A scene degradation processing method and device, electronic equipment and storage medium

Abstract

The application discloses a scene degradation processing method and device, equipment and storage medium, relates to the technical field of robot autonomous positioning, and comprises the following steps: acquiring a sequential point cloud set of a current frame, and extracting a first line feature set from the sequential point cloud set by using an incremental algorithm; acquiring an a priori point cloud map, and performing feature extraction on the a priori point cloud map by using a search incremental algorithm to obtain a second line feature set; calculating the inclination angle of each first line feature in the first line feature set, and judging whether the scene is degraded according to the inclination angle; when the scene is degraded, each first line feature in the first line feature set is matched with each second line feature in the second line feature set, and pose correction is performed according to the matching result. The application combines the incremental algorithm and the search incremental algorithm, so that the pose can be solved from the point cloud in the case of scene degradation.

Description

Technical Field

[0001] This application relates to the field of robot autonomous positioning technology, and in particular to a method, apparatus, electronic device and storage medium for handling scene degradation. Background Technology

[0002] With the development of science and technology, the application of mobile robots is becoming more and more widespread, and robot localization technology is the focus of technology. Simultaneous Localization and Mapping (SLAM) technology is mainly used to solve the problem of robot localization and map building in unknown environments. At present, the demand for AGV robots in China is strong, and their application in manufacturing, logistics, inspection and other industries is becoming more and more widespread and the demand is stable. However, the requirements for intelligent automation of products are also getting higher and higher. At the same time, the arrival of the 5G era has prompted AGV robots to once again explode rapidly.

[0003] In existing technologies, 2D laser SLAM is widely used in AGVs, mainly to calculate point cloud line features. However, in cases of scene degradation, such as narrow passages or long walls, the vehicle cannot solve for the pose from the point cloud. Furthermore, the original incremental method is only applicable to cases where the direction and point correspond one-to-one at each angle. For existing point cloud maps, the incremental method cannot be used simply. Summary of the Invention

[0004] This application provides a method for handling scene degradation, which aims to solve the problem in the prior art that the pose cannot be solved from the point cloud when the scene is degraded.

[0005] To achieve the above objectives, this application adopts the following technical solution:

[0006] This application provides a method for handling scene degradation, comprising the following steps:

[0007] Obtain the sequential point cloud set of the current frame, and use an incremental algorithm to extract the first line feature set from the sequential point cloud set;

[0008] A prior point cloud map is obtained, and a second line feature set is obtained by performing feature extraction on the prior point cloud map using an incremental search algorithm.

[0009] Calculate the tilt angle of each first line feature in the first line feature set, and determine whether the scene is degraded based on the tilt angle;

[0010] When scene degradation occurs, each first-line feature in the first-line feature set is matched with each second-line feature in the second-line feature set, and pose correction is performed based on the matching results.

[0011] Preferably, the step of extracting the first line feature set from the sequential point cloud set using an incremental algorithm includes:

[0012] The first point set is determined according to the minimum number of fitted points and the point cloud order, and the first line segment is fitted according to the point cloud data in the first point set. The point cloud data in the first point set is arranged according to the point cloud order.

[0013] Calculate the distance from each sequential point cloud in the current first point set to the first line segment. If any distance is greater than a first set threshold, remove the first point cloud in the current first point set and determine whether the sequential point cloud set has been traversed. If not, add a new point cloud to the current point set according to the point cloud order and refit the first line segment.

[0014] If all distances are not greater than the first set threshold, then determine whether the first line segment meets the shortest line length requirement. If so, add its first line feature to the first line feature set and remove all point clouds in the current point set. Determine whether the sequential point cloud set has been traversed. If not, add a new point cloud to the current first point set according to the point cloud order and refit the first line segment.

[0015] Preferably, the step of using an incremental search algorithm to extract features from the prior point cloud map to obtain a second line feature set includes:

[0016] The prior point cloud map is divided into several fixed grids, and each grid stores multiple unordered point clouds;

[0017] The current second point set is determined based on the minimum number of fitted points and the minimum nearest neighbor points, and the second line segment is fitted based on the point cloud data in the current second point set;

[0018] Calculate the distance from each unordered point cloud in the current second point set to the second line segment. If any distance is greater than the second set threshold, remove the first point cloud in the current second point set and determine whether the prior point cloud map has been traversed. If not, add a new point cloud to the current second point set according to the least nearest point and refit the second line segment.

