Roughness detection method and detection system for unhardened road surface, electronic device, and storage medium

By installing sensors on mobile vehicles to collect pavement point cloud data and performing automated analysis, the problem of low rugged detection efficiency of non-hardened pavement is solved, and accurate control of real-time road conditions and improved transportation efficiency are achieved.

CN117670822BActive Publication Date: 2024-11-01北京路凯智行科技有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311655439.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-12-05
Publication Date
2024-11-01
Estimated Expiration
2043-12-05

AI Technical Summary

Technical Problem

In the prior art, the detection of ruggedness of non-hardened road surfaces relies on manual patrols, with slow response speed and unquantifiable judgment basis, resulting in low efficiency, lack of unified scheduling, and high cost.

Method used

The sensor installed on the mobile vehicle collects pavement point cloud data, obtains vehicle position data through the positioning unit, performs timing matching and smoothness analysis, extracts obstacle point sets, filters dynamic targets, calculates pavement ruggedness index, and realizes automated detection. .

Benefits of technology

It realizes accurate control of real-time road conditions, and quantitative data comparison reduces manpower, energy and equipment losses, improves transportation efficiency, and reduces road maintenance costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117670822B_ABST
    Figure CN117670822B_ABST
Patent Text Reader

Abstract

The present disclosure relates to the technical field of data processing, and discloses a method for detecting the roughness of a non-hardened road surface, including: S1: collecting vehicle road surface point cloud data through a sensor on a vehicle, and obtaining vehicle posture data through a positioning unit; S2: performing time-series matching on the collected single-frame point cloud data and vehicle posture data; S3: extracting target road surface point cloud data based on the collected full-frame point cloud data of the road surface; S4: determining the position of the target road surface in the map based on the vehicle posture data; S5: performing smoothness analysis on the point cloud on a column in the extracted single-frame point cloud data, and performing smoothness analysis on the point clouds on other columns column by column to obtain an obstacle point set; S6: filtering dynamic targets on the target road surface based on the obtained obstacle point set; S7: obtaining the roughness of the target road surface based on the filtered result and the position of the target road surface. The method can accurately control the real-time road conditions of the road surface, and the data can be quantitatively compared.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular, to a method for detecting the roughness of an unhardened road surface and a system for detecting the roughness of an unhardened road surface, and also to an electronic device and a storage medium. Background Art

[0002] The road conditions at the open-pit mining site are relatively poor. Since the transportation operation surface will change with the progress of the ore collection and stripping, the transportation roads built in the mining area are basically dirt roads paved with local materials, and rarely professionally hardened. In addition, the frequent back-and-forth driving of heavy-loaded transport vehicles weighing tens or even hundreds of tons causes great damage to the road surface. Especially after watering operations or rainy and snowy weather, the road surface softens significantly, and wet mud can cause wheels to slip or accumulate obstacles such as soil piles and pits. At the same time, the rugged road conditions increase the bumps of heavy-loaded vehicles, causing the transported stones and soil to fall from the bucket, further deteriorating the road flatness and forming a vicious cycle.

[0003] Therefore, the mining site needs to mobilize engineering equipment such as graders, rollers, and bulldozers to repair the roads to ensure the smooth travel of transport vehicles to ensure operational safety and efficiency. At present, road repair operations in various mining areas are manually dispatched and assigned, and on-site personnel are required to regularly inspect the road surface and judge whether to call the road repair team based on experience. The traditional manual detection method has a slow real-time response speed, the judgment basis cannot be quantified, and it relies on the experience and attitude of the specific executors. There is a lack of unified scheduling, so the efficiency is not high. Summary of the invention

[0004] The purpose of the present disclosure is to solve at least one aspect of the above-mentioned problems and defects in the prior art.

[0005] According to one aspect of the present disclosure, a method for detecting roughness of an unhardened road surface is provided, comprising:

[0006] S1: collecting a frame of point cloud data with a frame time stamp of a road surface within a preset range in front of the vehicle through a sensor installed on the mobile vehicle, wherein the point cloud data includes a plurality of vertical columns of point clouds, each vertical column of point clouds includes a plurality of point clouds, and obtaining vehicle posture data of the mobile vehicle through a positioning unit;

[0007] S2: performing time sequence matching on the collected single-frame point cloud data and the vehicle posture data;

[0008] S3: extracting point cloud data of a target road surface to be detected based on the full-frame point cloud data of the road surface collected by the sensor;

[0009] S4: Determine the position of the target road surface to be detected in the map based on the vehicle posture data;

[0010] S5: performing a smoothness analysis on a point cloud in one column of the extracted single-frame point cloud data, and performing a smoothness analysis on the point clouds in other columns of the extracted single-frame point cloud data column by column to obtain an obstacle point set;

[0011] S6: filtering the dynamic targets on the target road surface to be detected based on the obtained obstacle point set;

[0012] S7: deriving the roughness of the target road surface to be detected based on the filtered result and the position of the target road surface to be detected in the map.

