wheel trajectory V diagram, the distance between the line-of-sight intersection point below the image and the vehicle (binocular camera) is less than the distance between the line-of-sight intersection point above the image and the vehicle, and the road surface should be continuous. Additionally, within a certain distance range, the parallax value at a closer distance is more accurate than that at a farther distance. Therefore, this embodiment proposes the concept of a cumulative reward map. Based on the reward map A, along the extension direction of the road surface, the cumulative reward values at each position in the reward map A are calculated in sequence to generate the cumulative reward map M, which not only utilizes the spatial information of the road surface but also reduces the influence of noise points in the parallax map on the reward path. Furthermore, the maximum cumulative reward value in each row of the cumulative reward map M is selected as the reward path, and the elevation data of the wheel trajectory at the corresponding position in the wheel driving area can be obtained through the difference between it and the road surface fitting equation.

[0053] Through the technical solution in this embodiment, based on the wheel trajectory V diagram and the corresponding reward mechanism, the spatial characteristics of the road surface shown in the V diagram are fully utilized to obtain the corresponding cumulative reward map M, and then the final reward path and the corresponding elevation data are obtained, reducing the influence of single-point parallax on the extraction of elevation data and helping to improve the accuracy of the wheel trajectory elevation data.

[0054] In any of the above embodiments, further, to improve the accuracy of the reward map A, the row data and column data of the wheel trajectory V diagram are processed separately, and then the two reward results are added together to obtain the final reward map A. In the above step 2, it specifically includes:

[0055] Step 21: Based on the valid points in the wheel trajectory V diagram, the reward values are filled row by row and column by column to obtain the row reward value and the column reward value respectively;

[0056] Step 22: Using the method of matrix addition, calculate the sum value between the row reward value and the column reward value, denoted as the reward map A.

[0057] Specifically, a preset reward function can be set as A(d, y), corresponding to the reward value at the coordinate point (d, y) in the wheel trajectory V diagram, where the preset reward function can be obtained by fitting based on historical data.

[0058] Those skilled in the art can understand that there will be parallax holes in the parallax map itself, that is, there will be points with a parallax value of 0 at certain positions in the parallax map, and the distribution of parallax in the V map is relatively random. For a certain row in the V map of the wheel trajectory, there may be "holes" (invalid points) between the valid points in the row, which will cause the discontinuity of the road surface, that is, the real path may pass through such points. Therefore, it is necessary to fill such invalid points. So, in this embodiment, a reward map is designed to calculate the row reward value and the column reward value respectively, so that the calculated reward value can be selected for filling, such as using the interpolation filling method, or directly filling during the process of calculating the row and column reward values. Then, by performing the summation calculation of the row and column reward values, a more accurate reward map A can be obtained.

[0059] It should be noted that the methods of filling the reward values row by row and column by column can be the same or different.

[0060] In any of the above embodiments, further, in order to improve the accuracy of the reward map A and solve the problem of parallax loss caused by parallax jitter, this embodiment also proposes a method for calculating the row and column reward values. In step 21 above, the calculation methods of the row reward value map and the column reward value map are the same. Among them, the calculation process of the row reward value map specifically includes:

[0061] Step 211: Traverse the valid points in each row of the V map of the wheel trajectory to determine the horizontal valid width w y of each row, and then count all the valid widths w y , and determine the median of the horizontal widths, denoted as w; among them, the valid width w y is the difference between the coordinates of the valid points at the left and right ends among the valid points in this row.

[0062] Step 212: Calculate the centroid position of the parallax of the current row according to the parallax value d of the valid points in each row of the V map of the wheel trajectory, that is, the data centroid of the valid points in the current row, and record its position as (x i , y j );

[0063] Step 213: Calculate the row reward basis corresponding to each row based on the centroid position of the parallax and the median of the horizontal width through Gaussian operation;

[0064] Specifically, this embodiment uses Gaussian operation to fill the parallax holes, and the corresponding operation conditions are set as:

[0065] 1. The filling interval is [d i - w / 2, d i + w / 2];

[0066] 2. The mean value of the filling Gaussian function is: d i(Row centroid position (x i , y j ) corresponding disparity value);

[0067] 3. The variance of the filled Gaussian function is: w / 6.

[0068] The specific Gaussian operation process will not be elaborated here.

[0069] Step 214: Based on the row reward basis, use the preset reward function A(d, y) to calculate the reward value row by row to obtain the row reward value map.

[0070] As Figure 4 shown, the reward value map obtained by the above method in this embodiment improves the possibility of retaining the best path, making it close to the real path. While solving the disparity hole problem in the V map, it helps to improve the accuracy of elevation data.

