CN116863447A

CN116863447A - Guardrail detection method and device, electronic equipment and storage medium

Info

Publication number: CN116863447A
Application number: CN202310953274.1A
Authority: CN
Inventors: 张晶华; 耿秀军; 张丹
Original assignee: Uisee Shanghai Automotive Technologies Ltd
Current assignee: Uisee Shanghai Automotive Technologies Ltd
Priority date: 2023-07-31
Filing date: 2023-07-31
Publication date: 2023-10-10

Abstract

The embodiment of the disclosure discloses a guardrail detection method, a device, electronic equipment and a storage medium, wherein at least one obstacle target is determined by acquiring a current frame point cloud, final node statistical information is determined through a corresponding point cloud block of each obstacle target, whether the obstacle target meets a length continuous condition is judged, whether the obstacle target which does not meet the length continuous condition meets a height continuous condition is judged according to the final node statistical information, the obstacle target meeting the length continuous condition and the obstacle target meeting the height continuous condition are determined to be candidate guardrail targets of the current frame, and each formal guardrail target of the current frame is determined by combining with each formal guardrail target of a history frame, so that guardrail detection results of the current frame are determined, guardrail detection based on laser radar point cloud is realized, and the problem of guardrail detection omission caused by shielding or other factor deletion of guardrail point clouds is solved.

Description

Guardrail detection method and device, electronic equipment and storage medium

Technical Field

The disclosure relates to the field of automatic driving environment sensing, in particular to a guardrail detection method, a guardrail detection device, electronic equipment and a storage medium.

Background

The current guardrail detection method mainly comprises a traditional rule-based method and a deep learning method. The traditional rule-based method only adopts a single height threshold value and a single width threshold value to judge, and if a certain section of guardrail point cloud is lost due to shielding or other factors, the section of guardrail is missed.

In addition, the deep learning method is used for specially detecting the guardrail targets through the corresponding models of data driving training, but the occupied computing resources are more, and for the limited computing resources of the automatic driving vehicle-mounted platform, the method is not suitable for adding one model to specially detect the guardrail targets, even if the guardrail targets are added into the conventional obstacle detection model, the guardrail targets have fewer quantity compared with other targets, the long tail effect is caused, and the guardrail detection effect is not ideal.

Disclosure of Invention

In order to solve the above technical problems or at least partially solve the above technical problems, the embodiments of the present disclosure provide a method, an apparatus, an electronic device, and a storage medium for detecting a guardrail, which solve the problem of missed detection of a guardrail caused by shielding or other missing factors of a point cloud of the guardrail, and perform rapid, effective and stable detection on a guardrail target, without increasing excessive computing load.

In a first aspect, an embodiment of the present disclosure provides a guardrail detection method, including:

acquiring a current frame point cloud, determining at least one obstacle target in the current frame point cloud, and determining final node statistical information of the obstacle targets based on point cloud blocks corresponding to the obstacle targets aiming at each obstacle target;

judging whether each obstacle target meets a length continuous condition based on final node statistical information of each obstacle target, and judging whether each obstacle target meets a height continuous condition based on final node statistical information of each obstacle target aiming at the obstacle target which does not meet the length continuous condition;

determining an obstacle target meeting the length continuous condition as a candidate guardrail target of the current frame;

and determining each formal guardrail target of the current frame based on each candidate guardrail target of the current frame and each formal guardrail target of the history frame, and determining the guardrail detection result of the current frame according to each formal guardrail target of the current frame.

In a second aspect, embodiments of the present disclosure also provide a guardrail detection device, the device comprising:

The node statistics module is used for acquiring the point cloud of the current frame, determining at least one obstacle target in the point cloud of the current frame, and determining final node statistics information of the obstacle targets based on point cloud blocks corresponding to the obstacle targets aiming at each obstacle target;

the continuous judging module is used for judging whether each obstacle target meets a length continuous condition or not based on the final node statistical information of each obstacle target, and judging whether the obstacle target meets a height continuous condition or not based on the final node statistical information of the obstacle target aiming at the obstacle target which does not meet the length continuous condition;

a candidate determining module, configured to determine, as candidate guard rail targets of the current frame, an obstacle target that satisfies the length continuous condition and an obstacle target that satisfies the height continuous condition;

and formally determining targets, wherein the formally determining targets are used for determining all formally guardrail targets of the current frame based on all candidate guardrail targets of the current frame and all formally guardrail targets of the history frame, and determining guardrail detection results of the current frame according to all formally guardrail targets of the current frame.

In a third aspect, embodiments of the present disclosure further provide an electronic device, including: one or more processors; a storage means for storing one or more programs; the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the guardrail detection method as described above.

In a fourth aspect, the disclosed embodiments also provide a computer readable storage medium having stored thereon a computer program which when executed by a processor implements the guardrail detection method as described above.

According to the guardrail detection method provided by the embodiment of the disclosure, at least one obstacle target is determined by acquiring the point cloud of the current frame, final node statistical information of the obstacle targets is determined for each obstacle target through the corresponding point cloud block of each obstacle target, whether the obstacle targets meet the length continuous condition is judged according to the final node statistical information, whether the obstacle targets which do not meet the length continuous condition meet the height continuous condition is judged according to the final node statistical information, the obstacle targets which meet the length continuous condition and the obstacle targets which meet the height continuous condition are determined to be candidate guardrail targets of the current frame, and each formal guardrail target of the current frame is determined by combining with each formal guardrail target of a history frame, so that guardrail detection results of the current frame are determined, guardrail detection based on the laser radar point cloud is realized.

Drawings

The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by reference to the following detailed description when taken in conjunction with the accompanying drawings. The same or similar reference numbers will be used throughout the drawings to refer to the same or like elements. It should be understood that the figures are schematic and that elements and components are not necessarily drawn to scale.

FIG. 1 is a flow chart of a guardrail detection method in an embodiment of the present disclosure;

FIG. 2 is a schematic diagram of a lateral point cloud in an embodiment of the present disclosure;

FIG. 3 is a schematic diagram of a longitudinal point cloud in an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of an traversal process according to an embodiment of the disclosure;

FIG. 5 is a schematic view of an obstacle target in an embodiment of the disclosure

FIG. 6 is a schematic structural view of a guardrail detection device according to an embodiment of the present disclosure;

fig. 7 is a schematic structural diagram of an electronic device in an embodiment of the disclosure.

Detailed Description

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure have been shown in the accompanying drawings, it is to be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustration purposes only and are not intended to limit the scope of the present disclosure.

It should be noted that the terms "first," "second," and the like in this disclosure are merely used to distinguish between different devices, modules, or units and are not used to define an order or interdependence of functions performed by the devices, modules, or units.