[0019] If all distances are not greater than the second set threshold, then determine whether the second line segment meets the shortest line length requirement. If so, save its second line feature and remove all point clouds in the current second point set. Determine whether the prior point cloud map has been traversed. If not, add a new point cloud to the current second point set according to the least nearest point and refit the second line segment.

[0020] After the traversal is complete, calculate the distance, angle, and overlap rate between all the saved second-line features, and remove all second-line features that do not meet the distance threshold, angle threshold, and overlap rate threshold.

[0021] Preferably, determining whether the scene is degraded based on the tilt angle includes:

[0022] Calculate the absolute value of the tilt angle difference between each adjacent first line feature, and determine whether each absolute value is less than the third set threshold. If so, the corresponding line features are parallel, and scene degradation occurs.

[0023] Preferably, the step of matching each first line feature in the first line feature set with each second line feature in the second line feature set includes:

[0024] The third line feature set is obtained by transforming all the first line features in the first line feature set to the global coordinate system using the two-dimensional pose formula.

[0025] Calculate the line angle difference, origin distance difference, and line coincidence degree between each third-line feature in the third-line feature set and each second-line feature in the second-line feature set;

[0026] The second line feature that matches each third line feature is determined based on the line angle difference, the origin distance difference, and the line overlap.

[0027] Preferably, the pose correction based on the matching result includes:

[0028] When the matching rate between the third line feature set and the second line feature set is greater than the matching threshold, the angle parameters and position parameters in the two-dimensional pose formula are corrected according to the line angle difference and the origin distance difference, and all first line features in the first line feature set are transformed to the global coordinate system using the corrected two-dimensional pose formula.

[0029] When the matching rate between the third line feature set and the second line feature set is less than the matching threshold, an integrator is used to calculate the pose of each first line feature.

[0030] Preferably, the method further includes:

[0031] When the scene is not degraded, radar matching is used to calculate the pose of each first-line feature.

[0032] A scene degradation processing device, comprising:

[0033] The first extraction module is used to obtain the sequential point cloud set of the current frame and extract the first line feature set from the sequential point cloud set using an incremental algorithm.

[0034] The second extraction module is used to obtain a prior point cloud map and use an incremental search algorithm to extract features from the prior point cloud map to obtain a second line feature set.

[0035] The judgment module is used to calculate the tilt angle of each first line feature in the first line feature set, and to determine whether the scene is degraded based on the tilt angle.

[0036] The matching module is used to match each first line feature in the first line feature set with each second line feature in the second line feature set when scene degradation occurs, and to perform pose correction based on the matching results.

[0037] An electronic device includes a memory and a processor, the memory being used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement a scene degradation processing method as described in any one of the preceding descriptions.

[0038] A computer-readable storage medium storing a computer program that, when executed by a computer, implements a scene degradation processing method as described in any one of the above descriptions.

[0039] The present invention has the following beneficial effects:

[0040] This application uses an incremental search algorithm to extract global line features from an established point cloud map, which ensures the continuity of point traversal by searching for neighboring points on the map through ordered containers. At the same time, by combining incremental algorithms with the incremental search algorithm, the pose can still be solved from the point cloud even in the case of scene degradation. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a flowchart of a scene degradation handling method in this application;

[0043] Figure 2 This is a diagram illustrating the failure of PL-ICP pose solving due to scene degradation.

[0044] Figure 3 This is a flowchart of the first-line feature set extraction process in this application;

[0045] Figure 4 This is a diagram showing the extraction effect of the first-line feature set in this application;

[0046] Figure 5 This is an image showing the effect of an incremental algorithm for extracting map line features.

[0047] Figure 6 This is a diagram showing the extraction effect of the second-line feature set in this application;

[0048] Figure 7 This is a diagram showing the pose correction result in this application;

[0049] Figure 8 This is a schematic diagram of a scene degradation processing device in this application;

[0050] Figure 9 This is a schematic diagram of an electronic device that implements a scene degradation processing method according to this application. Detailed Implementation

[0051] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0052] The terms “first,” “second,” etc., used in the claims and description of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate. This is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion so that a process, method, system, product, or apparatus that comprises a series of units is not necessarily limited to those units, but may include other units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0053] like Figure 1 As shown, this embodiment provides a method for processing scene degradation, including the following steps:

[0054] S110. Obtain the sequential point cloud set of the current frame, and use an incremental algorithm to extract the first line feature set from the sequential point cloud set.