[0013] According to an exemplary embodiment of the present disclosure, the sensor collects data on the road surface in a surface scanning manner.

[0014] According to an exemplary embodiment of the present disclosure, extracting point cloud data of a target road surface to be detected based on the full-frame point cloud data of the road surface collected by the sensor includes: performing intra-region screening on the full-frame point cloud data through a map filter to shield background information outside the drivable road surface area in the current frame data, and only retaining the point cloud data of the target road surface to be detected.

[0015] According to an exemplary embodiment of the present disclosure, in step S5, performing smoothness analysis on a point cloud in a vertical column of the extracted single-frame point cloud data includes the following steps:

[0016] S51: deleting the zero point on the column in the single-frame point cloud data, re-recording the remaining points as P1, P2, ..., Pk, if k < 2, it is considered that there are too few valid points on the column, and turning to the next column;

[0017] S52: Reset the obstacle point set, clear the temporary point set, initialize the accumulated height difference ΔT=0, and set the height difference threshold T and the gradient change threshold α;

[0018] S53: performing adjacent point fluctuation calculations on two adjacent points in sequence from one end to the other end along the vertical column, and obtaining an obstacle point set on the vertical column based on the results of the adjacent point fluctuation calculations.

[0019] According to an exemplary embodiment of the present disclosure, performing adjacent point fluctuation calculations on two adjacent points along the column from one end to the other end in sequence, and obtaining an obstacle point set on the column based on the result of the adjacent point fluctuation calculations includes the following steps:

[0020] S531: The point at the front of the two adjacent points is used as the comparison point, and the point at the back is used as the compared point. The gradient value between the two adjacent points is calculated by the following equation:

[0021]

[0022] Among them, x 1 ,y 1 , z 1 To compare the three-dimensional space coordinates of adjacent points;

[0023] x 2 ,y 2 , z 2 is the three-dimensional space coordinate of the adjacent point to be compared;

[0024] S532: If Then ΔT=|z 1 -z 2 |, put the current matched adjacent point into the temporary point set, and use the current matched adjacent point as the next round of matched adjacent point, and the next point adjacent to the current matched adjacent point as the next round of matched adjacent point, and recalculate the gradient value of the next round if Then for the next round of ΔT next Accumulate it with the ΔT of this round to get ΔT total , and put the next round of compared points into the temporary point set, if Then directly convert the accumulated ΔT total Compare with the height threshold T, if ΔT total <T, then for ΔT total and the temporary point set is cleared, if ΔT total >T, then all points in the temporary point set are transferred to the obstacle point set, and ΔT total and the temporary point set are cleared;

[0025] if Then the current point to be compared is used as the next point to be compared, and the next point adjacent to the current point to be compared is used as the next point to be compared, and the gradient value of the next round is recalculated.

[0026] According to an exemplary embodiment of the present disclosure, filtering a dynamic target on a target road surface to be detected based on the obtained obstacle point set includes:

[0027] S61: performing spatial distance clustering on all points in the obstacle point set, and clustering the obstacle point cloud into point cloud clusters;

[0028] S62: Calculate the minimum spatial bounding box of each point cloud cluster;

[0029] S63: Determine a dynamic object filtering parameter threshold according to each instance individual in the work site, and generate a dynamic filter rule;

[0030] S64: Based on the dynamic filter rule, clear the data corresponding to the point cloud cluster in the obstacle point set and the point cloud data of the target road surface to be detected.

[0031] According to an exemplary embodiment of the present disclosure, the road surface roughness index is determined by the following equation based on the obstacle point set that has been cleared:

[0032] Index=sizeB / (sizeG-sizeB),

[0033] Among them, Index is the roughness index of the target road surface to be detected;

[0034] sizeB is the final number of points in the obstacle point set after the removal process;

[0035] sizeG is the total number of points in the point cloud data of the target road surface to be detected after the cleaning process.

