Ground height estimation method, device and storage medium

Abstract

An embodiment of the present invention provides a ground height estimation method, apparatus, and storage medium. The ground height estimation method includes: obtaining an initial estimation of the ground height of the current point cloud frame and initial estimations of the ground height of at least one historical point cloud frame; determining the confidence level of the initial estimation of the ground height of the current point cloud frame and the confidence levels of the initial estimations of the ground height of the at least one historical point cloud frame; and fusing the initial estimation of the ground height of the current point cloud frame and the initial estimations of the ground height of the historical point cloud frames according to the confidence level of the initial estimation of the ground height of the current point cloud frame and the confidence levels of the initial estimations of the ground height of the at least one historical point cloud frame to obtain an estimation of the ground height of the current point cloud frame. The solution of the embodiment of the present invention improves the reliability and robustness of the estimation of the ground height of the current point cloud frame.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of computer technology, and in particular, to a method, apparatus, and storage medium for estimating ground height. Background Art

[0002] A lidar is a sensor that uses laser as a signal source. The pulsed laser emitted by the laser emitter is scattered when it hits target objects such as the ground, trees, roads, bridges, and buildings (e.g., obstacles). Among them, a part of the scattered light waves will be reflected onto the receiver of the lidar to form point cloud data, and the distance from the lidar to the target point can be calculated according to the laser ranging principle. Since the pulsed laser continuously scans the target object, the point cloud data of the target object can be obtained. After imaging processing with this point cloud data, an accurate three-dimensional stereoscopic image can be obtained.

[0003] Generally speaking, lidar can provide accurate distance measurement and three-dimensional geometric information, and is widely used in fields with high requirements for spatial ranging accuracy such as autonomous driving.

[0004] Due to this ranging characteristic of lidar, the height feature of the point cloud data can improve the reliability of obstacle estimation, and further enhance the target detection ability of lidar. Therefore, a reliable ground height estimation scheme is needed. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method, apparatus, and storage medium for estimating ground height to at least partially solve the above problems.

[0006] According to a first aspect of the embodiments of the present invention, a method for estimating ground height is provided, including: obtaining an initial estimate of the ground height of the current point cloud frame and initial estimates of the ground heights of at least one historical point cloud frame; determining the confidence levels of the initial estimate of the ground height of the current point cloud frame and the initial estimates of the ground heights of the at least one historical point cloud frame; and fusing the initial estimate of the ground height of the current point cloud frame and the initial estimates of the ground heights of the historical point cloud frames according to the confidence levels of the initial estimate of the ground height of the current point cloud frame and the initial estimates of the ground heights of the at least one historical point cloud frame to obtain an estimate of the ground height of the current point cloud frame.

[0007] According to a second aspect of the embodiments of the present invention, a ground height estimation device is provided, including: an acquisition module that acquires an initial estimation of the ground height of the current point cloud frame and initial estimations of the ground heights of at least one historical point cloud frame; a determination module that determines the confidence levels of the initial estimation of the ground height of the current point cloud frame and the initial estimations of the ground heights of the at least one historical point cloud frame; and a fusion module that fuses the initial estimation of the ground height of the current point cloud frame and the initial estimations of the ground heights of the historical point cloud frames according to the confidence levels of the initial estimation of the ground height of the current point cloud frame and the initial estimations of the ground heights of the at least one historical point cloud frame to obtain an estimation of the ground height of the current point cloud frame.

[0008] According to a third aspect of the embodiments of the present invention, an electronic device is provided, including: a processor, a memory, a communication interface, and a communication bus, where the processor, the memory, and the communication interface complete communication with each other through the communication bus; the memory is used to store at least one executable instruction, and the executable instruction causes the processor to perform operations corresponding to the method described in the first aspect.

[0009] According to a fourth aspect of the embodiments of the present invention, a computer storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, it implements the method described in the first aspect.

[0010] In the solution of the embodiments of the present invention, the estimation of the ground height of the current point cloud frame fuses the initial estimation of the ground height of the current point cloud frame and the initial estimations of the ground heights of at least one historical point cloud frame. In other words, the initial estimation data of the ground height before fusion includes information of the historical point cloud frames, thus improving the reliability of the estimation of the ground height of the current point cloud frame. In addition, the fusion of the initial estimations of the ground height based on the confidence levels of the initial estimations of the ground height further improves the robustness of the estimation of the ground height of the current point cloud frame. Description of the Drawings

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings.

[0012] Figure 1 It is a flowchart of the steps of a ground height estimation method according to an embodiment of the present invention;

[0013] Figure 2 It is a flowchart of the steps of another ground height estimation method according to the present invention;

[0014] Figure 3 A structural block diagram of a ground height estimation device according to another embodiment of the present invention;

[0015] Figure 4 A structural block diagram of a ground height estimation device according to another embodiment of the present invention; and

[0016] Figure 5 A structural schematic diagram of an electronic device according to another embodiment of the present invention. Detailed implementation manners

[0017] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments in the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art shall fall within the protection scope of the embodiments of the present invention.

[0018] The following further illustrates the specific implementation of the embodiments of the present invention in conjunction with the accompanying drawings of the embodiments of the present invention.

[0019] Figure 1 A step flow chart of a ground height estimation method according to an embodiment of the present invention. The solution of this embodiment can be applied to any suitable electronic device with data processing capabilities, including but not limited to: servers, mobile terminals (such as mobile phones, PADs, etc.) and PC machines, etc. Figure 1 The ground height estimation method includes:

[0020] S110: Obtain an initial estimate of the ground height of the current point cloud frame, and an initial estimate of the ground height of at least one historical point cloud frame.