The names of messages or information interacted between the various devices in the embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

Before describing in detail a guardrail detection method provided by an embodiment of the present disclosure, a description is given of a technical problem solved by the method. In the prior art, the guardrail detection method mainly comprises a traditional rule-based method and a deep learning method. For example, the rule-based method and the deep learning method can be seen in the following schemes:

in patent 1 (publication number CN111881752 a), the method projects point cloud on an image through an external reference matrix based on the image and the point cloud data, establishes a corresponding relation between image pixels and the point cloud, then detects a guardrail target in the image by using a deep learning method, distinguishes whether the guardrail is positioned on the left side or the right side of a vehicle according to a detection result, and simultaneously segments the point cloud data by using a point cloud semantic segmentation network to obtain a segmentation result of the guardrail target. And finally, the final detection result and the attribute of the guardrail target are obtained by combining the detection result and the segmentation result.

Patent 2 (publication number CN115327539 a), the method uses millimeter wave radar point cloud to detect the guardrail based on the characteristic that millimeter wave radar is sensitive to guardrail target perception. During detection, firstly, performing gridding treatment on the millimeter wave radar, then performing depth-first search on the gridded point cloud data, searching out a guardrail detection target with the longest length, and finally, fitting the detected guardrail target to obtain a curve equation of the guardrail.

Patent 3 (publication number CN116129359 a), this method proposes a method for off-line detection of guardrails using a deep learning method based on image data. The edge in the image is detected by using a canny algorithm, the edge detection image is spliced with the original image in the channel dimension, then the improved Resnet50 is used for detecting the guardrail target, and the detected target result is used for maintenance analysis of the guardrail.

Patent 4 (US 20230069618 A1) proposes a guardrail detection method based on point cloud, which firstly uses a plane fitting method to remove the ground point cloud, then constructs a 3D spherical voxel space based on the characteristics of mechanical laser radar rotation scanning, divides the point cloud into various voxels, and distinguishes whether the voxels contain the point cloud. And further selecting voxels which are possibly guardrails based on the characteristic that the guardrails are parallel to the vehicle, extracting points in the voxels, fitting to obtain candidate guardrail targets, and finally carrying out final verification on the guardrail candidate targets by utilizing the preset width and height of the guardrail targets.

The above patent 1 and patent 3 use a method based on deep learning, which is to train a corresponding model through data driving to detect the guardrail target specifically, and has the advantages of high accuracy, continuous iterative update of the model, but more occupied computing resources, because for the limited computing resources of the automatic driving vehicle-mounted platform, a model is not suitable for being additionally arranged to detect the guardrail target specifically, even if the guardrail target is added to a conventional obstacle detection model, the guardrail target has a small quantity compared with other targets, the guardrail target can cause long tail effect, and the guardrail detection effect is not ideal.

The above patent 2 and patent 4 adopt traditional rule-based methods, the patent 2 is based on the characteristics that millimeter wave radars are sensitive to metal target perception to rapidly position guardrails, then whether the targets are guardrail targets is judged according to the fitted point cloud length, the method is fast in processing speed, but the disadvantage is that millimeter wave radars are sensitive to metal targets, various metal targets are easy to form continuous point clouds, and the common millimeter wave radars lack target height information, so that the targets are difficult to verify in terms of height. In addition, the length threshold value of the guardrail judgment is difficult to select, so that false detection is easy to generate by the method. Patent 4 utilizes laser radar point cloud data to detect the guardrail, and the guardrail target can be effectively detected based on the characteristic that laser radar point cloud data coordinate position accuracy is high. However, the 3D spherical space voxel division method used by the method is long in time consumption and is not suitable for an automatic driving platform with high real-time requirements, the method depends on judgment of the flatness and straightness of the guardrail, and the detection effect of a non-planar guardrail or a guardrail target at a turning position can be greatly influenced. In addition, when the guardrail target is checked finally, only a single height threshold value and a single width threshold value are adopted for judgment, the continuity of the guardrail in the height direction and the width direction is not considered, and if a certain section of guardrail point cloud is lost due to shielding or other factors, the section of guardrail is missed.

Therefore, in order to solve the above-mentioned problem, the embodiment of the disclosure provides a guardrail detection method, which aims at the problems of large occupation of computing resources and long time consumption of a deep learning method, and adopts a traditional rule-based thought design algorithm, so that excessive computing load is not increased, and real-time operation can be realized under the limited computing resources of a vehicle-mounted platform. In addition, aiming at the defects that 3D space voxel division takes longer time, depends on flatness and straightness judgment, uses a single height and width threshold value for judgment and the like in the patent 4, the guardrail target is rapidly, effectively and stably detected through the processes of identifying the obstacle target, determining final node statistical information, judging whether the length continuous condition is met, judging whether the height continuous condition is met and the like, and the robustness of guardrail detection is further improved.

Fig. 1 is a flowchart of a guardrail detection method in an embodiment of the present disclosure. The method can be suitable for determining the corresponding guardrail detection result based on the point cloud data acquired by the vehicle-mounted laser radar, so as to carry out the path planning or obstacle avoidance control of the vehicle through the guardrail detection result. The method may be performed by a guardrail detection device, which may be implemented in software and/or hardware, which may be configured in an electronic device. As shown in fig. 1, the method specifically may include the following steps:

S110, acquiring a current frame point cloud, determining at least one obstacle target in the current frame point cloud, and determining final node statistical information of the obstacle targets based on point cloud blocks corresponding to the obstacle targets aiming at each obstacle target.

The current frame point cloud may be a point cloud of a laser radar carried by the vehicle under a vehicle body coordinate system acquired at the current moment. Specifically, obstacle point separation, grid map division, grid state determination and target clustering can be performed on the current frame point cloud, and at least one obstacle target in the current frame point cloud is identified.

In a specific embodiment, determining at least one obstacle target in the current frame point cloud comprises the steps of:

step 11, screening the point cloud of the current frame to obtain the point cloud positioned in the region of interest in the point cloud of the current frame;

step 12, eliminating the vehicle body point cloud, the ground point cloud and the suspension point cloud in the point cloud positioned in the region of interest based on preset vehicle body size parameters, a minimum height threshold and a maximum height threshold to obtain an obstacle point cloud;

step 13, dividing a grid map of the region of interest to obtain a corresponding grid map, and determining the state of each grid according to the corresponding obstacle points of each grid in the obstacle point cloud, wherein the state of each grid is an occupied state or a non-occupied state;

Step 14, determining at least one obstacle target according to the state of each grid.

The region of interest may be a preset stereoscopic region that does not limit the height. Specifically, the point cloud located in the region of interest can be screened from the point cloud of the current frame through the region of interest. Further, the vehicle body point cloud in the region of interest is eliminated through preset vehicle body size parameters, points smaller than the minimum height threshold value are eliminated from the point cloud in the region of interest, points larger than the maximum height threshold value are eliminated from the point cloud in the region of interest, and finally the eliminated point cloud in the region of interest is used as obstacle point cloud.

Further, the region of interest may be divided into h×w grid maps according to a set size, and for each grid in the grid map, a correspondence between the grid and the obstacle points in the obstacle point cloud located in the grid may be established, that is, the obstacle points in the grid may be indexed by the grid coordinates.