[0036] According to an exemplary embodiment of the present disclosure, in step S4, determining the position of the target road surface to be detected in the map based on the vehicle posture data includes: marking the current frame ground coordinate POS (X, Y) as the target ground positioning point, wherein the coordinates X and Y are calculated by the following equations:

[0037]

[0038]

[0039] Among them, X1, X2, ..., Xn are the horizontal coordinates corresponding to the point cloud of the current frame obtained by the positioning unit;

[0040] Y1, Y2, ..., Yn are the vertical coordinates corresponding to the current frame point cloud obtained by the positioning unit.

[0041] According to an exemplary embodiment of the present disclosure, deriving the roughness of the target road surface to be detected based on the filtered result and the position of the target road surface to be detected in the map includes:

[0042] S71: The ground coordinates POS (X, Y) of the current frame of the single-frame road point cloud data are converted into the map coordinate system as POS_MAP (X, Y) by the following equation: m , Y m ):

[0043] X m =cos(YAW)*X+sin(YAW)*r+X lo

[0044] Ym =cos(YAW)*Y-sin(YAW)*X+Y lo

[0045] Among them, X lo , Y lo and Yaw are the positioning coordinates and heading angle of the mobile vehicle in the map coordinate system obtained by the positioning unit, respectively;

[0046] X m and Y m are respectively the ground coordinates of the current frame of the single frame road point cloud data;

[0047] S72: The map road coordinate POS_MAP (X m , Y m ) is matched with the corresponding roughness index Index, and the roughness of the target road surface to be detected is output.

[0048] According to another aspect of the present disclosure, there is also provided a system for detecting roughness of an unhardened road surface, comprising:

[0049] A sensor is mounted on a mobile vehicle and is used to collect a frame of point cloud data with a frame time stamp of a road surface within a preset range in front of the vehicle, wherein the point cloud data includes a plurality of vertical columns of point clouds, and each vertical column of point clouds includes a plurality of point clouds;

[0050] A positioning unit, the positioning unit is used to obtain vehicle posture data of the mobile vehicle;

[0051] A road surface roughness determination module, wherein the road surface roughness calculation module is configured to: perform time-series matching of the collected single-frame point cloud data and the vehicle posture data; extract point cloud data of the target road surface to be detected based on the full-frame point cloud data of the road surface collected by the sensor; determine the position of the target road surface to be detected in the map based on the vehicle posture data; perform smoothness analysis on the point clouds on a column in the extracted single-frame point cloud data, and perform smoothness analysis on the point clouds on other columns in the extracted single-frame point cloud data column by column to obtain an obstacle point set; filter dynamic targets on the target road surface to be detected based on the obtained obstacle point set; and derive the roughness of the target road surface to be detected based on the filtered result and the position of the target road surface to be detected in the map.

[0052] According to another aspect of the present disclosure, an electronic device is provided, which includes: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the method for detecting roughness of a non-hardened road surface as described above.

[0053] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for detecting roughness of an unhardened road surface as described above are executed.

[0054] According to the roughness detection method and detection system for non-hardened road surfaces described in the above embodiments of the present disclosure, the real-time road conditions can be accurately controlled, the data can be quantified for comparison, and can be archived for query. Compared with the extensive manual management method, it effectively reduces the cost expenditure including manpower, energy, and equipment loss, and improves the transportation efficiency of the operating fleet, thereby achieving the goal of reducing costs and increasing efficiency of road maintenance operations through scientific and technological means.

[0055] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0057] Figure 1 The present invention is a flowchart of a vehicle-mounted roughness detection method for an unhardened road surface according to an exemplary embodiment of the present invention.

[0058] Figure 2 is a schematic diagram of smoothness analysis according to an exemplary embodiment of the present disclosure;

[0059] Figure 3 is a smoothness determination rule according to an exemplary embodiment of the present disclosure.

[0060] Figure 4 This is a visualization result diagram of rutting detection on a flat road surface.

[0061] Figure 5 is a schematic diagram of a vehicle-mounted roughness detection system for an unhardened road surface according to an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0062] In order to more clearly explain the purpose, technical solutions and advantages of the present disclosure, the embodiments of the present disclosure will be described in detail below in conjunction with the accompanying drawings. It should be understood that the following description of the embodiments is intended to explain and illustrate the overall concept of the present disclosure and should not be understood as a limitation of the present disclosure. In the specification and the drawings, the same or similar reference numerals refer to the same or similar parts or components. For the sake of clarity, the drawings are not necessarily drawn to scale, and some well-known parts and structures may be omitted in the drawings.