[0021] It should be understood that the sensors for obtaining such as the current point cloud frame or the historical point cloud frame include but are not limited to lidar. When the sensor is a lidar, the sensor may include one or more lidars installed on the acquisition vehicle. One or more lidars can be installed in any direction. Preferably, in an autonomous driving scenario, they can be installed to mainly emit or collect light in the direction of the ground, so as to be able to collect sufficient ground information. Generally speaking, the lidar emits signals around and collects point cloud data in units of point cloud frames. The point cloud data includes three-dimensional information within the acquisition range.

[0022] It should also be understood that the acquisition of the current point cloud frame in the text is performed after the acquisition of at least one frame of historical point cloud frames. Additionally, there is no limitation on the relationship between the acquisition time of the current point cloud frame in the text and whether the next point cloud frame of the current point cloud frame has been acquired. In other words, the current point cloud frame refers to the point cloud frame for which the ground height estimation is to be determined, and is not necessarily the most recently acquired point cloud frame. Additionally, the height of each point in the point cloud indicates the distance between that point and the lidar.

[0023] It should also be understood that the initial ground height estimation of the point cloud frame in the text can be the initial ground height estimation for single-frame point cloud data or the initial ground height estimation for multi-frame point cloud data. As a preferred embodiment, the lidar continuously acquires point cloud frames to obtain a sequence of point cloud frames. Correspondingly, the initial ground height estimation can be performed for each point cloud frame in the sequence of point cloud frames. For example, the initial ground height estimation is successively performed for each point cloud frame to obtain the initial ground height estimation of the first frame of point cloud, the initial ground height estimation of the second frame of point cloud, the initial ground height estimation of the third frame of point cloud,..., the initial ground height estimation of the Nth frame of point cloud, where N is a positive integer. In one example, the current point cloud frame is a single-frame point cloud, and the historical point cloud frames are at least one single-frame point cloud acquired historically. For example, the Nth frame is the current point cloud frame, and the previous N - 1 frames are the historical point cloud frames.

[0024] It should also be understood that for the initial ground height estimation of the point cloud frame, the initial ground height estimation can be performed for each grid of the point cloud frame. Each grid can correspond to a height estimation value for the initial ground height estimation, and this height estimation value can be the height statistical result of the height estimations of the points within the grid. For example, it can be the median value, average value, highest point value, lowest point value, etc. of the heights of the points. Clustering processing can be performed on the points within each grid, and height statistics can be performed on the points in the main cluster.

[0025] S120: Determine the confidence level of the initial ground height estimation of the current point cloud frame and the confidence level of the initial ground height estimation of the at least one frame of historical point cloud frames.

[0026] It should be understood that the confidence level in the text can also be referred to as the confidence degree. For example, the confidence level can be a confidence value represented by a probability value between 0 and 1, or a confidence value represented by other statistical metrics.

[0027] It should also be understood that the confidence level in the text can be a single confidence value corresponding to the point cloud frame (e.g., the confidence value of the initial ground height estimation of the point cloud frame), or a set of confidence values corresponding to the point cloud frame (e.g., the confidence values corresponding to different parts of the point cloud frame). For example, the confidence value can be determined for each grid of the point cloud frame to obtain a set of confidence values, and each confidence value corresponds to a grid of the point cloud frame.

[0028] S130: Based on the confidence of the initial ground height estimate of the current point cloud frame and the confidence of the initial ground height estimate of at least one frame of historical point cloud frames, fuse the initial ground height estimate of the current point cloud frame and the initial ground height estimate of the historical point cloud frames to obtain the ground height estimate of the current point cloud frame.

[0029] It should be understood that fusion can be performed for the grids corresponding to different point cloud frames, or for a corresponding set of grids. More specifically, the ground height estimate of a specific grid in the current point cloud frame after fusion is determined based on both the initial ground height estimate of the grid corresponding to this specific grid in the historical point cloud frame and the initial ground height estimate of this feature grid.

[0030] It should also be understood that the fusion process can be performed continuously or at a predetermined period. When performing fusion for a point cloud frame after the current point cloud frame (for example, the next point cloud frame), the ground height overestimate estimate as the fusion result of the current point cloud frame (the historical point cloud frame of the next point cloud frame) can be used as the initial ground height estimate of the historical point cloud frame (the historical point cloud frame of the next point cloud frame, in other words, the current point cloud frame) adopted for the ground height estimate fusion of the next point cloud frame, so as to achieve iterative update, which is beneficial to improving the data processing efficiency.

[0031] In the solution of the embodiment of the present invention, the ground height estimate of the current point cloud frame fuses the initial ground height estimate of the current point cloud frame and the initial ground height estimate of at least one frame of historical point cloud frames. In other words, the initial ground height estimate data before fusion includes the information of the historical point cloud frames, so the reliability of the ground height estimate of the current point cloud frame is improved. In addition, fusing the initial ground height estimates based on the confidence of the initial ground height estimates further improves the robustness of the ground height estimate of the current point cloud frame.

[0032] In some other examples, the method for obtaining the initial ground height estimate of the current point cloud frame includes: rasterizing the current point cloud frame; obtaining the initial ground height estimate of the current point cloud frame according to the point cloud height distribution within multiple grids and / or the local plane estimation result near the obstacle within the grid.

[0033] In some other examples, the initial ground height estimate of the historical point cloud frame is obtained by the following method: rasterizing the historical point cloud frame; obtaining the initial ground height estimate of the historical point cloud frame according to the point cloud height distribution within multiple grids and / or the local plane estimation result near the obstacle within the grid.

[0034] Specifically, the local plane estimation result near the obstacle can indicate plane fitting to a target plane based on multiple point clouds within the neighborhood of the obstacle. For an obstacle distributed across multiple grids, the fitting plane corresponding to the obstacle for each grid can be determined, and the height estimation for each of the multiple grids can be judged as a preset height variation range.