Further, the height information of all the obstacle points in each grid can be counted, the highest obstacle point and the lowest obstacle point in each grid are recorded, and the state of the grid is determined according to the heights of the highest obstacle point and the lowest obstacle point. Wherein the state of the grid may be an occupied state, which means that the grid is occupied by an obstacle, or a non-occupied state, which means that the grid is not occupied by an obstacle.

For each grid, if the height of the highest obstacle point in the grid is greater than a set height threshold value, and the relative height difference between the highest obstacle point and the lowest obstacle point is greater than the set height difference threshold value, the state of the grid may be determined to be an occupied state (e.g., may be denoted as NO-PASS); if the height of the highest obstacle point in the grid is less than the set height threshold, or no obstacle point is present in the grid, or the relative height difference between the highest obstacle point and the lowest obstacle point is not greater than the set height difference threshold, then the state of the grid may be determined to be a non-occupied state (e.g., may be denoted as PASS).

Further, after determining the states of the grids, the communication area of each obstacle target can be obtained by calculating the communication area of each grid occupying the states, and the coordinates, the length, the width and the maximum height of the grid occupying the obstacle target are determined through the communication area, so that the identification of at least one obstacle target in the point cloud of the current frame is realized. For each of the grids in the occupied state, if the state of the adjacent other grids of the grid is the occupied state, the grid and the other grids can be determined to be communicated, and in this way, the communication area of each obstacle target can be determined in the grid map.

Through the steps 11-14, the point cloud of the vehicle body, the point cloud of the ground and the suspended target point cloud which does not influence the passing can be filtered, only the obstacle point cloud which has influence on the passing property in the interested area is reserved, the efficiency of identifying the obstacle target is improved, the determination of the grid state and the target clustering can be realized, and the accuracy of the obstacle target is ensured.

In the embodiment of the present disclosure, after determining at least one obstacle target in the point cloud of the current frame, for each obstacle target, the final node statistics may be determined according to the point cloud block corresponding to the obstacle target. The point cloud block may be formed by all point clouds corresponding to the obstacle targets, the final node statistical information may include a guardrail node list, the number of guardrail nodes and the number of boundary nodes, the guardrail node list is formed by at least one guardrail node, the guardrail node may be a local point cloud block which is detected as a point cloud block which accords with the guardrail width characteristics, and the boundary node may be a guardrail node which is detected to be located on the boundary of the guardrail detection area.

In a specific embodiment, for each obstacle target, determining final node statistics of the obstacle target based on the point cloud block corresponding to the obstacle target includes the following steps:

Step 21, determining each transverse point cloud block corresponding to the obstacle target in the abscissa direction based on each grid corresponding to the obstacle target, wherein the abscissa direction is the direction perpendicular to the vehicle head, and the transverse point cloud blocks are each row of the point cloud blocks of the obstacle target;

step 22, starting from a first transverse point cloud block in each transverse point cloud block, judging whether transverse guardrail nodes and transverse boundary nodes exist according to the width of the transverse point cloud block and the boundary of the guardrail detection area until the width of the transverse point cloud block is not smaller than a preset guardrail width threshold value, and obtaining a transverse guardrail node list, the number of transverse guardrail nodes and the number of transverse boundary nodes;

step 23, determining each longitudinal point cloud block corresponding to the obstacle target in the ordinate direction based on each grid corresponding to the obstacle target, wherein the ordinate direction is the head direction;

step 24, starting from the first longitudinal point cloud block in each longitudinal point cloud block, judging whether longitudinal guardrail nodes and longitudinal boundary nodes exist according to the width of the longitudinal point cloud block and the boundary of the guardrail detection area until the width of the longitudinal point cloud block is not smaller than a preset guardrail width threshold value or all longitudinal point cloud blocks are judged to be finished, and obtaining a longitudinal guardrail node list, the number of longitudinal guardrail nodes and the number of longitudinal boundary nodes;

And step 25, if the number of the transverse guardrail nodes is larger than the number of the longitudinal guardrail nodes, determining final node statistical information of the barrier target according to the transverse guardrail node list, the number of the transverse guardrail nodes and the number of the transverse boundary nodes, otherwise, determining final node statistical information of the barrier target according to the longitudinal guardrail node list, the number of the longitudinal guardrail nodes and the number of the longitudinal boundary nodes.

The horizontal coordinate direction and the vertical coordinate direction can construct a plane parallel to the ground, the horizontal coordinate direction can be understood as a direction perpendicular to the vehicle head in the plane, and the vertical coordinate direction can be understood as the vehicle head direction. In the above step 21, the lateral point cloud may be a lateral partial point cloud among the point clouds of the obstacle target, that is, each row of the point clouds of the obstacle target.

For example, fig. 2 is a schematic diagram of a horizontal dot cloud block in an embodiment of the disclosure, as shown in fig. 2, where the dot cloud block of the obstacle target at the upper left side may be divided into four horizontal dot cloud blocks, that is, four horizontal dot cloud blocks are grids occupied by the obstacle target in the 6 th to 9 th rows, respectively, and the dot cloud block of the obstacle target at the lower right side may be divided into four horizontal dot cloud blocks, that is, four horizontal dot cloud blocks are grids occupied by the obstacle target in the 3 rd to 6 th rows, respectively.

Specifically, in the step 22, from the first horizontal point cloud block of all horizontal point cloud blocks, whether the horizontal point cloud block meets the width characteristics of the guardrail is determined according to the width of the horizontal point cloud block, if yes, the horizontal guardrail node is determined according to the horizontal point cloud block, and whether the horizontal guardrail node is a horizontal boundary node is further determined by combining with the boundary of the guardrail detection area.

For example, taking fig. 2 as an example, for the obstacle target at the upper left, it may be sequentially determined, starting from bottom to top, whether each horizontal point cloud block has a horizontal guardrail node and a horizontal boundary node.

For the step 22, optionally, starting from the first horizontal point cloud block in each horizontal point cloud block, judging whether there are horizontal guardrail nodes and horizontal boundary nodes according to the width of the horizontal point cloud block and the boundary of the guardrail detection area until the width of the horizontal point cloud block is not less than a preset guardrail width threshold value, to obtain a horizontal guardrail node list, the number of horizontal guardrail nodes and the number of horizontal boundary nodes, including:

taking a first transverse point cloud block in the abscissa direction as a current point cloud block, traversing each grid in the current point cloud block along the abscissa direction to obtain the width of the current point cloud block, and judging whether the width of the current point cloud block is smaller than a preset guardrail width threshold value;

If yes, determining the center point of the current point cloud block as a transverse guardrail node, updating the number of the transverse guardrail nodes, determining the transverse guardrail node as a transverse boundary node under the condition that the transverse guardrail node is positioned on the boundary of the guardrail detection area, and updating the number of the transverse boundary node;

and taking the next transverse point cloud block of the current point cloud block as a new current point cloud block, and returning to execute the step of traversing each grid in the current point cloud block along the transverse coordinate direction until the width of the transverse point cloud block is not smaller than a preset guardrail width threshold value, so as to obtain a transverse guardrail node list, the number of transverse guardrail nodes and the number of transverse boundary nodes.