[0055] S120. Obtain a priori point cloud map and use an incremental search algorithm to extract features from the priori point cloud map to obtain a second line feature set.

[0056] S130. Calculate the tilt angle of each first line feature in the first line feature set, and determine whether the scene is degraded based on the tilt angle.

[0057] S140. When scene degradation occurs, each first line feature in the first line feature set is matched with each second line feature in the second line feature set, and pose correction is performed based on the matching results.

[0058] This embodiment provides a scene degradation processing method that uses an incremental algorithm and a search-incremental algorithm to extract features from the sequential point cloud set and the prior point cloud map of the current frame, respectively, to obtain the point cloud line feature set (i.e., the first line feature set) and the map line feature set (i.e., the second line feature set) of the current frame. The sequential point cloud set of the current frame refers to the point cloud set transmitted by the radar in the current frame in sequence. l(a,b,c) is a two-dimensional line, and ax+by+c=0, where x is the two-dimensional pose abscissa and y is the two-dimensional pose ordinate. The first line feature is represented as l. n i (a n ,b n ,c n ), where i represents the point cloud of the i-th frame, n represents the n-th line feature in the coordinate system of the i-th frame point cloud, and the second line feature is represented as l. m w (a m ,b m ,c m Let w represent the world coordinate system and m represent the m-th line feature in the world coordinate system. Then, based on the tilt angle of each first line feature in the first line feature set, it is determined whether the scene is degraded. If the scene is degraded, feature matching is performed between the first line features in the first line feature set and the second line features in the second line feature set. If the matching result exceeds the matching threshold, the first line features are corrected using two degrees of freedom. In this embodiment, the global line features of the established point cloud map are obtained through an incremental search algorithm, ensuring the continuity of point traversal.

[0059] Furthermore, the current first point set is determined according to the minimum number of fitted points and the point cloud order, and the first line segment is fitted according to the point cloud data in the current first point set. The point cloud data in the current first point set are arranged according to the point cloud order.

[0060] Calculate the distance from each sequential point cloud in the current first point set to the first line segment. If any distance is greater than a first set threshold, remove the first point cloud in the current first point set and determine whether the sequential point cloud set has been traversed. If not, add a new point cloud to the current point set according to the point cloud order and refit the first line segment.

[0061] If all distances are not greater than the first set threshold, then determine whether the first line segment meets the shortest line length requirement. If so, add its first line feature to the first line feature set and remove all point clouds in the current point set. Determine whether the sequential point cloud set has been traversed. If not, add a new point cloud to the current first point set according to the point cloud order and refit the first line segment.

[0062] like Figure 2As shown in the figure, if there is a 20m long channel during the car's downward travel, the pose calculation will fail when using the point cloud library iterative nearest point algorithm (PL-ICP) to match the point cloud with the map. In this case, the car's pose estimation is provided by the IMU and odometry, which cannot eliminate the cumulative error of the trajectory. If the two sensors are not fully calibrated, the car will deviate from the predetermined trajectory shortly after the radar point cloud matching fails. In the figure, the red and blue point clouds are the map and the current frame point cloud, respectively. The blue arrow indicates the car's driving direction. The green circle indicates that the radar point cloud matching has failed in the long channel degradation scenario, resulting in a large pose error. At this time, the correct position of the car should be near the blue circle.