[0035] For example, the height of each grid in the point cloud frame can be estimated sequentially. When it is determined whether the current grid is distributed with a target obstacle, if so, it is determined whether the grids adjacent to the current grid are distributed with the target obstacle until all grids distributed with the obstacle are determined, and the height estimations of all grids are determined to be within the preset height variation range.

[0036] When it is determined that the current grid is not distributed with an obstacle, the height estimation of the current grid begins to be determined.

[0037] In another example, rasterization processing can be performed on the current point cloud frame and the historical point cloud frame. The height distribution of the point clouds within each grid of the rasterized current point cloud frame and the height distribution of the point clouds within each grid of the historical point cloud frame can be statistically analyzed respectively. The initial ground height estimations for each grid of the current frame and the initial ground height estimations for each grid of the historical frame can be determined respectively based on the height distribution of the point clouds within each grid of the current frame and the height distribution of the point clouds within each grid of the historical frame.

[0038] It should be understood that the current frame grid refers to the grid of the current point cloud frame. The historical frame point cloud frame grid refers to the grid of the historical point cloud frame. Specifically, based on the height distribution of the point clouds within each grid of the current frame, the degree of uniformity of the height distribution of the point clouds within each grid of the current frame can be determined. Based on the height distribution of the point clouds within each grid of the historical frame, the degree of uniformity of the height distribution of the point clouds within each grid of the historical frame can be determined. The initial ground height estimations for each grid of the current frame can be determined based on the degree of uniformity of the height distribution of the point clouds within each grid of the current frame, and the initial ground height estimations for each grid of the historical frame can be determined based on the degree of uniformity of the height distribution of the point clouds within each grid of the historical frame.

[0039] Since the height distribution of the point clouds reflects the confidence level of each point cloud being a ground point cloud, the initial ground height estimations for each grid of the current frame are determined based on the height distribution of the point clouds within each grid of the current frame, and the initial ground height estimations for each grid of the historical frame are determined based on the height distribution of the point clouds within each grid of the historical frame, which improves the reliability of the initial ground height estimation. In addition, based on the grid ground height estimation, both the efficiency of the ground height estimation is improved and the accuracy of the ground height estimation is ensured.

[0040] Specifically, for the point cloud with a relatively uniform height distribution, the confidence level that it is the ground point cloud is relatively high, and this height range can be used as the ground height. Additionally, according to the characteristics of the lidar device that collects the point cloud, the closer the range is to the lidar, the higher the sensing accuracy, and the higher the confidence level of the point cloud closer to the lidar in the point cloud frame.

[0041] In another example, determining the confidence level of the initial ground height estimate of the current point cloud frame and the confidence level of the initial ground height estimate of the at least one frame of historical point cloud frames includes: determining the confidence level of the initial ground height estimate of each grid according to the height distribution of the initial ground height estimates of multiple grids in the current point cloud frame; determining the confidence level of the initial ground height estimate of each grid according to the height distribution of the initial ground height estimates of multiple grids in the historical point cloud frame; the smoother the height distribution between any grid and its neighboring grids, the higher the confidence level of the initial ground height estimate of this grid.

[0042] In another example, the confidence level of the initial ground height estimate of each current frame grid can be determined according to the height distribution of the initial ground height estimates of each current frame grid, where the smoother the height distribution between a current frame grid and its neighboring grids, the higher the confidence level of the initial ground height estimate of this current frame grid. The confidence level of the initial ground height estimate of each historical frame grid can be determined according to the height distribution of the initial ground height estimates of each historical frame grid, where the smoother the height distribution between a historical frame grid and its neighboring grids, the higher the confidence level of the initial ground height estimate of this historical frame grid.

[0043] Since the confidence level calculation is performed for the grids, it not only improves the efficiency of the confidence level calculation but also ensures the accuracy of the confidence level calculation. In addition, the smoothness of the height distribution between the frame grid and its neighboring grids reflects the reliability of the ground height estimate of the frame grid, in other words, reflects the confidence level of the ground height estimate of the frame grid, and improves the efficiency of the confidence level calculation.

[0044] In some other examples, the method further includes: if the smoothness of the height distribution between any grid and its neighboring grids is lower than the preset range, then determining the interpolation of the confidence level of this grid as the confidence level of the initial ground height estimate of this grid.

[0045] In other words, if the smoothness of the height distribution between a current frame grid and its neighboring grids is lower than a preset range, the interpolation of the confidence level of the current frame grid is determined as the confidence level of the initial ground height estimate of the current frame grid. Or, if the smoothness of the height distribution between a historical frame grid and its neighboring grids is lower than a preset range, the interpolation of the confidence level of the neighboring grids of the historical frame grid is determined as the confidence level of the initial ground height estimate of the historical frame grid.

[0046] When the smoothness of the height distribution between a current frame grid and its neighboring grids is lower than a preset range, the reliability of the confidence levels of both the current frame grid and the neighboring grids is relatively high. Determining the interpolation of the confidence level of the current frame grid as the confidence level of the initial ground height estimate of the current frame grid improves the efficiency of confidence level calculation. Additionally, when the smoothness of the height distribution between a historical frame grid and its neighboring grids is lower than a preset range, the confidence level of the current frame grid is lower than that of its neighboring grids. Therefore, determining the interpolation of the confidence level of the neighboring grids of the historical frame grid as the confidence level of the initial ground height estimate of the historical frame grid improves the accuracy of confidence level calculation.