Specifically, the first horizontal point cloud block can be used as a current point cloud block, grids in the current point cloud block are traversed along the horizontal coordinate direction, the width of the current point cloud block is determined according to the number of the grids in the traversed current point cloud block, whether the width is smaller than a preset guardrail width threshold value is judged, if yes, the current point cloud block accords with the guardrail width characteristic, namely, the guardrail has the characteristic of smaller width compared with other targets, so that the current point cloud block can be determined to be a guardrail, the center point of the current point cloud block is used as a horizontal guardrail node, the number of the horizontal guardrail nodes is added, and the global coordinates and the maximum height of the horizontal guardrail node are recorded. The maximum height of the transverse guardrail node can be the height value of the highest point in the corresponding transverse point cloud block.

Further, whether the transverse guardrail node is located on the boundary of the guardrail detection area can be judged, if so, the transverse guardrail node is determined to be a transverse boundary node, and the number of the transverse boundary nodes is added with one. And taking the next transverse point cloud block of the current point cloud block as a new current point cloud block, repeating the process until the width of the transverse point cloud block is not smaller than a preset guardrail width threshold value, and stopping the iterative process at the moment to obtain a transverse guardrail node list, the number of transverse guardrail nodes and the number of transverse boundary nodes. By the method, the transverse traversal of each obstacle target is realized, and then the transverse guardrail node list, the transverse guardrail node number and the transverse boundary node number which are obtained by statistics of each obstacle target in the abscissa direction are obtained.

Specifically, in the step 23, if the width of the horizontal point cloud block is not smaller than the preset guardrail width threshold in the obstacle target, the horizontal traversal is changed to the longitudinal traversal, so as to count the longitudinal guardrail node list, the number of longitudinal guardrail nodes and the number of longitudinal boundary nodes of the obstacle target in the ordinate direction.

Specifically, starting from a first longitudinal point cloud block of the obstacle target, judging whether the longitudinal point cloud block meets the width characteristic of the guardrail according to the width of the longitudinal point cloud block, if so, determining longitudinal guardrail nodes according to the longitudinal point cloud block, and further judging whether the longitudinal guardrail nodes are longitudinal boundary nodes according to the boundary of the guardrail detection area until the width of the longitudinal point cloud block is not smaller than a preset guardrail width threshold or all longitudinal point cloud blocks are judged to be finished.

For example, fig. 3 is a schematic diagram of a longitudinal point cloud block in an embodiment of the disclosure, as shown in fig. 3, where the point cloud block of the obstacle target at the upper left side may be divided into five longitudinal point cloud blocks, that is, five longitudinal point cloud blocks are grids occupied by the obstacle target in columns 4 to 8, respectively, and the point cloud block of the obstacle target at the lower right side may be divided into one longitudinal point cloud block, that is, a grid occupied by the obstacle target in column 7. For the obstacle object at the upper left, whether the longitudinal guardrail nodes and the longitudinal boundary nodes exist in each longitudinal point cloud block can be judged in sequence from left to right, and the specific judging process can be seen from the judging step of the transverse point cloud block.

In the step 24, if there are longitudinal point cloud blocks with widths not less than the preset guardrail width threshold, or if all the longitudinal point cloud blocks are judged to be complete, the longitudinal traversal may be stopped, so as to obtain a longitudinal guardrail node list, the number of longitudinal guardrail nodes and the number of longitudinal boundary nodes.

Further, the number of the nodes of the transverse guardrail and the number of the nodes of the longitudinal guardrail can be compared, if the number of the nodes of the transverse guardrail is larger than the number of the nodes of the longitudinal guardrail, the guardrail direction of the obstacle target is represented as the ordinate direction, at the moment, the list of the nodes of the transverse guardrail, the number of the nodes of the transverse guardrail and the number of the nodes of the transverse boundary are taken as final node statistical information, the obstacle target is identified, otherwise, the guardrail direction of the obstacle target is represented as the abscissa direction, and the list of the nodes of the longitudinal guardrail, the number of the nodes of the longitudinal guardrail and the number of the nodes of the longitudinal boundary are taken as final node statistical information, and the obstacle target is identified.

Through the steps 21-25, node statistics of each obstacle target is realized, whether each obstacle target is a guardrail or not is conveniently detected according to statistical information, and accuracy of guardrail detection is guaranteed.

To further ensure that the final node statistics are obtained after the lateral and longitudinal traversals are completed, in one example, after obtaining the lateral guardrail node list, the number of lateral guardrail nodes, and the number of lateral boundary nodes, further includes: determining the current traversal line number; correspondingly, after the longitudinal guardrail node list, the longitudinal guardrail node number and the longitudinal boundary node number are obtained, the method further comprises the following steps: determining the current traversal column number; judging whether the current traversal line number and the current traversal column number are equal to preset values, and if not, executing the step of determining the final node statistical information of the obstacle target.

When the transverse traversal is stopped (i.e., the width of the transverse point cloud block is not less than the preset guardrail width threshold), the transverse traversal can be considered to be completed, the current traversal line number can be determined at this time, and when the longitudinal traversal is stopped (i.e., the width of the longitudinal point cloud block is not less than the preset guardrail width threshold or all the longitudinal point cloud blocks are judged to be completed), the longitudinal traversal can be considered to be completed, the current traversal line number can be determined at this time, further, if the current traversal line number is not equal to the preset value, and the current traversal line number is not equal to the preset value, the transverse traversal and the longitudinal traversal can be determined to be completed, and at this time, the final node statistical information can be determined according to the comparison result of the number of the transverse guardrail nodes and the number of the longitudinal guardrail nodes. By the example, the final node statistical information can be prevented from being obtained under the condition that the transverse traversal and the longitudinal traversal are not completed, and the accuracy of the final node statistical information is ensured.

For example, taking the upper left object in fig. 2 as an example, the preset guard bar width threshold may be 3, first, the lateral traversal starts from row=6, when traversing to row=9, the width of the lateral point cloud block exceeds the preset guard bar width threshold, at this time, the current traversal line number may be determined to be 9, and the longitudinal traversal is turned to, then, the longitudinal traversal starts from col=4, and when col=4, the width of the longitudinal point cloud block exceeds the preset guard bar width threshold, at this time, the current traversal line number may be determined to be 4. And finally, judging whether the current traversal line number and the current traversal column number are equal to preset values (like-1), if not, indicating that the target has completed transverse traversal and longitudinal traversal, and obtaining a transverse guardrail node list, a transverse guardrail node number, a transverse boundary node number, a longitudinal guardrail node list, a longitudinal guardrail node number and a longitudinal boundary node number.