[0063] Unless otherwise defined, the technical terms or scientific terms used in the present disclosure should be understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. The wording "one" or "an" does not exclude multiple. "Including" or "comprising" and similar words mean that the elements or objects appearing in front of the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right", "top" or "bottom" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly. When an element such as a layer, a film, a region or a substrate substrate is referred to as being "on" or "under" another element, the element may be "directly" located "on" or "under" another element, or there may be an intermediate element.

[0064] like Figure 1 As shown, a method for detecting roughness of an unhardened road surface according to the present disclosure includes:

[0065] S1: collecting a frame of point cloud data with a frame time stamp of a road surface within a preset range in front of the vehicle through a sensor installed on the mobile vehicle, wherein the point cloud data includes a plurality of vertical columns of point clouds, each vertical column of point clouds includes a plurality of point clouds, and obtaining vehicle posture data of the mobile vehicle through a positioning unit;

[0066] S2: performing time sequence matching on the collected single-frame point cloud data and the vehicle posture data;

[0067] S3: extracting point cloud data of a target road surface to be detected based on the full-frame point cloud data of the road surface collected by the sensor;

[0068] S4: Determine the position of the target road surface to be detected in the map based on the vehicle posture data;

[0069] S5: performing a smoothness analysis on a point cloud in one column of the extracted single-frame point cloud data, and performing a smoothness analysis on the point clouds in other columns of the extracted single-frame point cloud data column by column to obtain an obstacle point set;

[0070] S6: filtering the dynamic targets on the target road surface to be detected based on the obtained obstacle point set;

[0071] S7: deriving the roughness of the target road surface to be detected based on the filtered result and the position of the target road surface to be detected in the map.

[0072] Specifically, in step S1, in view of the actual conditions of large road slope, poor flatness and bumpy driving in mining transportation operations, it is preferred to use surface scanning instead of line scanning to collect ground data. The single-frame data from the surface scanning can avoid the splicing offset caused by the violent shaking of the vehicle body. Therefore, a sensor that can generate array three-dimensional point cloud data, such as a multi-line laser radar or a three-dimensional camera, can be used, and the sensor can be mounted at the front position of the mobile vehicle and scan the ground at a downward FOV angle.

[0073] In this embodiment, an RTK (Real Time Kinematic) positioning unit or other positioning equipment is usually used to obtain the vehicle posture data of the mobile vehicle. However, the frequency of the obtained vehicle posture data is usually more than 10 times that of sensors such as radar. Therefore, when the acquired single-frame point cloud data and the vehicle posture data are time-series matched, the RTK timestamp with the smallest absolute value of the difference with the timestamp tc of the single-frame point cloud data sensed by the sensor is selected from the RTK timestamps t1, t2...tn to be matched, and the vehicle posture corresponding to the RTK timestamp is used as the vehicle posture corresponding to the current frame point cloud data. The vehicle posture information includes the positioning coordinates LOCAL (X lo , Y lo ) and heading angle YAW.

[0074] In step S3, the point cloud data of the target road surface to be detected is extracted based on the full-frame point cloud data collected by the sensor. Specifically, for example, a map filter can be used to perform regional screening on the full-frame point cloud data based on radar external parameters, vehicle posture data, etc., to shield the background information such as retaining walls, mountains, buildings, etc. outside the drivable road surface area in the current frame point cloud data, and only retain the point cloud data of the target road surface to be detected. For the specific implementation method and usage of the map filter, please refer to the patent NO.CN115984264B in the name of the applicant: "Map-based static obstacle filtering method and map filter", which is fully incorporated into this article by reference.

[0075] In step S5, the smoothness analysis of the point cloud on a column in the extracted single frame point cloud data includes the following steps:

[0076] S51: deleting the zero point on the column in the single-frame point cloud data, re-recording the remaining points as P1, P2, ..., Pk, if k < 2, it is considered that there are too few valid points on the column, and turning to the next column;

[0077] S52: Reset the obstacle point set, clear the temporary point set, initialize the accumulated height difference ΔT=0, and set the height difference threshold T and the gradient change threshold α;

[0078] S53: performing adjacent point fluctuation calculations on two adjacent points in sequence from one end to the other end along the vertical column, and obtaining an obstacle point set on the vertical column based on the results of the adjacent point fluctuation calculations.