[0047] In some other examples, determining the confidence level of the initial ground height estimate of the current point cloud frame includes: determining a plurality of acquisition distances respectively corresponding to the initial ground height estimates of a plurality of grids of the current point cloud frame; wherein, the acquisition distance indicates the distance between the point cloud conforming to the initial ground height estimate and the sensor that acquires the current point cloud frame. Determining the confidence level of the initial ground height estimate of the current point cloud frame according to the plurality of acquisition distances. The smaller an acquisition distance is, the greater the confidence level of the initial ground height estimate corresponding to that acquisition distance is. The confidence level of the initial ground height estimate of at least one historical point cloud frame is determined in the following manner: determining a plurality of acquisition distances respectively corresponding to the initial ground height estimates of a plurality of grids of the historical point cloud frame; wherein, the acquisition distance indicates the distance between the point cloud conforming to the initial ground height estimate and the sensor that acquires the historical point cloud frame; determining the confidence level of the initial ground height estimate of the historical point cloud frame according to the plurality of acquisition distances; wherein, the smaller an acquisition distance is, the greater the confidence level of the initial ground height estimate corresponding to that acquisition distance is.

[0048] Since the smaller the distance between the point cloud of the initial ground height estimate and the lidar that acquires the point cloud frame is, the greater the confidence level is, the confidence levels corresponding to the current point cloud frame and the historical point cloud frame are determined according to a plurality of current acquisition distances and a plurality of historical acquisition distances, which improves the accuracy of confidence level calculation.

[0049] In some other examples, according to the confidence in the initial estimate of the ground height of the current point cloud frame and the confidence in the initial estimate of the ground height of the at least one historical point cloud frame, the initial estimate of the ground height of the current point cloud frame and the initial estimate of the ground height of the historical point cloud frame are fused to obtain the estimate of the ground height of the current point cloud frame, including: determining the grids of the current point cloud frame and the grids of the historical point cloud frame at corresponding positions; if the difference between the confidence in the grid of the current point cloud frame and the confidence in the grid of the historical point cloud frame exceeds a preset range, then determining the initial estimate of the ground height of the grid with the larger confidence among the grid of the current point cloud frame and the grid of the historical point cloud frame as the estimate of the ground height of the current point cloud frame.

[0050] When the difference between the confidence in the grid of the current frame and the confidence in the grid of the historical frame exceeds the preset range, the smoothness of the change of the difference over time is relatively large. Thus, the grid of the point cloud frame with relatively small confidence reflects a lower confidence level. Therefore, determining the initial estimate of the ground height of the grid with the larger confidence among the grid of the current frame and the grid of the historical frame as the estimate of the ground height of the current point cloud frame improves the fusion efficiency of the initial estimate of the ground height.

[0051] Specifically, when fusing the initial estimates of the ground height of each point cloud frame, for each grid at the corresponding position, the confidence distribution of each grid can be determined. According to the position of the confidence in the grid of the current point cloud frame at this corresponding position in the confidence distribution of each grid, the proportion of the initial estimate of the ground height of the grid of the current point cloud frame at this corresponding position in the fused estimate of the ground height is determined. For example, when the position in the confidence distribution of each grid is less than the proportional value corresponding to the first preset threshold (for example, the confidence in this grid is lower than 80% of all grids), the initial estimate of the ground height can be ignored (not adopted) in the fused estimate of the ground height. For example, when the position in the confidence distribution of each grid is greater than the proportional value corresponding to the second preset threshold (for example, the confidence in this grid is higher than 70% of all grids), for this grid, the initial estimate of the ground height can be determined as the fused estimate of the ground height. For example, for grids at other positions, at least one of smoothing processing, interpolation processing, mean processing, and weighted processing can be performed on the initial estimates of the ground height corresponding to each grid to obtain the proportion of the initial estimate of the ground height of this grid in the fused estimate of the ground height. The above method can be executed for each grid at the corresponding position.

[0052] In some other examples, the method further includes: if the difference between the confidence of the grid of the current point cloud frame and the confidence of the grid of the historical point cloud frame does not exceed a preset range, determining the confidence interpolation between the confidence of the grid of the current point cloud frame and the confidence of the grid of the historical point cloud frame as the ground height estimate of the current point cloud frame.

[0053] In other words, in order to achieve the confidence according to the respective initial estimates of the ground height, the initial estimate of the ground height of the current point cloud frame and the initial estimate of the ground height of the historical point cloud frame are fused to obtain the ground height estimate of the current point cloud frame. If the difference between the confidence of the grid of the current frame and the confidence of the grid of the historical frame does not exceed a preset range, the confidence interpolation between the confidence of the grid of the current frame and the confidence of the grid of the historical frame can also be determined as the ground height estimate of the current point cloud frame.

[0054] Since the difference between the confidence of the grid of the current frame and the confidence of the grid of the historical frame at the corresponding position reflects the change characteristics of a specific position over time, the smoother the change of the difference over time, the higher the confidence of both the grid of the current frame and the grid of the historical frame. When the difference between the confidence of the grid of the current frame and the confidence of the grid of the historical frame exceeds the preset range, the smoothness of the change of the above difference over time is relatively small. Therefore, the initial estimate of the ground height of the grid with the larger confidence in the grid of the current frame and the grid of the historical frame is determined as the ground height estimate of the current point cloud frame, improving the fusion efficiency between the current frame and the historical frame.

[0055] In some other examples, the method further includes: determining a first ground height distribution of the initial estimates of the ground height of the grids of a plurality of current point cloud frames passed by the target trajectory, and a second ground height distribution of the initial estimates of the ground height of the grids of a plurality of historical point cloud frames passed by the target trajectory; performing jitter correction on at least one of the initial estimates of the ground height of the grids of a plurality of current point cloud frames and a plurality of historical point cloud frames, so that the first ground height distribution and the second ground height distribution are consistent.