In the embodiment of the disclosure, in the process of traversing the obstacle target transversely, considering the situation that the widths of all the transverse point clouds are smaller than the preset guardrail width threshold value, namely the widths of the transverse point clouds are smaller than the preset guardrail width threshold value when traversing to the last transverse point clouds, the possibility that the guardrail direction of the obstacle target is the ordinate direction is extremely high, and the obstacle target can not be traversed longitudinally any more.

For example, optionally, the method provided by the embodiment of the present disclosure further includes: if the widths of all the transverse point cloud blocks are smaller than the preset guardrail width threshold value, determining final node statistical information of the obstacle targets according to the transverse guardrail node list, the number of the transverse guardrail nodes and the number of the transverse boundary nodes.

That is, if the widths of all the horizontal point cloud blocks are smaller than the preset guardrail width threshold, the horizontal guardrail node list, the horizontal guardrail node number and the horizontal boundary node number can be directly used as final node statistical information, longitudinal traversal is not needed, and node statistical efficiency is improved.

For example, fig. 4 is a schematic diagram of a traversal process in the embodiment of the disclosure, as shown in fig. 4, for each obstacle target, whether the search direction (search_dir, where the preset value may be 0, and 0 represents the transverse traversal, and 1 represents the longitudinal traversal) is 0 may be firstly determined, if yes, the transverse traversal is required, further, whether the width of each transverse point cloud block is smaller than a preset guardrail width threshold value is determined, if the width is smaller than the preset guardrail width threshold value, the guardrail nodes are recorded and the number of guardrail nodes is updated, further, whether the guardrail nodes are located at the boundary is determined, if yes, the boundary nodes are recorded and the number of boundary nodes is updated, further, whether the number of guardrail nodes (denoted as pos_cnt) is equal to the number of transverse nodes (denoted as row_cnt), that is, the number of transverse point cloud blocks included in the point cloud blocks corresponding to the obstacle target), if no, the transverse traversal is continued, and if no, the transverse traversal is required, the traversing is continued, and the obtained guardrail node list, the number of guardrail nodes and the final number of boundary nodes are used as statistical information of the boundary nodes.

If the width is not less than the preset guardrail width threshold value, the search direction is adjusted (from 0 to 1), the jump line number (recorded as switch_row, namely the line of the transverse point cloud block which is currently traversed) is recorded, whether the jump line number is not equal to the preset value or not and whether the jump line number is not equal to the preset value or not is judged (namely the initial values of the jump line number and the jump line number can be set to be-1 or not), if yes, the traverse in the transverse and longitudinal directions is completed, if no, the continuous traverse is needed, and the step of judging whether the search direction is 0 or not is returned to the longitudinal traverse. The process of traversing longitudinally may refer to the process of traversing transversely, and will not be described in detail herein.

For example, fig. 5 is a schematic diagram of an obstacle target in the embodiment of the disclosure, taking fig. 5 as an example, assuming that a preset guard rail width threshold is 4, the number of transverse nodes row_cnt of the obstacle target is 3, the number of longitudinal nodes col_cnt is 8, and the number of hops switch_row and the number of hops switch_col take preset values of-1. Firstly, traversing from a 1 st horizontal point cloud block (curr_cow=3) of an obstacle target, stopping traversing when traversing to a 3 rd horizontal point cloud block (curr_cow=5), wherein the width of the horizontal point cloud block is not smaller than a preset guardrail width threshold value, recording the number pos_cnt=2 of guardrail nodes, recording switch_row=5 (namely, the 3 rd horizontal point cloud block is positioned on a 5 th row), judging whether the switch_row and the switch_col are not equal to-1, if yes, stopping traversing, and if not, jumping to longitudinal traversing.

Further, the traversal is started from the 1 st longitudinal point cloud block (curr_col=3) of the obstacle target, the traversal is started until the 8 th longitudinal point cloud block (curr_col=10), the width is smaller than a preset guardrail width threshold value, at the moment, the number of guardrail nodes pos_cnt=8, the switch_col=10 is recorded, whether pos_cnt is equal to col_cnt is judged, and if yes, the traversal is ended. Or judging whether the switch_row and the switch_col are not equal to-1, if so, stopping traversing.

S120, judging whether each obstacle target meets the length continuous condition based on the final node statistical information of each obstacle target, and judging whether the obstacle target meets the height continuous condition based on the final node statistical information of the obstacle target aiming at the obstacle target which does not meet the length continuous condition.

Wherein the length continuation condition may be used to determine whether the obstacle targets are continuous in their guardrail direction, and the height continuation condition is used to determine whether the obstacle targets are continuous in height.

In a specific embodiment, determining whether each obstacle target satisfies the length continuation condition based on the final node statistics of each obstacle target includes: for each obstacle target, determining the distance between two guardrail nodes furthest from each other based on a guardrail node list in the final node statistical information of the obstacle target, and determining the distance as the maximum length of the obstacle target; if the maximum length of the obstacle target is greater than the preset length threshold, determining that the obstacle target meets the length continuous condition.

Specifically, the global coordinate and the maximum height of each guardrail node can be recorded in the guardrail node list, the distance between two guardrail nodes farthest from each guardrail node can be obtained through the guardrail node list, and the distance is taken as the maximum length of an obstacle target.

Further, if the maximum length is greater than the preset length threshold, it indicates that the obstacle target is continuous in the guardrail direction thereof, and the length continuous condition is satisfied. By the method, the length continuity of each obstacle target is accurately judged, and the accuracy of the identified candidate guardrail targets is ensured.

In the embodiment of the disclosure, for the obstacle target meeting the length continuous condition, it can be determined as a candidate guardrail target, whether the obstacle target meets the height continuous condition is not required to be judged, and for the obstacle target not meeting the length continuous condition, in order to avoid the condition that the guardrail point cloud is missed due to shielding or other factor missing, whether the guardrail meets the height continuous condition can be continuously judged by considering that the guardrail has the characteristic of being high besides the length continuous condition, and if the height continuous condition is met, the obstacle target can be determined as the candidate guardrail target.

In a specific embodiment, for an obstacle target that does not satisfy the length continuity condition, determining whether the obstacle target satisfies the height continuity condition based on final node statistics of the obstacle target includes:

aiming at the obstacle targets which do not meet the length continuous condition, judging whether the number of guardrail nodes in the final node statistical information of the obstacle targets is larger than a preset first number and whether the number of boundary nodes is smaller than a preset second number;

if yes, determining the maximum height of the guardrail nodes and the maximum height difference between the guardrail nodes based on the guardrail node list in the final node statistical information of the obstacle targets, judging whether the maximum height is smaller than a preset maximum height threshold value or not, and if yes, determining that the obstacle targets meet the high continuous condition.

For the barrier target which does not meet the length continuous condition, whether the number of guardrail nodes of the barrier target is larger than the preset first number and whether the number of boundary nodes is smaller than the preset second number can be judged, if so, the barrier target is likely to be a guardrail, and further, whether the barrier target meets the height continuous condition can be judged by combining the maximum height of the guardrail nodes of the barrier target and the maximum height difference between the guardrail nodes.