[0063] In this embodiment, an incremental algorithm is used to extract the line features of the sequential point cloud set in the current frame, such as... Figure 3 As shown, specifically, the current first point set is determined according to the minimum number of fitted points and the point cloud order. The minimum number of fitted points is the minimum number of points required to form a line, which can be selected according to actual needs. In this embodiment, the minimum number of fitted points is 7. "According to the point cloud order" means selecting the point cloud data corresponding to the current first point set based on the sequential order of the point clouds in the current frame. A fitted point count of 7 means selecting the first 7 point cloud data from the current frame's sequential point cloud set as the current first point set. A straight line is fitted based on these 7 point cloud data to obtain the first line segment. The distance from each of these 7 point clouds to the first line segment is calculated. If the distance of any one of them exceeds a first threshold, the first point cloud in the current first point set is removed. Then, it is determined whether all point clouds in the current frame's sequential point cloud set have been traversed. If not, the next point cloud in the current frame's sequential point cloud set is selected according to the point cloud order. The first point set is the 8th point in the current frame's sequential point cloud set. Line fitting continues. If all distances do not exceed the first threshold, it is determined whether the first line segment meets the shortest line length requirement. If not, all points in the current first point set are removed, and it is checked whether all points in the current frame's sequential point cloud set have been traversed. If not, 7 more points are selected in sequence as the current first point set, and line fitting is performed. If the first line segment meets the shortest line length requirement, its first line feature is added to the first line feature set. Then, all points in the current first point set are removed, and it is checked whether all points in the current frame's sequential point cloud set have been traversed. If not, 7 more points are selected in sequence as the current first point set, and line fitting is performed. If all points in the current frame's sequential point cloud set have been traversed, the first line feature set is directly output. The incremental algorithm in this embodiment is suitable for point clouds emitted sequentially by radar. Each frame of data has exactly one point in each angle direction. The time complexity of this algorithm is O(n), enabling fast feature extraction with the following results: Figure 4 As shown, the meanings of each mark in the diagram are as follows: Figure 3 same.

[0064] Using formula The tilt angle of each first-line feature in the first-line feature set is calculated. If the absolute value of the difference between the tilt angles of two adjacent first-line features is less than a third set threshold, the first-line features are considered to be parallel, resulting in scene degradation. In this embodiment, the third set threshold is set to 10°.

[0065] Furthermore, the prior point cloud map is divided into several fixed grids, and each grid stores multiple unordered point clouds;

[0066] The current second point set is determined based on the minimum number of fitted points and the minimum nearest neighbor points, and the second line segment is fitted based on the point cloud data in the current second point set;

[0067] Calculate the distance from each unordered point cloud in the current second point set to the second line segment. If any distance is greater than the second set threshold, remove the first point cloud in the current second point set and determine whether the prior point cloud map has been traversed. If not, add a new point cloud to the current second point set according to the least nearest point and refit the second line segment.

[0068] If all distances are not greater than the second set threshold, then determine whether the second line segment meets the shortest line length requirement. If so, save its second line feature and remove all point clouds in the current second point set. Determine whether the prior point cloud map has been traversed. If not, add a new point cloud to the current second point set according to the least nearest point and refit the second line segment.

[0069] After the traversal is complete, calculate the distance, angle, and overlap rate between all the saved second-line features, and remove all second-line features that do not meet the distance threshold, angle threshold, and overlap rate threshold.

[0070] like Figure 5As shown, since point cloud data in the map often contains multiple points in various angle directions, and all point clouds have already been re-divided according to cell order, incremental algorithms are not suitable. To address this, this embodiment uses a search-based incremental algorithm to extract global line features from the map. Specifically, a priori map is first loaded, and the point cloud is stored in a grid of fixed size. In this embodiment, the grid resolution is 5cm, allowing for rapid location of each grid cell based on grid coordinates. The initial line points are not selected sequentially, but rather by continuously searching for neighboring points in the grid map from the first point to select the next line point, updating parameters, and saving the second line feature, until all points on the priori map have been traversed. The traversal process is the same as described above. The incremental algorithm for extracting line features from the sequential point cloud set of the current frame is basically the same, except that the selection method of the point cloud added to the set is different each time, which will not be elaborated here. Next, the extracted second line features are merged and removed. Specifically, the distance, angle, and overlap rate between the second line features are judged to meet the threshold requirements. In this embodiment, the distance threshold is 5cm, the angle threshold is ±(5-10) degrees, and the overlap rate threshold is 80%. When the distance is less than 5cm, the angle is less than ±(5-10) degrees, and the overlap rate is higher than 80%, the second line features are merged. Finally, the second line feature set is output, thereby retaining the most representative second line features within a certain range as much as possible. The extraction effect is as follows. Figure 6 As shown, blue represents map point clouds, and pink represents extracted map line features, i.e., second-line features.