[0056] In other words, in order to achieve the confidence according to the respective initial estimates of the ground height, the initial estimate of the ground height of the current point cloud frame and the initial estimate of the ground height of the historical point cloud frame are fused to obtain the ground height estimate of the current point cloud frame. It is also possible to determine a first ground height distribution of the initial estimates of the ground height of the grids of a plurality of current frames passed by the target trajectory, and a second ground height distribution of the initial estimates of the ground height of the grids of a plurality of historical frames passed by the target trajectory; performing jitter correction on at least one of the initial estimates of the ground height of the grids of a plurality of current frames and a plurality of historical frames, so that the first ground height distribution and the second ground height distribution are consistent.

[0057] Since the first ground height distribution and the second ground height distribution of the target trajectory effectively reflect the motion state of the lidar when collecting each point cloud frame, when the first ground height distribution and the second ground height distribution are inconsistent, the motion state of the lidar appears in a normal point cloud data acquisition state (e.g., a jitter state). Jitter correction is performed on at least one of the initial ground height estimates of several current frame grids and several historical frame grids to obtain consistent first and second ground height distributions, which is beneficial to improving the subsequent fusion accuracy.

[0058] Figure 2 FIG. is a flowchart of the steps of a ground height estimation method according to another embodiment of the present invention. The solution of this embodiment can be applied to any suitable electronic device with data processing capabilities, including but not limited to: servers, mobile terminals (such as mobile phones, PADs, etc.) and PCs, etc. Figure 2 The ground height estimation method includes:

[0059] S210: Perform rasterization processing on the current point cloud frame and the historical point cloud frame.

[0060] S220: Respectively for the rasterized current point cloud frame and historical point cloud frame, count the point cloud height distribution within each current frame grid and the point cloud height distribution within each historical frame grid.

[0061] S230: Respectively for each point cloud height distribution, determine the initial ground height estimate of each current frame grid and the initial ground height estimate of each historical frame grid. Specifically, the current frame grid refers to the grid of the current point cloud frame. The historical frame point cloud frame grid refers to the historical point cloud frame grid. Specifically, based on the point cloud height distribution within each current frame grid, the uniformity of the point cloud height distribution of each current frame grid can be determined. Based on the point cloud height distribution within each historical frame grid, the uniformity of the point cloud height distribution of each historical frame grid can be determined. Based on the uniformity of the point cloud height distribution of each current frame grid, the initial ground height estimate of each current frame grid can be determined. Based on the uniformity of the point cloud height distribution of each historical frame grid, the initial ground height estimate of each historical frame grid can be determined.

[0062] S240: According to the height distribution of the initial ground height estimate of each current frame grid, determine the confidence level of the initial ground height estimate of each current frame grid. The smoother the height distribution between a current frame grid and its neighboring grids, the higher the confidence level of the initial ground height estimate of this current frame grid.

[0063] S250: Determine the confidence of the initial ground height estimate for each historical frame grid according to the height distribution of the initial ground height estimate of each historical frame grid. The smoother the height distribution between a historical frame grid and its neighboring grids, the higher the confidence of the initial ground height estimate of this historical frame grid. When the smoothness of the height distribution between a current frame grid and its neighboring grids is lower than the preset range, the reliability of the confidence of both the current frame grid and the neighboring grids is relatively high. Interpolate the confidence of the current frame grid to determine the confidence of the initial ground height estimate of the current frame grid, which improves the efficiency of confidence calculation. Additionally, when the smoothness of the height distribution between a historical frame grid and its neighboring grids is lower than the preset range, and the confidence of the current frame grid is lower than that of its neighboring grids, the confidence of the neighboring grid of this historical frame grid is interpolated to determine the confidence of the initial ground height estimate of this historical frame grid, which improves the accuracy of confidence calculation.

[0064] S260: Determine the current frame grid and the historical frame grid at the corresponding positions in the current point cloud frame and the historical point cloud frames.

[0065] S270: If the difference between the confidence of the current frame grid and the confidence of the historical frame grid exceeds the preset range, determine the initial ground height estimate of the grid with the higher confidence as the ground height estimate of the current point cloud frame. Specifically, when fusing the initial ground height estimates of each point cloud frame, for each grid with a corresponding position, the confidence distribution of each grid can be determined. According to the position of the confidence of the grid at this corresponding position in the current point cloud frame in the confidence distribution of each grid, determine the proportion of the initial ground height estimate of the grid at this corresponding position in the current point cloud frame in the fused ground height estimate. For example, when the position in the confidence distribution of each grid is less than the proportional value corresponding to the first preset threshold (e.g., the confidence of this grid is lower than 80% of all grids), the initial ground height estimate can be ignored (not used) in the fused ground height estimate. For example, when the position in the confidence distribution of each grid is greater than the proportional value corresponding to the second preset threshold (e.g., the confidence of this grid is higher than 70% of all grids), the initial ground height estimate can be determined as the fused ground height estimate for this grid.

[0066] S280: If the difference between the confidence of the current frame grid and the confidence of the historical frame grid does not exceed the preset range, interpolate the confidence to determine the ground height estimate of the current point cloud frame. Specifically, for grids at other positions, at least one of smoothing processing, interpolation processing, mean processing, and weighted processing can be performed on the initial ground height estimates corresponding to each grid to obtain the proportion of the initial ground height estimate of this grid in the fused ground height estimate. The above method can be executed for each grid with a corresponding position.

[0067] Figure 3 It is a structural block diagram of a ground height estimation device according to another embodiment of the present invention. The solution of this embodiment can be applied to any suitable electronic device with data processing capabilities, including but not limited to: servers, mobile terminals (such as mobile phones, PADs, etc.), and PCs, etc. Figure 3 The ground height estimation device includes:

[0068] An acquisition module 310, which acquires an initial estimate of the ground height of the current point cloud frame and an initial estimate of the ground height of at least one historical point cloud frame.