[0079] Among them, in step S53, performing adjacent point fluctuation calculation on two adjacent points along the column from one end to the other end in sequence, and obtaining the obstacle point set on the column based on the result of the adjacent point fluctuation calculation includes the following steps:

[0080] S531: The point at the front of the two adjacent points is used as the comparison point, and the point at the back is used as the compared point. The gradient value between the two adjacent points is calculated by the following equation:

[0081]

[0082] Among them, x1, y1, z1 are the three-dimensional space coordinates of the adjacent points to be compared;

[0083] x2, y2, z2 are the three-dimensional space coordinates of the adjacent points to be compared;

[0084] S532: If Then ΔT=|z 1 -z 2 |, put the current matched adjacent point into the temporary point set, and use the current matched adjacent point as the next round of matched adjacent point, and the next point adjacent to the current matched adjacent point as the next round of matched adjacent point, and recalculate the gradient value of the next round if like Figure 3 As shown in (b), the next round of ΔT next Accumulate it with the ΔT of this round to get ΔT total , and put the next round of compared points into the temporary point set, if Then directly convert the accumulated ΔT total Compare with the height threshold T, if ΔT total <T, such as Figure 3As shown in (d) in the figure, for ΔT total and the temporary point set is cleared, if ΔT total >T, then all points in the temporary point set are transferred to the obstacle point set, and ΔT total and the temporary point set are cleared;

[0085] if like Figure 3 As shown in (a) and (c) in the figure, the current matched point is used as the next matched point, and the next point adjacent to the current matched point is used as the next matched point, and the gradient value of the next round is recalculated.

[0086] In the prior art, conventional RANSAC (random consistency) is usually used to fit the plane equation of the ground, and then the ratio of surface points (matching the ground plane) and outliers (not on the ground plane) is calculated to infer the flatness and smoothness of the area. However, due to the large number of steep slopes at the mining site, multiple surfaces will be fitted at the slope change, and the plane and slope surfaces are normal working surfaces. At the same time, there will be large dynamic objects such as dump trucks and construction machinery at the mining site to generate work activities. When these objects are close to the sensors of this system, they will generate large-area interference information such as inclined planes and vertical planes, resulting in abnormal ground plane fitting results. According to an embodiment of the present disclosure, the ground is fitted by adopting a special ray method based on sensor structured information, that is, by taking the vertical columns (cols) of the single-frame road point cloud Cloud_G data matrix as a ray, and performing smoothness analysis on the ray, such as Figure 2 As shown, the ray height difference threshold T and the ray / car body normal angle (gradient value) α are set to determine the flatness of the ground. After completing the smoothness analysis of the single column data point by point according to the above steps, the smoothness analysis process is repeated column by column until all points in the road point cloud Cloud_G participate in the analysis to obtain the final obstacle point set Cloud_B. The embodiments of the present disclosure can effectively overcome the above problems in the prior art by using a special ray method based on sensor structured information for ground fitting.

[0087] Since the obstacle point set Cloud_B generated in step S5 may also include a varying number of dynamic target noise points, in step S6, the dynamic targets on the target road surface to be detected are filtered based on the obtained obstacle point set, so as to analyze the dynamic targets such as people, vehicles, construction machinery, animals, etc. that appear accidentally on the road surface and exclude them from the final analysis process, thereby improving the accuracy of detection. Specifically, filtering the dynamic targets on the road surface to be detected based on the obtained obstacle point set includes:

[0088] S61: Perform spatial distance clustering on all points in the obstacle point set Cloud_B, and cluster the obstacle point cloud into point cloud clusters; the obstacle point set Cloud_B generated in step S5 may also include a varying number of dynamic target noise points, so perform Euclidean spatial distance clustering on all points in Cloud_B, and cluster the obstacle point cloud into point cloud clusters. The Euclidean spatial distance clustering method is already a mature algorithm in the technical field. The performance acceleration preprocessing method KD-Tree (i.e., K-dimension tree, nearest neighbor search) used in the present invention completes the clustering operation. The open source function library pcl provides pcl::extractEuclideanClusters to complete the relevant operations;

[0089] S62: Calculate the minimum spatial bounding box of each point cloud cluster; specifically, firstly, obtain the two-dimensional bounding box of the BEV (top view) of each point cloud cluster using the cv::minAreaRect method in the open source function library Opencv according to the projection of the point cloud cluster on the XY horizontal plane, and then traverse all points in the cluster to search for the maximum and minimum values ​​of the z-axis as the bounding box boundaries in the vertical direction, and calculate the minimum spatial bounding box of each point cloud cluster accordingly;