[0071] Furthermore, the three-dimensional pose formula is used to transform all the first-line features in the first-line feature set to the global coordinate system to obtain the third-line feature set;

[0072] Calculate the line angle difference, origin distance difference, and line coincidence degree between each third-line feature in the third-line feature set and each second-line feature in the second-line feature set;

[0073] The second line feature that matches each third line feature is determined based on the line angle difference, the origin distance difference, and the line overlap.

[0074] After determining scene degradation based on the tilt angle of the first-line features, the third-line feature set is obtained by transforming all first-line features in the first-line feature set to the global coordinate system using a two-dimensional pose formula. wi (R(θ),t(x,y)) will be the first line feature l n i Transform to the global coordinate system to obtain the third line feature l m w ', that is, a m w′ x+b m w′ y+cm w′ =0, where θ is the two-dimensional pose angle, PC is the laser point cloud, and the two-dimensional line feature l m w :a m w x+b m w y+c m w =0, then calculate the line angle difference Δ between each third line feature and the second line feature. θ , distance difference from origin △ d And the line coincidence degree σ, where:

[0075]

[0076] abs() represents the absolute value of the value inside the parentheses.

[0077]

[0078] Where, p m w' and p ms w They are l m w' and l m w The starting point of the line, p me w′ and p me w They are l m w' and l m w The end point of the line.

[0079] Only lines that are close in angle, close in distance, and have a high degree of overlap are matched.

[0080] Furthermore, when the matching rate between the third line feature set and the second line feature set is greater than the matching threshold, the angle parameters and position parameters in the two-dimensional pose formula are corrected according to the line angle difference and the origin distance difference, and all first line features in the first line feature set are transformed to the global coordinate system using the corrected two-dimensional pose formula;

[0081] When the matching rate between the third line feature set and the second line feature set is less than the matching threshold, an integrator is used to calculate the pose of each first line feature.

[0082] When the overall matching rate between the third line feature set and the second line feature set exceeds the matching threshold, pose correction is performed. The matching rate is the number of successfully matched lines in the current frame divided by the number of bus lines in the current frame. In this embodiment, the matching threshold is 80%, meaning that angle correction is performed when the overall matching rate between the third line feature set and the second line feature set exceeds 80%. First, based on Δ... θ Correction T wi θ, such that the transformed l m w 'and l m w Parallelism is then performed, followed by distance correction, based on △ d Correction T wi x and y, such that l m w' With l m w Once the pose correction is completed, the trolley can pass through the channel normally without any deviation in distance or direction, as long as it is on a straight line. Figure 7 As shown, its control group is Figure 2 The green circle indicates that the car's pose is normal after passing through a long, degraded scene and can be successfully matched with the map point cloud. The other marked parts have the same meaning. If the matching rate does not exceed the matching threshold, the integrator is used to calculate the pose. Similarly, if no scene degradation occurs, the pose is directly calculated based on the radar matching.

[0083] like Figure 8 As shown, this embodiment provides a scene degradation processing device, including:

[0084] The first extraction module 10 is used to obtain the sequential point cloud set of the current frame and extract the first line feature set from the sequential point cloud set using an incremental algorithm.

[0085] The second extraction module 20 is used to obtain a prior point cloud map and use an incremental search algorithm to extract features from the prior point cloud map to obtain a second line feature set.

[0086] The judgment module 30 is used to calculate the tilt angle of each first line feature in the first line feature set, and to determine whether the scene is degraded based on the tilt angle.

[0087] The matching module 40 is used to match each first line feature in the first line feature set with each second line feature in the second line feature set when the scene degrades, and to perform pose correction based on the matching results.

[0088] One embodiment of the above-mentioned device may be as follows: a first extraction module 10 acquires a sequential point cloud set of the current frame and extracts a first line feature set from the sequential point cloud set using an incremental algorithm; a second extraction module 20 acquires a prior point cloud map and extracts features from the prior point cloud map using a search incremental algorithm to obtain a second line feature set; a judgment module 30 calculates the tilt angle of each first line feature in the first line feature set obtained by the first extraction module 10 and determines whether the scene is degraded based on the tilt angle; when the scene is degraded, the matching module 40 matches each first line feature in the first line feature set obtained by the first extraction module 10 with each second line feature in the second line feature set obtained by the second extraction module 20, and performs pose correction based on the matching result.