[0069] A determination module 320, which determines the confidence level of the initial estimate of the ground height of the current point cloud frame and the confidence level of the initial estimate of the ground height of the at least one historical point cloud frame.

[0070] A fusion module 330, which fuses the initial estimate of the ground height of the current point cloud frame and the initial estimate of the ground height of the historical point cloud frame according to the confidence level of the initial estimate of the ground height of the current point cloud frame and the confidence level of the initial estimate of the ground height of the at least one historical point cloud frame, to obtain an estimate of the ground height of the current point cloud frame.

[0071] In the solution of the embodiment of the present invention, the estimate of the ground height of the current point cloud frame fuses the initial estimate of the ground height of the current point cloud frame and the initial estimate of the ground height of at least one historical point cloud frame. In other words, the initial estimate data of the ground height before fusion includes the information of the historical point cloud frame, so the reliability of the estimate of the ground height of the current point cloud frame is improved. In addition, the fusion of the initial estimates of the ground height is performed based on the confidence level of the initial estimate of the ground height, which further improves the robustness of the estimate of the ground height of the current point cloud frame.

[0072] In some other examples, the acquisition module may specifically be used for: rasterizing the current point cloud frame; obtaining the initial estimate of the ground height of the current point cloud frame according to the height distribution of the point cloud in multiple grids and / or the local plane estimation result near the obstacle in the grid. The acquisition module may also specifically be used for rasterizing the historical point cloud frame; obtaining the initial estimate of the ground height of the historical point cloud frame according to the height distribution of the point cloud in multiple grids and / or the local plane estimation result near the obstacle in the grid.

[0073] In some other examples, the determination module is specifically configured to: determine the confidence of the initial ground height estimate for each grid according to the height distribution of the initial ground height estimates of multiple grids in the current point cloud frame; determine the confidence of the initial ground height estimate for each grid according to the height distribution of the initial ground height estimates of multiple grids in the historical point cloud frame; the smoother the height distribution between any grid and its neighboring grids, the higher the confidence of the initial ground height estimate for that grid.

[0074] In some other examples, the determination module is further configured to: if the smoothness degree of the height distribution between any grid and its neighboring grids is lower than a preset range, then determine the interpolation of the confidence of that grid as the confidence of the initial ground height estimate for that grid.

[0075] In some other examples, the determination module is specifically configured to: determine the multiple acquisition distances corresponding to the initial ground height estimates of multiple grids in the current point cloud frame; wherein, the acquisition distance indicates the distance between the point cloud conforming to the initial ground height estimate and the sensor that acquires the current point cloud frame; determine the confidence of the initial ground height estimate of the current point cloud frame according to the multiple acquisition distances; wherein, the smaller an acquisition distance is, the greater the confidence of the initial ground height estimate corresponding to that acquisition distance is; determine the confidence of the initial ground height estimate of the at least one frame of historical point cloud frame by the following method: determine the multiple acquisition distances corresponding to the initial ground height estimates of multiple grids in the historical point cloud frame; wherein, the acquisition distance indicates the distance between the point cloud conforming to the initial ground height estimate and the sensor that acquires the historical point cloud frame; determine the confidence of the initial ground height estimate of the historical point cloud frame according to the multiple acquisition distances; wherein, the smaller an acquisition distance is, the greater the confidence of the initial ground height estimate corresponding to that acquisition distance is.

[0076] In some other examples, the fusion module is specifically configured to: determine the grids of the current point cloud frame and the grids of the historical point cloud frame corresponding to the same position; if the difference between the confidence of the grid of the current point cloud frame and the confidence of the grid of the historical point cloud frame exceeds a preset range, then determine the initial ground height estimate of the grid with the greater confidence among the grid of the current point cloud frame and the grid of the historical point cloud frame as the ground height estimate of the current point cloud frame.

[0077] In some other examples, the fusion module is further configured to: if the difference between the confidence of the grid of the current point cloud frame and the confidence of the grid of the historical point cloud frame does not exceed a preset range, then determine the confidence interpolation between the confidence of the grid of the current point cloud frame and the confidence of the grid of the historical point cloud frame as the ground height estimate of the current point cloud frame.

[0078] In some other examples, the fusion module is further configured to: determine a first ground height distribution of an initial estimate of the ground height of the grid passed by the target trajectory through a plurality of current point cloud frames, and a second ground height distribution of an initial estimate of the ground height of the grid passed by the target trajectory through a plurality of historical point cloud frames; perform jitter correction on at least one of the initial estimates of the ground height of the grids of the plurality of current point cloud frames and the grids of the plurality of historical point cloud frames, so that the first ground height distribution and the second ground height distribution are consistent.

[0079] The device of this embodiment is used to implement the corresponding methods in the foregoing multiple method embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be elaborated here. In addition, the implementation of the functions of each module in the device of this embodiment can be referred to the description of the corresponding parts in the foregoing method embodiments, which will not be elaborated here either.

[0080] Figure 4 It is a structural block diagram of a ground height estimation device according to another embodiment of the present invention. Figure 4 The ground height estimation device includes a ground height initial estimation module, a confidence estimation module, a jitter elimination module using a vibration filtering algorithm module, a fusion module using a confidence fusion algorithm, and a height estimation output module. Among them, the ground height initial estimation module can correspond to Figure 3 the acquisition module, the confidence estimation module can correspond to Figure 3 the determination module, and the jitter elimination module, the fusion module, and the height estimation output module can correspond to Figure 3 the fusion module.