[0090] S63: Formulate dynamic object filtering parameter thresholds based on each instance individual in the work site, and generate dynamic filter rules; for example, assuming that x, y, and z are the length, width, and height of the bounding box, respectively, the filtering rule example is as follows:

[0091] z>1.4 m;

[0092] x×y×4<z×z square meters;

[0093] x×y×z<12 cubic meters;

[0094] S64: Based on the dynamic filter rules, the data of the corresponding point cloud clusters in the obstacle point set Cloud_B and the point cloud data Cloud_G of the target road to be detected are cleared. If the bounding box parameters do not meet the dynamic filter rules, it is determined to be a non-road dynamic object, and the data of the corresponding point cloud clusters in the obstacle point set Cloud_B and the point cloud data Cloud_G of the target road to be detected are cleared. For example, an object with a Z value of 1.0 meters should not be a road dynamic object, because the height of people and cars, that is, the value of Z, must be greater than 1.4 meters. The filter threshold is not a fixed value, and should be set according to different road rules. If there are compact cars on the road (small cars used in mining areas are basically off-road SUVs and pickups with high chassis), the height is less than 1.4 meters and the volume is less than 12 cubic meters. In this case, the filter cannot be removed, and it is necessary to consider lowering the filter threshold.

[0095] According to the embodiments of the present disclosure, the obstacle point set Cloud_B that has been cleared is the set of road outliers determined by the algorithm. Non-smooth objects such as ruts, fallen rocks, loose soil, etc. that cause ground convexities and concavities are the source of outliers. According to the point cloud pixel sampling principle of the three-dimensional sensor, the physical volume, floor area and number of non-smooth objects are directly proportional to the number of outliers. Therefore, the output roughness metric can be quantified as the ratio of the number of road outliers to the number of surface points on the ground plane. That is: the road roughness index is determined by the following equation based on the obstacle point set that has been cleared:

[0096] Index=sizeB / (sizeG-sizeB),

[0097] Among them, Index is the roughness index of the target road surface to be detected;

[0098] sizeB is the final number of points in the obstacle point set after removal;

[0099] sizeG is the total number of points (including surface points and outliers) in the point cloud data of the target road surface to be detected after the cleaning process.

[0100] In step S4, determining the position of the target road surface to be detected in the map based on the vehicle posture data includes: marking the current frame ground coordinate POS (X, Y) as the target ground positioning point, wherein the coordinates X and Y are calculated by the following equations:

[0101]

[0102]

[0103] Among them, X1, X2, ..., Xn are the horizontal coordinates corresponding to the point cloud of the current frame obtained by the positioning unit;

[0104] Y1, Y2, ..., Yn are the vertical coordinates corresponding to the current frame point cloud obtained by the positioning unit.

[0105] According to an exemplary embodiment, deriving the roughness of the target road surface to be detected based on the filtered result and the position of the target road surface to be detected in the map includes the following steps:

[0106] S71: The ground coordinates POS (X, Y) of the current frame of the single-frame road point cloud data are converted into the map coordinate system as POS_MAP (X, Y) by the following equation: m , Y m ):

[0107] X m=cos(YAW)*X+sin(YAW)*Y+X lo

[0108] Y m =cos(YAW)*Y-sin(YAW)*X+Y lo

[0109] Among them, X lo , Y lo and Yaw are the positioning coordinates and heading angle of the mobile vehicle in the map coordinate system obtained by the positioning unit, respectively;

[0110] X m and Y m are respectively the ground coordinates of the current frame of the single frame road point cloud data;

[0111] S72: The map road coordinate POS_MAP (X m , Y m ) is matched with the corresponding roughness index Index, and the roughness of the target road surface to be detected is output.

[0112] Preferably, in an exemplary embodiment, an image acquisition device is also provided on the mobile vehicle to compare the outputted roughness of the target road surface to be detected with the optical image acquired and displayed by the image device, such as Figure 4 to further verify the accuracy of the results.