[0089] like Figure 9 As shown, an electronic device includes a memory 901 and a processor 902. The memory 901 is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor 902 to implement the above-described scene degradation processing method.

[0090] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic device described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0091] A computer-readable storage medium storing a computer program that, when executed by a computer, implements a scene degradation processing method as described above.

[0092] For example, a computer program can be divided into one or more modules / units. One or more modules / units are stored in memory 901 and executed by processor 902. Data I / O interface transmission is completed by input interface 905 and output interface 906 to complete the present invention. One or more modules / units can be a series of computer program instruction segments capable of performing specific functions. The instruction segments are used to describe the execution process of the computer program in the computer device.

[0093] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device may include, but is not limited to, a memory 901 and a processor 902. Those skilled in the art will understand that this embodiment is merely an example of a computer device and does not constitute a limitation on the computer device. It may include more or fewer components, or a combination of certain components, or different components. For example, the computer device may also include an input device 907, a network access device, a bus, etc.

[0094] The processor 902 can be a Central Processing Unit (CPU), or other general-purpose processors 902, digital signal processors 902 (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor 902 can be a microprocessor 902, or any conventional processor 902, etc.

[0095] The memory 901 can be an internal storage unit of a computer device, such as a hard drive or RAM. The memory 901 can also be an external storage device of a computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory 901 can include both internal and external storage units. The memory 901 is used to store computer programs and other programs and data required by the computer device. The memory 901 can also be used for temporary storage in the output device 908. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM) 903, random access memory (RAM) 904, discs, or optical discs.

[0096] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions within the technical scope disclosed in the present invention should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for handling scene degradation, characterized in that, Includes the following steps: Obtain the sequential point cloud set of the current frame, and use an incremental algorithm to extract the first line feature set from the sequential point cloud set; A prior point cloud map is obtained, and a second line feature set is obtained by performing feature extraction on the prior point cloud map using an incremental search algorithm. Calculate the tilt angle of each first line feature in the first line feature set, and determine whether the scene is degraded based on the tilt angle, including: calculating the absolute value of the difference between the tilt angles of each adjacent first line feature, and determining whether each absolute value is less than a third set threshold. If so, the corresponding line features are parallel, and scene degradation occurs. When the scene degrades, the three-dimensional pose formula is used to transform all the first-line features in the first-line feature set to the global coordinate system to obtain the third-line feature set. Calculate the line angle difference, origin distance difference, and line coincidence degree between each third-line feature in the third-line feature set and each second-line feature in the second-line feature set; The second line features that match each third line feature are determined based on the line angle difference, the origin distance difference, and the line coincidence, and pose correction is performed based on the matching results. The step of extracting the first line feature set from the sequential point cloud set using an incremental algorithm includes: The first point set is determined according to the minimum number of fitted points and the point cloud order, and the first line segment is fitted according to the point cloud data in the first point set. The point cloud data in the first point set is arranged according to the point cloud order. Calculate the distance from each sequential point cloud in the current first point set to the first line segment. If any distance is greater than a first set threshold, remove the first point cloud in the current first point set and determine whether the sequential point cloud set has been traversed. If not, add a new point cloud to the current point set according to the point cloud order and refit the first line segment. If all distances are not greater than the first set threshold, then determine whether the first line segment meets the shortest line length requirement. If so, add its first line feature to the first line feature set and remove all point clouds in the current point set. Determine whether the sequential point cloud set has been traversed. If not, add a new point cloud to the current first point set according to the point cloud order and refit the first line segment. The second feature set obtained by using an incremental search algorithm to extract features from the prior point cloud map includes: The prior point cloud map is divided into several fixed grids, and each grid stores multiple unordered point clouds; The current second point set is determined based on the minimum number of fitted points and the minimum nearest neighbor points, and the second line segment is fitted based on the point cloud data in the current second point set; Calculate the distance from each unordered point cloud in the current second point set to the second line segment. If any distance is greater than the second set threshold, remove the first point cloud in the current second point set and determine whether the prior point cloud map has been traversed. If not, add a new point cloud to the current second point set according to the least nearest point and refit the second line segment. If all distances are not greater than the second set threshold, then determine whether the second line segment meets the shortest line length requirement. If so, save its second line feature and remove all point clouds in the current second point set. Determine whether the prior point cloud map has been traversed. If not, add a new point cloud to the current second point set according to the least nearest point and refit the second line segment. After the traversal is complete, calculate the distance, angle, and overlap rate between all the saved second-line features, and remove all second-line features that do not meet the distance threshold, angle threshold, and overlap rate threshold.