Specifically, if the maximum height of the guardrail nodes is smaller than the preset maximum height threshold value, and the maximum height difference between the guardrail nodes is smaller than the preset maximum height difference threshold value, it can be determined that the obstacle targets are continuous in height, and the height continuity condition is satisfied, namely, the barrier targets have the height continuity. By the method, accurate judgment of the high continuity of each obstacle target is achieved, and accuracy of the identified candidate guardrail targets is guaranteed.

If the maximum height of the guardrail node of the obstacle target is not less than the preset maximum height threshold, or if the maximum height is not less than the preset maximum height threshold, it may be continuously determined whether the guardrail node satisfies the altitude continuous condition by combining the number of continuous nodes.

Optionally, the method provided by the embodiment of the present disclosure further includes: if the maximum height is not less than the preset maximum height threshold value, or the maximum height difference is not less than the preset maximum height difference threshold value, determining the height difference between adjacent guardrail nodes based on a guardrail node list in the final node statistical information of the obstacle target; determining the number of continuous nodes according to the height difference between all adjacent guardrail nodes, and determining that the barrier target meets the height continuous condition if the number of the continuous nodes is larger than the preset continuous number; the preset continuous number is the product of the number of guardrail nodes in the final node statistical information and the height consistency proportionality coefficient.

That is, the height difference between any adjacent two guardrail nodes may be continuously determined, and for each height difference, if the height difference is smaller than the set height difference threshold, the corresponding two guardrail nodes may be determined as continuous nodes, and the number of continuous nodes may be updated. Further, whether the number of continuous nodes is larger than the preset continuous number is judged, and if yes, it can be determined that the obstacle target meets the high continuous condition.

It should be noted that the preset continuous number may be determined according to a product of the number of guardrail nodes and the height consistency scaling factor, where a value range of the height consistency scaling factor may be [0,1]. For example, the height uniformity scaling factor may be a set value, or the height uniformity scaling factor may be determined according to a maximum length of the obstacle target, for example, the height uniformity scaling factor may be obtained according to coefficients respectively corresponding to lengths described in the calibration table.

By the method, whether the barrier targets meet the high continuous condition can be further judged by combining the number of the continuous nodes, so that the continuity between the adjacent nodes is checked, and the detection accuracy of the candidate guardrail targets is further ensured.

S130, determining the obstacle targets meeting the length continuous condition as candidate guardrail targets of the current frame.

Specifically, for an obstacle target that satisfies the length continuation condition, it may be determined as a candidate guardrail target of the current frame, and for an obstacle target that satisfies the height continuation condition, it may be determined as a candidate guardrail target of the current frame.

And S140, determining all formal guardrail targets of the current frame based on all candidate guardrail targets of the current frame and all formal guardrail targets of the history frame, and determining guardrail detection results of the current frame according to all formal guardrail targets of the current frame.

Specifically, each candidate guardrail target of the current frame and each formal guardrail target of the historical frame can be matched, each formal guardrail target of the current frame is determined according to the matching result, and then curve fitting is carried out on each formal guardrail target of the current frame, so that the guardrail detection result of the current frame is obtained.

In a specific embodiment, determining each formal guardrail target of the current frame based on each candidate guardrail target of the current frame and each formal guardrail target of the historical frame comprises the steps of:

step 31, projecting each candidate guardrail target of the current frame into a road network map to obtain the position of the candidate guardrail target on the road network map;

Step 32, performing position matching on each candidate guardrail target of the current frame and each formal guardrail target of the history frame, and determining each candidate guardrail target of the current frame successfully matched in position as each formal guardrail target of the current frame;

step 33, combining the candidate guardrail targets positioned in the same straight line aiming at the candidate guardrail targets of the current frame with failed position matching to obtain combined guardrail targets, and determining the combined guardrail targets with the lengths larger than a preset threshold value as the formal guardrail targets of the current frame.

Specifically, in the step 31, each candidate guardrail target may be projected onto the road network map, and the relative offset distance of each candidate guardrail target with respect to the road network line may be calculated, so as to obtain the position of each candidate guardrail target in the map, and each candidate guardrail target may be initialized according to the position, that is, the transformation of each candidate guardrail target from the vehicle body coordinate system into the map coordinate system may be implemented.

Further, each candidate guardrail target of the current frame is matched with each formal guardrail target of the history frame (previous k frames) in position, and the successfully matched candidate guardrail target can be determined as the formal guardrail target of the current frame.

For candidate guardrail targets which fail to match, all candidate guardrail targets positioned in the same straight line can be combined to obtain combined guardrail targets, and the combined guardrail targets with the lengths larger than a preset threshold value are determined to be all formal guardrail targets of the current frame.

By the mode, the screening of the formal guardrail targets is realized by combining the positions of the formal guardrail targets of the history frames and combining the candidate guardrail targets possibly connected in the same straight line, and the accuracy of the formal guardrail targets is ensured.

Considering that there may be a formal guardrail target that is not matched successfully for a long time in each formal guardrail target of the history frame, for example, in an area where a guardrail is driven out due to driving of a vehicle, in order to further ensure accuracy of guardrail detection in the current frame, in an optional implementation manner, the method provided by the embodiment of the disclosure further includes: updating the corresponding state identifier aiming at the candidate guardrail targets or the combined guardrail targets determined to be formal guardrail targets;

correspondingly, before the position matching is carried out on each candidate guardrail target of the current frame and each formal guardrail target of the historical frame, the method further comprises the following steps: among the formal guardrail targets in the history frame, the formal guardrail targets with the number of frames, of which the state identification is not updated, exceeding the set number of frames are removed.

That is, for the candidate guardrail targets successfully matched or the combined guardrail targets obtained after combination, the corresponding state identifiers can be updated. Wherein the status identifier is used to describe whether the guardrail targets match successfully or merge.

Furthermore, before the candidate guardrail targets of the current frame are matched with the formal guardrail targets of the history frame, the formal guardrail targets of which the frame numbers are not updated and exceed the set frame numbers can be removed from the formal guardrail targets of the history frame, so that the candidate guardrail targets of the current frame are prevented from being matched with guardrails which leave the field of view of the vehicle, and the accuracy of identifying the formal guardrail targets of the current frame is further ensured.

After obtaining each formal guardrail target of the current frame, fitting each formal guardrail target of the current frame by adopting a least square method to obtain a guardrail curve after fitting, and taking the guardrail curve as a guardrail detection result of the current frame.

In the embodiment of the disclosure, the guardrail target can be stably and effectively detected by adopting multiple judging modes such as a length continuous condition, a height continuous condition and the like, and meanwhile, the operation real-time requirement of the automatic driving embedded platform can be met. And moreover, a guardrail maximum width screening algorithm based on a two-dimensional planar grid map is provided, potential guardrail targets in all directions are searched through window cutting traversal, the guardrail targets are ensured to be effectively detected, and missed detection is not easy to occur. In addition, a guardrail screening algorithm combining the height and length continuity of guardrail targets is provided, a single threshold value is not relied on, and the judgment result of the guardrail targets is more accurate and more robust.

