Beam transporting vehicle obstacle detection method and system based on laser radar point cloud clustering

Abstract

The embodiment of the invention provides a beam transporting vehicle obstacle detection method and system based on laser radar point cloud clustering, and the method comprises the steps: scanning the target direction of a beam transporting vehicle based on a laser radar, and obtaining the point cloud data of a target space region; obtaining target parameters Eps and MinPts on the basis of a genetic algorithm in combination with a clustering algorithm and a fitness function; performing optimization processing on the target parameters Eps and MinPts based on the point cloud density distribution characteristics; and on the basis of the optimized target parameters Eps and MinPts, clustering recognition calculation is performed by using a clustering algorithm to obtain obstacle information in the target space region, so that automatic generation of a neighborhood radius (Eps) parameter and a minimum point number (MinPts) parameter can be realized, the parameter accuracy is high, the clustering recognition precision is improved while the clustering recognition calculation rate is improved, and the method is suitable for large-scale popularization and application. And rapid and accurate obstacle detection is facilitated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of obstacle detection, and in particular to a girder transport vehicle obstacle detection method and system based on laser radar point cloud clustering. BACKGROUND

[0002] Girder transport vehicles, such as bridge erection girder transport vehicles, high-speed rail girder transport vehicles, etc., may face complex construction environments such as narrow roads, temporary supports, construction equipment, etc. when transporting large precast girders (therefore, an obstacle detection method is needed to ensure safe transportation. Currently, laser point cloud data of the obstacle detection area of the girder transport vehicle is clustered and recognized by a clustering algorithm to obtain obstacle information. However, in the existing clustering algorithm, the neighborhood radius (Eps) parameter and the minimum point number (MinPts) parameter need to be manually adjusted, which is time-consuming and has low accuracy, resulting in slow clustering recognition calculation speed and low clustering recognition accuracy, which is not conducive to fast and accurate obstacle detection. SUMMARY

[0003] The present application provides a girder transport vehicle obstacle detection method and system based on laser radar point cloud clustering, which can automatically generate the neighborhood radius (Eps) parameter and the minimum point number (MinPts) parameter, has high parameter accuracy, improves the clustering recognition calculation speed and the clustering recognition accuracy, and is conducive to fast and accurate obstacle detection.

[0004] In a first aspect, the present application provides a girder transport vehicle obstacle detection method based on laser radar point cloud clustering, comprising: Based on the target direction of the laser radar scanning the girder transport vehicle, point cloud data of a target space region is obtained; Based on a genetic algorithm combined with a clustering algorithm and a fitness function, target parameters Eps and MinPts are obtained; Based on the point cloud density distribution characteristics, the target parameters Eps and MinPts are optimized; Based on the optimized target parameters Eps and MinPts, a clustering algorithm is used for clustering recognition calculation to obtain obstacle information in the target space region.

[0005] In a second aspect, the present application provides a girder transport vehicle obstacle detection system based on laser radar point cloud clustering, comprising: A point cloud data acquisition module is configured to obtain point cloud data of a target space region based on the target direction of the laser radar scanning the girder transport vehicle; A target parameter determination module is configured to obtain target parameters Eps and MinPts based on a genetic algorithm combined with a clustering algorithm and a fitness function; The target parameter optimization module is configured to optimize the target parameters Eps and MinPts based on the point cloud density distribution characteristics. The clustering identification module is configured to perform clustering identification calculation based on the optimized target parameters Eps and MinPts by using a clustering algorithm, and obtain the obstacle information in the target space region.

[0006] In a third aspect, the present application provides an electronic device, which comprises a memory and a processor, the memory stores a computer program, and the processor implements the steps of the beam transport vehicle obstacle detection method based on laser radar point cloud clustering provided in the first aspect when executing the computer program.

[0007] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the beam transport vehicle obstacle detection method based on laser radar point cloud clustering provided in the first aspect when executed by a processor.