[0113] According to another aspect of the present disclosure, Figure 5 As shown, a roughness detection system for an unhardened road surface is also provided, comprising: a sensor 1, the sensor is mounted on a mobile vehicle 4, and is used to collect a frame of point cloud data with a frame time stamp of a road surface within a preset range in front of the vehicle, wherein the point cloud data includes a plurality of vertical columns of point clouds, and each vertical column of point clouds includes a plurality of point clouds; a positioning unit 2, the positioning unit 2 is used to obtain vehicle posture data of the mobile vehicle 4; and a road surface roughness determination module 3, the road surface roughness calculation module 3 is configured to: perform time sequence matching on the collected single frame point cloud data and the vehicle posture data; based on the sensor The method comprises the following steps: extracting point cloud data of a target road surface to be detected from the full-frame point cloud data of the road surface collected by the sensor; determining the position of the target road surface to be detected in the map based on the vehicle posture data; performing smoothness analysis on the point clouds on one column in the extracted single-frame point cloud data, and performing smoothness analysis on the point clouds on other columns in the extracted single-frame point cloud data column by column to obtain an obstacle point set; filtering dynamic targets on the target road surface to be detected based on the obtained obstacle point set; and deriving the ruggedness of the target road surface to be detected based on the filtered result and the position of the target road surface to be detected in the map.

[0114] According to another aspect of the present disclosure, an electronic device is also provided, including: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the non-hardened road roughness detection method as described above.

[0115] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for detecting roughness of an unhardened road surface as described above are executed.

[0116] According to the roughness detection method and detection system of the non-hardened road surface described in the above embodiment of the present disclosure, the real-time road condition of the road surface can be accurately controlled, the data can be quantified and compared, and can be archived and queried. Compared with the extensive management method of manual work, the cost expenditure including manpower, energy, and equipment loss is effectively reduced, and the transportation efficiency of the operation team is improved, achieving the purpose of reducing costs and increasing efficiency of road maintenance operations through scientific and technological means. The method can be applied to non-hardened roads in mining environments, for example, and can also be applied to hardened roads with slightly better road conditions.

[0117] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0118] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0119] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for detecting roughness of an unhardened road surface, comprising: S1: collecting a frame of point cloud data with a frame timestamp of a road surface within a preset range in front of the vehicle through a laser radar installed on the moving vehicle, wherein the point cloud data includes a plurality of vertical columns of point clouds, each vertical column of point clouds includes a plurality of point clouds, and obtaining vehicle posture data of the moving vehicle through a positioning unit; S2: performing time sequence matching on the collected single-frame point cloud data and the vehicle posture data; S3: extracting point cloud data of a target road surface to be detected based on the full-frame point cloud data of the road surface collected by the laser radar; S4: Determine the position of the target road surface to be detected in the map based on the vehicle posture data; S5: performing smoothness analysis on the point clouds on one column of the extracted single-frame point cloud data, and performing smoothness analysis on the point clouds on other columns of the extracted single-frame point cloud data column by column to obtain an obstacle point set, and performing smoothness analysis on the point clouds on one column of the extracted single-frame point cloud data includes the following steps: S51: Delete the zero point on the column in the single frame point cloud data, and re-record the remaining points as P1, P2, ... Pk. If k<2, it is considered that there are too few valid points on the column, and go to the next column. S52: Reset the obstacle point set, clear the temporary point set, initialize the accumulated height difference ΔT=0, and set the height difference threshold T and the gradient change threshold α; S53: performing adjacent point fluctuation calculation on two adjacent points in sequence from one end to the other end along the longitudinal column, and obtaining an obstacle point set on the longitudinal column based on the result of the adjacent point fluctuation calculation; S6: filtering the dynamic targets on the target road surface to be detected based on the obtained obstacle point set; S7: deriving the roughness of the target road surface to be detected based on the filtered result and the position of the target road surface to be detected in the map.

2. The ruggedness detection method according to claim 1, wherein: The laser radar collects data on the road surface in a surface scanning manner.

3. The ruggedness detection method according to claim 1, wherein: Extracting point cloud data of a target road surface to be detected based on the full-frame point cloud data of the road surface collected by the laser radar includes: performing intra-region screening on the full-frame point cloud data through a map filter to shield background information outside the drivable road surface area in the current frame data, and only retaining the point cloud data of the target road surface to be detected.