And the detected candidate guardrail targets are projected onto the map through coordinate system transformation on the basis of the existing map by utilizing a mode of matching with the map, and the positions of the guardrails are further checked, so that the accuracy of guardrail detection results is ensured.

According to the guardrail detection method provided by the embodiment of the disclosure, at least one obstacle target is determined by acquiring the point cloud of the current frame, final node statistical information of the obstacle targets is determined for each obstacle target through the corresponding point cloud block of each obstacle target, whether the obstacle targets meet the length continuous condition is judged according to the final node statistical information, whether the obstacle targets which do not meet the length continuous condition meet the height continuous condition is judged according to the final node statistical information, the obstacle targets meeting the length continuous condition and the obstacle targets meeting the height continuous condition are determined to be candidate guardrail targets of the current frame, and each formal guardrail target of the current frame is determined by combining with each formal guardrail target of a history frame, so that guardrail detection results of the current frame are determined, guardrail detection based on laser radar point cloud is realized.

In addition, the method provided by the embodiment of the disclosure can accurately detect the guardrail target through various feature checks, and meanwhile, compared with a data-driven deep learning method, the detection speed is higher, the occupied amount of calculation resources is less, the collection of marking data is not needed, and the cost is saved; the guardrail detection result can be subjected to strong verification based on the existing map data, and the robustness of guardrail detection is greatly improved.

Fig. 6 is a schematic structural view of a guardrail detection device according to an embodiment of the present disclosure. As shown in fig. 6: the device comprises: a node statistics module 610, a continuation determination module 620, a candidate determination module 630, and a formal determination goal 640, wherein:

the node statistics module 610 is configured to obtain a current frame point cloud, determine at least one obstacle target in the current frame point cloud, and determine, for each obstacle target, final node statistics information of the obstacle target based on a point cloud block corresponding to the obstacle target;

a continuous judging module 620, configured to judge whether each of the obstacle targets satisfies a length continuous condition based on final node statistics information of each of the obstacle targets, and judge whether each of the obstacle targets satisfies a height continuous condition based on final node statistics information of the obstacle targets for obstacle targets that do not satisfy the length continuous condition;

A candidate determining module 630, configured to determine, as candidate guard rail targets of the current frame, an obstacle target that satisfies the length continuation condition and an obstacle target that satisfies the height continuation condition;

and the formal determination target 640 is configured to determine each formal guardrail target of the current frame based on each candidate guardrail target of the current frame and each formal guardrail target of the history frame, and determine a guardrail detection result of the current frame according to each formal guardrail target of the current frame.

The guardrail detection device provided by the embodiment of the disclosure can execute steps in the guardrail detection method provided by the embodiment of the disclosure, and has the execution steps and beneficial effects, which are not repeated here.

Fig. 7 is a schematic structural diagram of an electronic device in an embodiment of the disclosure. Referring now in particular to fig. 7, a schematic diagram of an electronic device 500 suitable for use in implementing embodiments of the present disclosure is shown. The electronic device shown in fig. 7 is merely an example and should not be construed to limit the functionality and scope of use of the disclosed embodiments.

As shown in fig. 7, an electronic device 500 may include a processing means (e.g., a central processor, a graphics processor, etc.) 501 that may perform various suitable actions and processes to implement the methods of embodiments as described in the present disclosure according to a program stored in a Read Only Memory (ROM) 502 or a program loaded from a storage means 508 into a Random Access Memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic apparatus 500 are also stored. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.

In particular, according to embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a non-transitory computer readable medium, the computer program comprising program code for performing the method shown in the flowchart, thereby implementing the guardrail detection method as described above. In such an embodiment, the computer program may be downloaded and installed from a network via the communication means 509, or from the storage means 508, or from the ROM 502. The above-described functions defined in the methods of the embodiments of the present disclosure are performed when the computer program is executed by the processing device 501.

It should be noted that the computer readable medium described in the present disclosure may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this disclosure, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present disclosure, however, the computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, fiber optic cables, RF (radio frequency), and the like, or any suitable combination of the foregoing.

The computer readable medium may be contained in the electronic device; or may exist alone without being incorporated into the electronic device. The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to perform the steps of any of the embodiments described above.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

Scheme 1, a guardrail detection method, the method includes:

Solution 2, the method according to solution 1, the determining at least one obstacle target in the current frame point cloud, includes:

Screening the current frame point cloud to obtain a point cloud positioned in the region of interest in the current frame point cloud;

removing the vehicle body point cloud, the ground point cloud and the suspended point cloud in the region of interest based on preset vehicle body size parameters, a minimum height threshold and a maximum height threshold to obtain obstacle point cloud;

dividing the grid map of the region of interest to obtain a corresponding grid map, and determining the state of each grid according to the corresponding obstacle point of each grid in the obstacle point cloud, wherein the state of each grid is an occupied state or a non-occupied state;

at least one obstacle target is determined based on the status of each of the grids.

Scheme 3, the method according to scheme 2,

for each obstacle target, determining final node statistics of the obstacle targets based on point clouds corresponding to the obstacle targets, including:

for each obstacle target, determining each transverse point cloud block corresponding to the obstacle target in the abscissa direction based on each grid corresponding to the obstacle target, wherein the abscissa direction is the direction perpendicular to the vehicle head, and the transverse point cloud blocks are each row of the point cloud blocks of the obstacle target;

Starting from a first transverse point cloud block in the transverse point cloud blocks, judging whether transverse guardrail nodes and transverse boundary nodes exist or not according to the width of the transverse point cloud blocks and the boundary of the guardrail detection area until the width of the transverse point cloud blocks is not smaller than a preset guardrail width threshold value, and obtaining a transverse guardrail node list, the number of transverse guardrail nodes and the number of transverse boundary nodes;

determining each longitudinal point cloud block corresponding to the obstacle target in the ordinate direction based on each grid corresponding to the obstacle target, wherein the ordinate direction is the head direction;

starting from a first longitudinal point cloud block in the longitudinal point cloud blocks, judging whether longitudinal guardrail nodes and longitudinal boundary nodes exist according to the width of the longitudinal point cloud blocks and the boundary of a guardrail detection area until the width of the longitudinal point cloud blocks is not smaller than a preset guardrail width threshold value or all longitudinal point cloud blocks are judged to be finished, and obtaining a longitudinal guardrail node list, the number of longitudinal guardrail nodes and the number of longitudinal boundary nodes;

if the number of the transverse guardrail nodes is larger than the number of the longitudinal guardrail nodes, determining final node statistical information of the obstacle target according to the transverse guardrail node list, the number of the transverse guardrail nodes and the number of the transverse boundary nodes, and if not, determining final node statistical information of the obstacle target according to the longitudinal guardrail node list, the number of the longitudinal guardrail nodes and the number of the longitudinal boundary nodes.