[0008] In a fifth aspect, the present application provides a computer program product, which comprises a computer program, and the computer program implements the steps of the beam transport vehicle obstacle detection method based on laser radar point cloud clustering provided in the first aspect when executed by a processor.

[0009] In the present application, the beam transport vehicle obstacle detection method based on laser radar point cloud clustering can automatically generate the neighborhood radius (Eps) parameter and the minimum point number (MinPts) parameter, has high parameter accuracy, improves the clustering identification calculation rate and the clustering identification accuracy, is conducive to fast and accurate obstacle detection, and is further conducive to the beam transport vehicle to perform obstacle avoidance processing in the target direction based on the obstacle detection information, and ensures the safe passage of the beam transport vehicle. BRIEF DESCRIPTION OF DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0011] In order to more completely understand the present application and its beneficial effects, the following will be described in conjunction with the drawings, wherein the same reference numerals in the following description represent the same parts.

[0012] Figure 1 A flowchart of a beam transport vehicle obstacle detection method based on laser radar point cloud clustering provided in the embodiments of the present application.

[0013] Figure 2A schematic diagram of a target space area of a beam transport vehicle and an obstacle provided in an embodiment of the present application.

[0014] Figure 3 A flowchart of step 200 provided in an embodiment of the present application.

[0015] Figure 4 A flowchart of step 230 provided in an embodiment of the present application.

[0016] Figure 5 A flowchart of step 240 provided in an embodiment of the present application.

[0017] Figure 6 Another flowchart of step 200 provided in an embodiment of the present application.

[0018] Figure 7 A schematic diagram of a result of clustering and identifying point cloud data of a target space area of a beam transport vehicle in an embodiment of the present application.

[0019] Figure 8 A structural schematic diagram of a beam transport vehicle obstacle detection system based on laser radar point cloud clustering provided in an embodiment of the present application.

[0020] Figure 9 A structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] The technical solutions of the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0022] Figure 1 A flowchart of a beam transport vehicle obstacle detection method based on laser radar point cloud clustering provided in an embodiment of the present application. Please refer to Figure 1 The beam transport vehicle obstacle detection based on laser radar point cloud clustering provided in the embodiments of the present application includes steps 100 to 400, which will be described in detail below.

[0023] Step 100, based on laser radar scanning a target direction of a beam transport vehicle, point cloud data of a target space area is obtained.

[0024] Step 200, based on a genetic algorithm and combined with a clustering algorithm and a fitness function, target parameters Eps and MinPts are obtained.

[0025] At step 300, the target parameters Eps and MinPts are optimized based on the point cloud density distribution characteristics.

[0026] At step 400, the clustering algorithm is used to perform clustering recognition calculation based on the optimized target parameters Eps and MinPts, and the obstacle information in the target space region is obtained.

[0027] In the embodiments of the present application, the beam transport vehicle obstacle detection method based on laser radar point cloud clustering can automatically generate the neighborhood radius (Eps) parameter and the minimum point number (MinPts) parameter, has high parameter accuracy, improves the clustering recognition calculation rate and the clustering recognition accuracy, is conducive to fast and accurate obstacle detection, and is further conducive to the beam transport vehicle to perform obstacle avoidance processing in the target direction based on the obstacle detection information, and ensures the safe passage of the beam transport vehicle.

[0028] The following describes Figure 1 When the beam transport vehicle obstacle detection processing based on laser radar point cloud clustering is performed in the embodiments, further optional specific embodiments are performed in each step.

[0029] At step 100, the point cloud data of the target space region is obtained based on the laser radar scanning the target direction of the beam transport vehicle.

[0030] In the present embodiment, the target direction can be the driving direction of the beam transport vehicle, such as at least one of the forward direction and the reverse direction. The target direction can also be a non-driving direction of the beam transport vehicle, such as a side direction, a bottom direction, specifically at least one of a left side direction, a right side direction, a left front side direction, a right front side direction, a left rear side direction, a right rear side direction, and a bottom direction. In this way, the beam transport vehicle can be detected in all directions.