4. The ruggedness detection method according to claim 1, wherein: Performing adjacent point fluctuation calculations on two adjacent points along the column from one end to the other end, and obtaining an obstacle point set on the column based on the results of the adjacent point fluctuation calculations comprises the following steps: S531: The point at the front of the two adjacent points is used as the comparison point, and the point at the back is used as the compared point. The gradient value between the two adjacent points is calculated by the following equation: Among them, x1, y1, z1 are the three-dimensional space coordinates of the adjacent points to be compared; x2, y2, z2 are the three-dimensional space coordinates of the adjacent points to be compared; S532: If then ΔT = |z1 - z2|, put the currently compared adjacent point into the temporary point set, and use the currently compared adjacent point as the compared adjacent point for the next round. The next point adjacent to the currently compared adjacent point is used as the point to be compared for the next round, and recalculate the gradient value for the next round. If next > α, then for the ΔT of the next round next accumulate it with the ΔT of this round to obtain ΔT total , and put the point to be compared for the next round into the temporary point set. If then directly compare the already accumulated ΔT total with the height threshold T. If ΔT total < T, then clear ΔT total and the temporary point set. If ΔT total > T, then transfer all the points in the temporary point set to the obstacle point set, and clear ΔT total and the temporary point set; if Then the current point to be compared is used as the next point to be compared, and the next point adjacent to the current point to be compared is used as the next point to be compared, and the gradient value of the next round is recalculated.

5. The ruggedness detection method according to claim 1, wherein: Filtering the dynamic targets on the target road surface to be detected based on the obtained obstacle point set includes: S61: performing spatial distance clustering on all points in the obstacle point set, and clustering the obstacle point cloud into point cloud clusters; S62: Calculate the minimum spatial bounding box of each point cloud cluster; S63: Determine a dynamic object filtering parameter threshold according to each instance individual in the work site, and generate a dynamic filter rule; S64: Based on the dynamic filter rule, clear the data corresponding to the point cloud cluster in the obstacle point set and the point cloud data of the target road surface to be detected.

6. The ruggedness detection method according to claim 5, wherein: The road roughness index is determined based on the cleared obstacle point set using the following equation: Index=sizeB / (sizeG-sizeB), Among them, Index is the roughness index of the target road surface to be detected; sizeB is the final number of points in the obstacle point set after the removal process; sizeG is the total number of points in the point cloud data of the target road surface to be detected after the cleaning process.

7. The ruggedness detection method according to claim 6, wherein: In step S4, determining the position of the target road surface to be detected in the map based on the vehicle posture data includes: marking the current frame ground coordinate POS (X, Y) as the target ground positioning point, wherein the coordinates X and Y are calculated by the following equations: Among them, X1, X2, ..., Xn are the horizontal coordinates corresponding to the point cloud of the current frame obtained by the positioning unit; Y1, Y2, ..., Yn are the vertical coordinates corresponding to the current frame point cloud obtained by the positioning unit.

8. The ruggedness detection method according to claim 7, wherein: Determining the roughness of the target road surface to be detected based on the filtered result and the position of the target road surface to be detected in the map includes: S71: The ground coordinates POS(X, Y) of the current frame of the single-frame road point cloud data are converted into the map coordinate system as POS_MAP(X, Y) by the following equation: m ,Y m ): X m =cos(YAW)*X+sin(YAW)*Y+X lo AND m =cos(YAW)*Y-sin(YAW)*X+Y lo Among them, X lo , Y lo and Yaw are the positioning coordinates and heading angle of the mobile vehicle in the map coordinate system obtained by the positioning unit, respectively; X m and Y m are respectively the ground coordinates of the current frame of the single frame road point cloud data; S72: The map road coordinate POS_MAP (X m ,Y m ) is matched with the corresponding roughness index Index, and the roughness of the target road surface to be detected is output.

9. A system for detecting the roughness of an unhardened road surface using the roughness detection method of claim 1, comprising: A laser radar, which is installed on a mobile vehicle and is used to collect a frame of point cloud data with a frame time stamp of a road surface within a preset range in front of the vehicle, wherein the point cloud data includes a plurality of vertical columns of point clouds, and each vertical column of point clouds includes a plurality of point clouds; A positioning unit, the positioning unit is used to obtain vehicle posture data of the mobile vehicle; A road surface roughness determination module, wherein the road surface roughness calculation module is configured to: perform time-series matching of the collected single-frame point cloud data and the vehicle posture data; extract point cloud data of the target road surface to be detected based on the full-frame point cloud data of the road surface collected by the laser radar; determine the position of the target road surface to be detected in the map based on the vehicle posture data; perform smoothness analysis on the point clouds on a column in the extracted single-frame point cloud data, and perform smoothness analysis on the point clouds on other columns in the extracted single-frame point cloud data column by column to obtain an obstacle point set; filter dynamic targets on the target road surface to be detected based on the obtained obstacle point set; and derive the roughness of the target road surface to be detected based on the filtered result and the position of the target road surface to be detected in the map.

10. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the method for detecting the roughness of a non-hardened road surface as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for detecting the roughness of an unhardened road surface as claimed in any one of claims 1 to 8 are executed.