[0081] As Figure 4 shown, first, the point cloud frame sequence continuously collected by the lidar (K + 1 frames of point clouds including current point cloud frames and historical point cloud frames) is input into the ground height initial estimation module. Correspondingly, the initial estimation of the ground height can be performed on each point cloud frame in the point cloud frame sequence respectively. For example, the initial estimation of the ground height is successively performed on each point cloud frame to obtain the initial estimation of the ground height of the first point cloud frame, the initial estimation of the ground height of the second point cloud frame, the initial estimation of the ground height of the third point cloud frame,..., the initial estimation of the ground height of the (K + 1)th point cloud frame, where K is a positive integer.

[0082] Then, the ground height initial estimation module inputs each initial estimation of the ground height into the confidence estimation module. In the confidence estimation module, the grid confidence of each frame in the above K + 1 frames of point clouds can be determined, and the grid confidence of the initial estimation of the ground height of each point cloud frame is obtained.

[0083] Then, in the jitter elimination module, the jitter correction of the initial ground height estimate of each point cloud frame can be performed on the lidar to obtain the corrected initial ground height estimate and the confidence of the corrected initial ground height estimate. In the determination module, the corrected initial ground height estimate can be fused using the confidence of the corrected initial ground height estimate to obtain the ground height estimate of the current point cloud frame. The ground height estimate of the current point cloud frame can be input through the height estimate output module.

[0084] The device of this embodiment is used to implement the corresponding methods in the foregoing multiple method embodiments and has the beneficial effects of the corresponding method embodiments, which will not be elaborated here. In addition, the functional implementation of each module in the device of this embodiment can refer to the description of the corresponding part in the foregoing method embodiments, which will not be elaborated here either.

[0085] Refer to Figure 5 , which shows a schematic structural diagram of an electronic device according to another embodiment of the present invention. The specific implementation of the electronic device in the specific embodiments of the present invention is not limited.

[0086] As Figure 5 shown, the electronic device may include: a processor 502, a communications interface 504, a memory 506, and a communication bus 508.

[0087] Among them: the processor 502, the communications interface 504, and the memory 506 communicate with each other through the communication bus 508.

[0088] The communications interface 504 is used to communicate with other electronic devices or servers.

[0089] The processor 502 is used to execute the program 510, and specifically can execute the relevant steps in the foregoing method embodiments.

[0090] Specifically, the program 510 may include program code, and the program code includes computer operation instructions.

[0091] The processor 502 may be a central processing unit CPU, or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention. One or more processors included in the intelligent device may be of the same type of processor, such as one or more CPUs; or may be of different types of processors, such as one or more CPUs and one or more ASICs.

[0092] A memory 506 for storing a program 510. The memory 506 may include high-speed RAM memory and may also include non-volatile memory, such as at least one disk memory.

[0093] Specifically, the program 510 may be used to cause the processor 502 to perform the following operations: obtain an initial estimate of the ground height of the current point cloud frame and an initial estimate of the ground height of at least one historical point cloud frame; determine the confidence of the initial estimate of the ground height of the current point cloud frame and the confidence of the initial estimate of the ground height of the at least one historical point cloud frame; and fuse the initial estimate of the ground height of the current point cloud frame and the initial estimate of the ground height of the historical point cloud frame according to the confidence of the initial estimate of the ground height of the current point cloud frame and the confidence of the initial estimate of the ground height of the at least one historical point cloud frame to obtain an estimate of the ground height of the current point cloud frame.

[0094] In addition, for the specific implementation of each step in the program 510, reference may be made to the corresponding steps and descriptions in the corresponding units in the foregoing method embodiments, which will not be elaborated herein. Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the devices and modules described above may refer to the corresponding process descriptions in the foregoing method embodiments, which will not be elaborated herein.

[0095] It should be noted that, according to the needs of implementation, the various components / steps described in the embodiments of the present invention may be split into more components / steps, or two or more components / steps or partial operations of components / steps may be combined into new components / steps to achieve the purpose of the embodiments of the present invention.

[0096] The method according to the embodiments of the present invention described above may be implemented in hardware, firmware, or may be implemented as software or computer code stored in a recording medium (such as a CD ROM, RAM, floppy disk, hard disk, or magneto-optical disk), or may be implemented as computer code originally stored in a remote recording medium or a non-transitory machine-readable medium and downloaded through a network and to be stored in a local recording medium, so that the method described herein may be stored on such a software process on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware (such as an ASIC or an FPGA). It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component (such as RAM, ROM, flash memory, etc.) that can store or receive software or computer code, and when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the method described herein is implemented. In addition, when a general-purpose computer accesses the code for implementing the method shown herein, the execution of the code converts the general-purpose computer into a dedicated computer for executing the method shown herein.

[0097] Those of ordinary skill in the art can realize that the units and method steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the embodiments of the present invention.

[0098] The above embodiments are only used to illustrate the embodiments of the present invention, rather than to limit the embodiments of the present invention. Those of ordinary skill in the relevant technical field can also make various changes and modifications without departing from the spirit and scope of the embodiments of the present invention. Therefore, all equivalent technical solutions also belong to the scope of the embodiments of the present invention. The patent protection scope of the embodiments of the present invention shall be defined by the claims.

Claims

1. A method for estimating ground height, comprising: obtaining an initial estimate of the ground height of the current point cloud frame, and initial estimates of the ground height of at least one historical point cloud frame; determining the confidence level of the initial estimate of the ground height of the current point cloud frame, and the confidence level of the initial estimates of the ground height of the at least one historical point cloud frame; fusing the initial estimate of the ground height of the current point cloud frame and the initial estimates of the ground height of the historical point cloud frames according to the confidence level of the initial estimate of the ground height of the current point cloud frame and the confidence levels of the initial estimates of the ground height of the at least one historical point cloud frame, to obtain the estimated ground height of the current point cloud frame, wherein the fusion is performed for corresponding grids of different point cloud frames, and the estimated ground height of a specific grid of the current point cloud frame after fusion is determined based on both the initial estimate of the ground height of the grid corresponding to this specific grid in the historical point cloud frame and the initial estimate of the ground height of this specific grid.