[0071] In any of the above embodiments, further, since there may be noise points in the disparity, therefore, in this embodiment, a threshold selection method can be used for noise reduction processing. Therefore, the above method further includes:

[0072] When it is determined that the number of valid points in any row of the wheel track V map is less than or equal to the first quantity threshold, the reward value of the corresponding row in the row reward value map is recorded as 0; and / or

[0073] When it is determined that the number of valid points in any column of the wheel track V map is less than or equal to the second quantity threshold, the reward value of the corresponding column in the column reward value map is recorded as 0.

[0074] It should be noted that the value of the first quantity threshold can be the same as or different from the value of the second quantity threshold, and its value size can be set according to actual needs.

[0075] Those skilled in the art can understand that in order to increase the weight of the reward value at certain positions in the reward map, and considering the different meanings represented by the row and column data in the wheel track V map, different weights can be set for the rows and columns respectively. For example, the row weight is set to 0.8 and the column weight is set to 0.2.

[0076] In any of the above embodiments, further, in order to increase the positive impact of some regions with better disparity (i.e., more valid points) on the reward path, the weight of the reward value is increased by using the disparity density. Therefore, the above method further includes:

[0077] Count the number of valid points in the wheel track V map column by column to obtain the thickness of valid points in each column, and generate a column histogram;

[0078] Select the weighted evaluation value in the column histogram; wherein, the weighted evaluation value can be determined as the median of the column histogram or a value satisfying a certain proportion.

[0079] When it is determined that the weighted evaluation value is less than or equal to the weighted threshold, the reward values in the corresponding rows of the column reward value map are weighted according to the number of valid points in any row of the wheel trajectory V map, and the column reward value map is updated.

[0080] In any of the above embodiments, further, in order to implement dynamic path planning, based on the design idea of the minimum unit path and combined with the characteristics of the road surface space, this embodiment also proposes a method for generating the cumulative reward map M. In step 3 above, it specifically includes:

[0081] Step 31: Starting from the lower right corner of the reward map A, in the order from right to left and from bottom to top (that is, first determine the row and then traverse from right to left), according to the preset path rule, sequentially confirm the cumulative reward values of each point in any valid interval of the reward map A. The calculation formula for the cumulative reward value is:

[0082]

[0083] In the formula, (d i , y j ) is the point coordinate of the i-th row and j-th column in the reward map A, A(d i , y j ) is the reward value corresponding to the point coordinate (d i , y j ), M(d i+1 , y j ), M(d i , y j+1 ), M(d i+1 , y j+1 ) are the cumulative reward values corresponding to the point coordinates (d i+1 , y j ), (d i , y j+1 ), (d i+1 , y j+1 ) respectively;

[0084] Step 32: Generate the cumulative reward map M based on the cumulative reward value.

[0085] Specifically, in this embodiment, starting from the lower right corner of the reward map A, that is, the position (DMAX, YMAX), in the order from right to left and from bottom to top, according to the preset path rule, the valid points in the reward map A are traversed, and the reward values are accumulated. Among them, the valid points are the points with non-zero parallax values in the wheel trajectory V map. For any valid point (d i , yj ) For this purpose, the preset path rule is one of the following:

[0086] i. Right side: (d i+1 , y j )

[0087] ii. Bottom side: (d i , y j+1 )

[0088] iii. Lower right side: (d i+1 , y j+1 )

[0089] Therefore, the calculation formula for the cumulative reward value of any valid point is:

[0090]

[0091] That is, select the maximum value of its own reward value plus the cumulative reward value of the corresponding point under the legal path. The generated cumulative reward map M is as Figure 5 shown.

[0092] After traversing all valid points, record the position of the maximum cumulative reward value in the last traversed row of the cumulative reward map M as the path recovery starting point, and start to recover the reward path from top to bottom (YMIN--->YMAX) in each row of the wheel track V diagram.

[0093] Since the obtained road surface fitting equation is a straight line in the V parallax diagram, set the corresponding equation as: y = d * k + b, where d is the abscissa of the V diagram, y is the ordinate of the V diagram, k is the slope, and b is the intercept. Among them, k = -C / B, b = -D / B.

[0094] Therefore, when obtaining the reward path, the corresponding set is P = {(d i , y j ) | (d i , y j ) belongs to the maximum cumulative reward value trajectory line}.

[0095] For any point, the road surface position can be obtained through the road surface fitting equation: y road = k * d i + b. The corresponding trajectory position y p can also be determined through the reward path.

[0096] Therefore, take the difference δ y = y p - y road , and through binocular ranging and geometric triangle relationship, perform transformation to obtain the true physical height of the road surface, that is, the corresponding elevation data at the wheel track.