[0031] Figure 2 A schematic diagram of the target space region and the obstacle of the beam transport vehicle provided in the embodiments of the present application is provided. Please refer to Figure 2 The target space region is a preset space range in the target direction of the beam transport vehicle, and the point cloud data of the target region is obtained based on the laser radar. The obstacle information of the target region is detected through the point cloud data. The position of the target space region is a space range set relative to the position of the beam transport vehicle. During the driving process of the beam transport vehicle, the three-dimensional coordinates of the target space region dynamically change with the movement of the beam transport vehicle. The target space region is a space range in which obstacle detection must be performed to ensure the safe passage of the beam transport vehicle. For example, for a beam transport vehicle with a width of 7.5 meters, when the target direction is the forward direction, the target space region can be set as a region with a length of 15 meters, a width of 7.6 meters, and the same height as the beam transport vehicle in front of the beam transport vehicle.

[0032] The point cloud is a collection of reflection points in a target space region scanned by a laser radar, and each point in the point cloud is each reflection point. The point cloud data includes a collection of data of each reflection point in the target space region after the laser radar scans the target space region, and the data of each reflection point includes the three-dimensional coordinates and the reflection intensity of the reflection point.

[0033] In some embodiments, based on the target direction of the girder transport vehicle scanned by the laser radar, the step of obtaining the point cloud data of the target space region can further include a preprocessing step. The preprocessing step includes: Step 510, obtaining the point cloud data of the target space region, using statistical filtering method to remove invalid point cloud points, using voxel filtering to reduce the amount of point cloud data, and obtaining first point cloud data; Step 520, using RANSAC algorithm to remove ground point cloud, and obtaining second point cloud data, which is the preprocessed point cloud data.

[0034] In step 510, the statistical filtering method is used to remove invalid point cloud points, including: calculating the average distance between each point cloud point in the point cloud and its nearest neighbor point, removing outliers with a distance exceeding a threshold value, such as 3 times the standard deviation, thereby removing invalid point cloud points in the point cloud data.

[0035] Using voxel filtering to reduce the amount of point cloud data includes: after removing invalid point cloud points in the point cloud data, dividing the point cloud into a voxel grid, and retaining a representative point cloud point in each voxel grid, such as a center point cloud point or a gravity point cloud point, which can significantly reduce the amount of point cloud data and maintain the shape feature, and obtain the first point cloud data.

[0036] In step 520, the step of removing ground point cloud using RANSAC algorithm includes: first, establishing a mathematical model, and marking the point cloud data points conforming to the mathematical model as ground points; then removing the ground point cloud data and retaining only the non-ground point cloud data as the second point cloud data output.

[0037] In this embodiment, preprocessing the point cloud data of the target space region can remove irrelevant point cloud point data in the point cloud data, reduce the amount of point cloud data, reduce the data processing amount of the clustering processing step, and further improve the calculation efficiency of the clustering identification step of the subsequent target parameter determination step.

[0038] In some embodiments, the girder transport vehicle obstacle detection method based on laser radar point cloud clustering can include steps 100, preprocessing step, step 200, step 300, and step 400 in sequence.

[0039] In some embodiments, after the step of obtaining point cloud data of the target space region based on the target direction of the girder transport vehicle scanned by the laser radar, a three-dimensional tree structure data construction step can be further included. The three-dimensional tree structure data construction step includes: Based on the dimension of the point cloud data, a three-dimensional tree structure data is constructed by a spatial partitioning algorithm to obtain the point cloud data of the three-dimensional tree structure.

[0040] For example, the spatial partitioning algorithm can be any one of kdtree, Octree, Rtree, and Ball Tree algorithm.

[0041] In this embodiment, the three-dimensional tree structure organizes unordered point cloud into a hierarchical structure through recursive spatial division, reduces the complexity of large-scale search from O(n) to O(log n), thereby realizing efficient neighbor query and spatial operation, and further accelerating the calculation efficiency of the subsequent target parameter determination step and clustering identification step.