2. The method for processing scene degradation according to claim 1, characterized in that, The pose correction based on the matching results includes: When the matching rate between the third line feature set and the second line feature set is greater than the matching threshold, the angle parameters and position parameters in the two-dimensional pose formula are corrected according to the line angle difference and the origin distance difference, and all first line features in the first line feature set are transformed to the global coordinate system using the corrected two-dimensional pose formula. When the matching rate between the third line feature set and the second line feature set is less than the matching threshold, an integrator is used to calculate the pose of each first line feature.

3. The method for processing scene degradation according to claim 2, characterized in that, The method further includes: when the scene is not degraded, using radar matching to calculate the pose of each first-line feature.

4. A scene degradation processing device, characterized in that, include: The first extraction module is used to obtain the sequential point cloud set of the current frame and extract the first line feature set from the sequential point cloud set using an incremental algorithm. The second extraction module is used to obtain a prior point cloud map and use an incremental search algorithm to extract features from the prior point cloud map to obtain a second line feature set. The judgment module is used to calculate the tilt angle of each first line feature in the first line feature set, and to determine whether the scene is degraded based on the tilt angle. This includes: calculating the absolute value of the difference between the tilt angles of each adjacent first line feature, and determining whether each absolute value is less than a third set threshold. If so, the corresponding line features are parallel, and scene degradation occurs. The matching module is used to transform all the first-line features in the first-line feature set to the global coordinate system to obtain the third-line feature set when the scene degrades. Calculate the line angle difference, origin distance difference, and line coincidence degree between each third-line feature in the third-line feature set and each second-line feature in the second-line feature set; The second line features that match each third line feature are determined based on the line angle difference, the origin distance difference, and the line coincidence, and pose correction is performed based on the matching results. The step of extracting the first line feature set from the sequential point cloud set using an incremental algorithm includes: The first point set is determined according to the minimum number of fitted points and the point cloud order, and the first line segment is fitted according to the point cloud data in the first point set. The point cloud data in the first point set is arranged according to the point cloud order. Calculate the distance from each sequential point cloud in the current first point set to the first line segment. If any distance is greater than a first set threshold, remove the first point cloud in the current first point set and determine whether the sequential point cloud set has been traversed. If not, add a new point cloud to the current point set according to the point cloud order and refit the first line segment. If all distances are not greater than the first set threshold, then determine whether the first line segment meets the shortest line length requirement. If so, add its first line feature to the first line feature set and remove all point clouds in the current point set. Determine whether the sequential point cloud set has been traversed. If not, add a new point cloud to the current first point set according to the point cloud order and refit the first line segment. The second feature set obtained by using an incremental search algorithm to extract features from the prior point cloud map includes: The prior point cloud map is divided into several fixed grids, and each grid stores multiple unordered point clouds; The current second point set is determined based on the minimum number of fitted points and the minimum nearest neighbor points, and the second line segment is fitted based on the point cloud data in the current second point set; Calculate the distance from each unordered point cloud in the current second point set to the second line segment. If any distance is greater than the second set threshold, remove the first point cloud in the current second point set and determine whether the prior point cloud map has been traversed. If not, add a new point cloud to the current second point set according to the least nearest point and refit the second line segment. If all distances are not greater than the second set threshold, then determine whether the second line segment meets the shortest line length requirement. If so, save its second line feature and remove all point clouds in the current second point set. Determine whether the prior point cloud map has been traversed. If not, add a new point cloud to the current second point set according to the least nearest point and refit the second line segment. After the traversal is complete, calculate the distance, angle, and overlap rate between all the saved second-line features, and remove all second-line features that do not meet the distance threshold, angle threshold, and overlap rate threshold.

5. An electronic device, characterized in that, The system includes a memory and a processor, the memory being used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement a scene degradation processing method as described in any one of claims 1 to 3.

6. A computer-readable storage medium storing a computer program, characterized in that, The computer program enables the computer to implement a scene degradation processing method as described in any one of claims 1 to 3 when executed.

Images