[0097] In any of the above embodiments, further, the method further includes:

[0098] Performing an average weighted operation on the wheel track elevation data corresponding to the obtained multiple frames of V-disparity maps to generate a final wheel track elevation curve.

[0099] As Figure 6 shown, considering the instability of single-frame data, an average weighted operation can also be performed on multiple consecutive V-disparity maps, making the output elevation curve smoother, as Figure 6 shown by the black solid line in

[0100] Embodiment 2

[0101] As Figure 7 shown, this embodiment provides a wheel track elevation extraction system 100, and the system 100 includes:

[0102] A fitting unit 101, which is configured to obtain a road surface fitting equation and a wheel track V map based on the V-disparity map and vehicle driving information;

[0103] A reward map unit 102, which is configured to generate a reward map A corresponding to the wheel track V map by using a preset reward function based on the valid points in the wheel track V map;

[0104] An accumulated reward map unit 103, which is configured to calculate the accumulated reward values of each point within the valid interval in the reward map A according to a preset path rule to generate an accumulated reward map M; wherein, the valid interval is determined by the valid points in the wheel track V map;

[0105] A reward path unit 104, which is configured to search for the maximum accumulated reward value row by row in the accumulated reward map M and generate a reward path based on the maximum accumulated reward value;

[0106] An elevation extraction unit 105, which is configured to calculate the difference between the reward path and the road surface fitting equation to obtain the wheel track elevation data.

[0107] In any of the above embodiments, further, the elevation extraction unit 105 is further configured to:

[0108] Performing an average weighted operation on the wheel track elevation data corresponding to the obtained multiple frames of V-disparity maps to generate a final wheel track elevation curve.

[0109] In any of the above embodiments, further, the reward map unit 102 is further configured to:

[0110] Based on the valid points in the wheel trajectory V diagram, using a preset reward function, calculate the reward values row by row and column by column to obtain the row reward value map and the column reward value map respectively;

[0111] Use the method of matrix addition to calculate the sum value between the row reward value and the column reward value, denoted as reward map A.

[0112] In any of the above embodiments, further, the calculation methods of the row reward value map and the column reward value map are the same. Among them, the calculation process of the row reward value map specifically includes:

[0113] Traverse the valid points in each row of the wheel trajectory V diagram to determine the horizontal effective width and the median of the horizontal width;

[0114] Calculate the centroid position of the parallax of the current row according to the parallax value of the valid points in each row of the wheel trajectory V diagram;

[0115] Calculate the row reward basis based on the centroid position of the parallax and the median of the horizontal width using Gaussian operation;

[0116] Based on the row reward basis, use the preset reward function to calculate the reward values row by row to obtain the row reward value map.

[0117] Embodiment 3

[0118] On the basis of the above embodiments, this embodiment also provides an electronic device, which includes: a processor;

[0119] A memory for storing executable instructions of the processor;

[0120] Among them, when the processor is configured to execute the executable instructions, it implements the wheel trajectory elevation extraction method described in any of the above embodiments.

[0121] Those skilled in the art can understand that to implement all or part of the processes in the above embodiment methods, it can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above control methods. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (Hard Disk Drive, abbreviated as HDD) or a solid-state drive (Solid-State Drive, SSD), etc.; the storage medium can also include a combination of the above types of memories.

[0122] So far, the embodiments of the present application have been described in detail. To avoid obscuring the concept of the present application, some details well known in the art have not been described. Those skilled in the art can fully understand how to implement the technical solutions disclosed herein based on the above description.

[0123] Although some specific embodiments of the present application have been described in detail by way of examples, those skilled in the art should understand that the above examples are for illustrative purposes only and not for limiting the scope of the present application.

[0124] The steps in the present application can be adjusted, combined, and deleted according to actual needs.

[0125] Although the present application has been disclosed in detail with reference to the accompanying drawings, it should be understood that these descriptions are merely exemplary and not used to limit the application of the present application. The protection scope of the present application is defined by the appended claims and may include various modifications, adaptations, and equivalent solutions made to the invention without departing from the protection scope and spirit of the present application.

Claims

1. A method for extracting the elevation of a wheel trajectory, characterized in that, The method includes: Step 1: Based on the V-disparity map and vehicle driving information, obtain the road surface fitting equation and the wheel trajectory V map; Step 2: Based on the valid points in the wheel trajectory V map, use a preset reward function to generate a reward map A corresponding to the wheel trajectory V map; wherein, the valid points are points with non-zero disparity values; Step 3: According to a preset path rule, calculate the cumulative reward values of each point in the valid interval in the reward map A to generate a cumulative reward map M, where the valid interval is determined by the valid points in the wheel trajectory V map; Step 4: In the cumulative reward map M, search for the maximum cumulative reward value row by row, and generate a reward path based on the maximum cumulative reward value; Step 5: Calculate the difference between the reward path and the road surface fitting equation to obtain wheel trajectory elevation data.