The method according to claim 4, wherein starting from a first horizontal point cloud block in each horizontal point cloud block, judging whether there are horizontal guardrail nodes and horizontal boundary nodes according to the width of the horizontal point cloud block and the boundary of the guardrail detection area until the width of the horizontal point cloud block is not less than a preset guardrail width threshold value, and obtaining a horizontal guardrail node list, the number of horizontal guardrail nodes and the number of horizontal boundary nodes, including:

if yes, determining the center point of the current point cloud block as a transverse guardrail node, updating the number of the transverse guardrail nodes, and determining the transverse guardrail node as a transverse boundary node under the condition that the transverse guardrail node is positioned on the boundary of the guardrail detection area, and updating the number of the transverse boundary nodes;

and taking the next transverse point cloud block of the current point cloud block as a new current point cloud block, and returning to execute the step of traversing each grid in the current point cloud block along the abscissa direction until the width of the transverse point cloud block is not smaller than a preset guardrail width threshold value, so as to obtain a transverse guardrail node list, the number of transverse guardrail nodes and the number of transverse boundary nodes.

Scheme 5, the method according to scheme 3, the method further comprising:

and if the widths of all the transverse point cloud blocks are smaller than the preset guardrail width threshold, determining final node statistical information of the obstacle target according to the transverse guardrail node list, the transverse guardrail node number and the transverse boundary node number.

Solution 6, according to the method of solution 3, after the obtaining the list of lateral guardrail nodes, the number of lateral guardrail nodes, and the number of lateral boundary nodes, further includes:

determining the current traversal line number;

correspondingly, after the longitudinal guardrail node list, the longitudinal guardrail node number and the longitudinal boundary node number are obtained, the method further comprises the following steps:

determining the current traversal column number;

judging whether the current traversal line number and the current traversal column number are equal to preset values, and if not, executing the step of determining the final node statistical information of the obstacle target.

The method according to claim 7, wherein the determining whether each of the obstacle targets satisfies a length continuation condition based on the final node statistics of each of the obstacle targets, includes:

for each obstacle target, determining a distance between two guardrail nodes furthest from each other based on a guardrail node list in final node statistics of the obstacle targets, and determining the distance as a maximum length of the obstacle targets;

And if the maximum length of the obstacle target is greater than the preset length threshold, determining that the obstacle target meets the length continuous condition.

In an aspect 8, according to the method of aspect 3, for an obstacle target that does not satisfy the length continuation condition, determining whether the obstacle target satisfies a height continuation condition based on final node statistical information of the obstacle target includes:

aiming at the obstacle targets which do not meet the continuous length condition, judging whether the number of guardrail nodes in the final node statistical information of the obstacle targets is larger than a preset first number and whether the number of boundary nodes is smaller than a preset second number;

if yes, determining the maximum height of the guardrail nodes and the maximum height difference between the guardrail nodes based on the guardrail node list in the final node statistical information of the obstacle targets, judging whether the maximum height is smaller than a preset maximum height threshold value or not, and if yes, determining that the obstacle targets meet the height continuous condition.

Solution 9, the method according to solution 8, the method further comprising:

if the maximum height is not smaller than the preset maximum height threshold value or the maximum height difference is not smaller than the preset maximum height difference threshold value, determining the height difference between adjacent guardrail nodes based on a guardrail node list in final node statistical information of the obstacle target;

Determining the number of continuous nodes according to the height difference between all adjacent guardrail nodes, and if the number of the continuous nodes is larger than the preset continuous number, determining that the obstacle target meets the height continuous condition;

the preset continuous number is the product of the number of guardrail nodes in the final node statistical information and the height consistency proportionality coefficient.

The method according to claim 10, wherein the determining each formal guardrail target of the current frame based on each candidate guardrail target of the current frame and each formal guardrail target of the history frame includes:

for each candidate guardrail target of the current frame, projecting the candidate guardrail target into a road network map to obtain the position of the candidate guardrail target on the road network map;

performing position matching on each candidate guardrail target of the current frame and each formal guardrail target of the history frame, and determining each candidate guardrail target of the current frame successfully matched in position as each formal guardrail target of the current frame;

and combining the candidate guardrail targets positioned in the same straight line aiming at the candidate guardrail targets of the current frame with failed position matching to obtain combined guardrail targets, and determining the combined guardrail targets with the lengths larger than a preset threshold value as the formal guardrail targets of the current frame.

Solution 11, the method according to solution 10, the method further comprising:

updating the corresponding state identifier aiming at the candidate guardrail targets or the combined guardrail targets determined to be formal guardrail targets;

correspondingly, before the matching of the positions of the candidate guardrail targets of the current frame and the formal guardrail targets of the historical frame, the method further comprises:

among the formal guardrail targets in the history frame, the formal guardrail targets with the number of frames, of which the state identification is not updated, exceeding the set number of frames are removed.

Scheme 12, a guardrail detection device, comprising:

Scheme 13, an electronic device, the electronic device comprising:

one or more processors;

a storage means for storing one or more programs;

the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the methods of any of aspects 1-11.

Scheme 14, a computer readable storage medium having stored thereon a computer program which when executed by a processor implements the method of any of the schemes 1-11.

The foregoing description is only of the preferred embodiments of the present disclosure and description of the principles of the technology being employed. It will be appreciated by persons skilled in the art that the scope of the disclosure referred to in this disclosure is not limited to the specific combinations of features described above, but also covers other embodiments which may be formed by any combination of features described above or equivalents thereof without departing from the spirit of the disclosure. Such as those described above, are mutually substituted with the technical features having similar functions disclosed in the present disclosure (but not limited thereto).

Claims

1. A method of guardrail detection, the method comprising:

2. The method of claim 1, wherein the determining at least one obstacle target in the current frame point cloud comprises:

3. The method of claim 2, wherein the determining, for each of the obstacle targets, final node statistics of the obstacle targets based on the point clouds corresponding to the obstacle targets comprises:

4. The method of claim 3, wherein said determining whether there are lateral guardrail nodes and lateral border nodes from the width of the lateral point cloud and the boundary of the guardrail detection area starting from the first lateral point cloud of each lateral point cloud until the width of the lateral point cloud is not less than a preset guardrail width threshold value, to obtain a lateral guardrail node list, a number of lateral guardrail nodes, and a number of lateral border nodes, includes:

5. A method according to claim 3, characterized in that the method further comprises:

6. The method of claim 3, further comprising, after said deriving the list of lateral guardrail nodes, the number of lateral guardrail nodes, and the number of lateral border nodes:

determining the current traversal line number;

determining the current traversal column number;

7. The method of claim 3, wherein the determining whether each of the obstacle targets satisfies a length continuation condition based on final node statistics of each of the obstacle targets comprises:

8. A guardrail detection device, comprising:

9. An electronic device, the electronic device comprising:

one or more processors;

a storage means for storing one or more programs;

the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of any of claims 1-7.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method according to any of claims 1-7.