2. The method according to claim 1, wherein, the manner of obtaining the initial estimate of the ground height of the current point cloud frame includes: performing rasterization processing on the current point cloud frame; obtaining the initial estimate of the ground height of the current point cloud frame according to the height distribution of point clouds within multiple grids, and / or the local plane estimation result near obstacles within the grid; wherein the initial estimate of the ground height of the historical point cloud frame is obtained by the following method: performing rasterization processing on the historical point cloud frame; obtaining the initial estimate of the ground height of the historical point cloud frame according to the height distribution of point clouds within multiple grids, and / or the local plane estimation result near obstacles within the grid.

3. The method according to claim 2, wherein, the determination of the confidence level of the initial estimate of the ground height of the current point cloud frame, and the confidence levels of the initial estimates of the ground height of the at least one historical point cloud frame includes: determining the confidence level of the initial estimate of the ground height of each grid according to the height distribution of the initial estimates of the ground height of multiple grids of the current point cloud frame; determining the confidence level of the initial estimate of the ground height of each grid according to the height distribution of the initial estimates of the ground height of multiple grids of the historical point cloud frame; the smoother the height distribution between any grid and its neighboring grids, the higher the confidence level of the initial estimate of the ground height of this grid.

4. The method according to claim 3, wherein, the method further includes: if the smoothness degree of the height distribution between any grid and its neighboring grids is lower than a preset range, then interpolating the confidence level of this grid to determine the confidence level of the initial estimate of the ground height of this grid.

5. The method according to claim 2, wherein, the determination of the confidence level of the initial estimate of the ground height of the current point cloud frame includes: determining multiple acquisition distances respectively corresponding to the initial estimates of the ground height of multiple grids of the current point cloud frame; wherein the acquisition distance indicates the distance between the point cloud conforming to the initial estimate of the ground height and the sensor that acquires the current point cloud frame; Determine the initial confidence of the ground height of the current point cloud frame according to the multiple acquisition distances; wherein, the smaller an acquisition distance is, the greater the confidence of the initial estimate of the ground height corresponding to this acquisition distance is; Determine the confidence of the initial estimate of the ground height of the at least one historical point cloud frame in the following manner: Determine the multiple acquisition distances respectively corresponding to the initial estimates of the ground heights of the multiple grids of the historical point cloud frame; wherein, the acquisition distance indicates the distance between the point cloud conforming to the initial estimate of the ground height and the sensor that acquired the historical point cloud frame; Determine the initial confidence of the ground height of the historical point cloud frame according to the multiple acquisition distances; wherein, the smaller an acquisition distance is, the greater the confidence of the initial estimate of the ground height corresponding to this acquisition distance is.

6. The method according to claim 3, wherein, The step of fusing the initial estimate of the ground height of the current point cloud frame and the initial estimate of the ground height of the historical point cloud frame according to the confidence of the initial estimate of the ground height of the current point cloud frame and the confidence of the initial estimate of the ground height of the at least one historical point cloud frame to obtain the estimated ground height of the current point cloud frame includes: Determine the grids of the current point cloud frame and the grids of the historical point cloud frame belonging to the corresponding positions; If the difference between the confidence of the grid of the current point cloud frame and the confidence of the grid of the historical point cloud frame exceeds a preset range, then determine the initial estimate of the ground height of the grid with the greater confidence among the grid of the current point cloud frame and the grid of the historical point cloud frame as the estimated ground height of the current point cloud frame.

7. The method according to claim 6, wherein, The method further includes: If the difference between the confidence of the grid of the current point cloud frame and the confidence of the grid of the historical point cloud frame does not exceed a preset range, then determine the confidence interpolation between the confidence of the grid of the current point cloud frame and the confidence of the grid of the historical point cloud frame as the estimated ground height of the current point cloud frame.

8. The method according to claim 6, wherein, The method further includes: Determine the first ground height distribution of the initial estimates of the ground heights of the grids of several current point cloud frames passed by the target trajectory, and the second ground height distribution of the initial estimates of the ground heights of the grids of several historical point cloud frames passed by the target trajectory; Perform jitter correction on at least one of the initial estimates of the ground heights of the grids of several current point cloud frames and the grids of several historical point cloud frames so that the first ground height distribution and the second ground height distribution are consistent.

9. A ground height estimation device, comprising: An acquisition module, which acquires the initial estimate of the ground height of the current point cloud frame and the initial estimate of the ground height of at least one historical point cloud frame; A determination module, which determines the confidence of the initial estimate of the ground height of the current point cloud frame and the confidence of the initial estimate of the ground height of the at least one historical point cloud frame; The fusion module fuses the initial ground height estimate of the current point cloud frame and the initial ground height estimates of the at least one historical point cloud frame according to the confidence of the initial ground height estimate of the current point cloud frame and the confidence of the initial ground height estimates of the at least one historical point cloud frame, to obtain the ground height estimate of the current point cloud frame. Wherein, the fusion is performed for the grids corresponding to different point cloud frames, and the ground height estimate of a specific grid of the current point cloud frame after fusion is determined based on both the initial ground height estimate of the grid corresponding to this specific grid of the historical point cloud frame and the initial ground height estimate of this specific grid.

10. A computer storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the method according to any one of claims 1-8.

Images