2. The wheel track elevation extraction method according to claim 1, characterized in that The method further includes:

3. The method for extracting the elevation of the wheel track according to claim 1, wherein Perform an average weighted operation on the wheel trajectory elevation data corresponding to multiple frames of V-disparity maps obtained to generate a final wheel trajectory elevation curve. In the said Step 2, it specifically includes: Step 21: Based on the valid points in the wheel trajectory V map, use the preset reward function to calculate the reward values row by row and column by column, and respectively obtain a row reward value map and a column reward value map; 4. The method for extracting the elevation of the wheel trajectory according to claim 3, characterized in that, Step 22: Use the method of matrix addition to calculate the sum value between the row reward value and the column reward value, and denote it as the reward map A. In the said Step 21, the calculation methods of the row reward value map and the column reward value map are the same. Among them, the calculation process of the row reward value map specifically includes: Step 211: Traverse the valid points in each row of the wheel trajectory V map to determine the horizontal valid width and the median of the horizontal width; Step 212: Calculate the disparity centroid position of the current row according to the disparity values of the valid points in each row of the wheel trajectory V map; Step 213: According to the disparity centroid position and the median of the horizontal width, calculate the row reward basis based on Gaussian operation; 5. The method for extracting the elevation of the wheel track according to claim 3 or 4, characterized in that Step 214: Based on the row reward basis, use the preset reward function to calculate the reward values row by row to obtain the row reward value map. The method further includes: When it is determined that the number of valid points in any row of the wheel trajectory V map is less than or equal to the first quantity threshold, record the reward value of the corresponding row in the row reward value map as 0; and / or 6. The method for extracting the elevation of the wheel trajectory according to claim 3 or 4, characterized in that, When it is determined that the number of valid points in any column of the wheel trajectory V map is less than or equal to the second quantity threshold, record the reward value of the corresponding column in the column reward value map as 0. The method further includes: Statistically count the number of valid points in the wheel trajectory V map column by column to generate a column histogram; Select the weighted evaluation value in the column histogram; 7. The method for extracting the elevation of a wheel trajectory according to any one of claims 1 to 4, characterized in that When it is determined that the weighted evaluation value is less than or equal to the weighted threshold, perform a weighted operation on the reward values of the corresponding rows in the column reward value map according to the number of valid points in any row of the wheel trajectory V map to update the column reward value map. The said Step 3 specifically includes: Step 31: Taking the lower right corner of the reward map A as the starting position, confirm the cumulative reward value of each point in any valid interval in the reward map A in the order from right to left and from bottom to top. The calculation formula of the cumulative reward value is as follows: where (d i , y j ) is the point coordinate of the i-th row and j-th column in the reward map A, and A(d i , y j ) is the reward value corresponding to the point coordinate (d i , y j ), and M(d i+1 , y j ), M(d i , y j+1 ), M(d i+1 , y j+1 ) are the cumulative reward values corresponding to the point coordinates (d i+1 , y j ), (d i , y j+1 ), (d i+1 , y j+1 ) respectively; Step 32: Generate the cumulative reward map M based on the cumulative reward value.

8. A wheel track elevation extraction system, characterized in that, The system includes: A fitting unit configured to obtain a road surface fitting equation and a wheel trajectory V map based on the V disparity map and vehicle driving information; A reward map unit configured to generate a reward map A corresponding to the wheel trajectory V map by using a preset reward function based on the valid points in the wheel trajectory V map; wherein the valid points are points with non-zero disparity values; A cumulative reward map unit configured to calculate the cumulative reward value of each point in the valid interval in the reward map A according to a preset path rule and generate a cumulative reward map M; wherein the valid interval is determined by the valid points in the wheel trajectory V map; A reward path unit configured to search for the maximum cumulative reward value row by row in the cumulative reward map M and generate a reward path based on the maximum cumulative reward value; An elevation extraction unit configured to calculate the difference between the reward path and the road surface fitting equation to obtain wheel trajectory elevation data.

9. The wheel track elevation extraction system according to claim 8, characterized in that, The elevation extraction unit is further configured to: Perform an average weighted operation on the wheel trajectory elevation data corresponding to multiple frames of V disparity maps obtained to generate a final wheel trajectory elevation curve.

10. An electronic device, characterized in that, It includes: A processor; A memory for storing processor-executable instructions; Wherein, the processor is configured to implement the wheel trajectory elevation extraction method described in any one of the above claims 1-7 when executing the executable instructions.

Images