[0042] In some embodiments, the girder transport vehicle obstacle detection method based on laser radar point cloud clustering can include the following steps in sequence: step 100, preprocessing step, three-dimensional tree structure data construction step, step 300, and step 400.

[0043] In other embodiments, the girder transport vehicle obstacle detection method based on laser radar point cloud clustering can include the following steps in sequence: step 100, three-dimensional tree structure data construction step, step 300, and step 400.

[0044] In some embodiments, after the step of obtaining point cloud data of the target space region based on the target direction of the girder transport vehicle scanned by the laser radar, a sampling step can be further included. The sampling step includes: Based on the density distribution characteristics of the point cloud data, the point cloud data is divided into multiple density regions; A sample is extracted from each density region according to a proportion to form a sample data set; The sample data set is:

[0045] wherein, represents taking the absolute value, is the sample data set, is the jth density region data set, is the sampling proportion of the jth density region data set, and j is a natural number from 1 to m.

[0046] In the embodiment, the point cloud data is sampled. The first aspect is to divide the point cloud data according to density, and select part of the point cloud data in each density division, so as to reduce the amount of point cloud data without destroying the density characteristics of the point cloud data, that is, to preserve the basic point cloud shape, thereby facilitating improving the calculation efficiency of the clustering recognition under the premise of ensuring the accuracy of the clustering recognition. The second aspect is to remove the sign and direction of the point cloud data by taking the absolute value, so as to ensure the integrity of the point cloud data, thereby facilitating ensuring the accuracy of the clustering recognition.

[0047] In some embodiments, the girder transport vehicle obstacle detection method based on laser radar point cloud clustering can sequentially include: step 100, a sampling step, step 300, and step 400.

[0048] In some embodiments, the girder transport vehicle obstacle detection method based on laser radar point cloud clustering can sequentially include: step 100, a sampling step, step 300, and step 400.

[0049] In some embodiments, the girder transport vehicle obstacle detection method based on laser radar point cloud clustering can sequentially include: step 100, a sampling step, step 300, and step 400.

[0050] In some embodiments, the girder transport vehicle obstacle detection method based on laser radar point cloud clustering can sequentially include: step 100, a sampling step, step 300, and step 400.

[0051] It can be understood that the point cloud density is the number of point clouds in a space region of the same size. In a space region of the same size, the high-density area has a large number of point clouds, which is generally a near region relative to the laser radar. The low-density area has a small number of point clouds, which is generally a far region relative to the laser radar.

[0052] It can be further understood that when the laser radar scans a closer object, the closer object can return a larger number of point clouds, and the larger number of returned point clouds results in a larger point cloud density, thereby forming a high-density area in the point cloud data. When the laser radar scans a farther object, the farther object can return a smaller number of point clouds, and the smaller number of returned point clouds results in a smaller point cloud density, thereby forming a low-density area in the point cloud data.

[0053] In some embodiments, in the divided plurality of density regions, the smaller the density region of the point cloud point density, the greater the corresponding sampling ratio. For example, through experiments, the point cloud data can be divided into high, medium and low density regions, and the sampling ratio of the low density region is at least 74%, the sampling ratio of the medium density region is at least 52%, and the sampling ratio of the high density region is at least 31%, so as to simultaneously consider the integrity of the data and the calculation efficiency of the subsequent target parameter determination step and the clustering identification step.

[0054] Please refer to Table 1, in example 1, the sampling ratios of the low, medium and high density regions are all 100%, in example 2, the sampling ratios of the low, medium and high density regions are all 50%, and in example 3, the sampling ratios of the low, medium and high density regions are 74%, 52% and 31% respectively.

[0055] Among them, the silhouette coefficient is used to represent the clustering identification accuracy, the higher the silhouette coefficient, the higher the clustering identification accuracy. The clustering identification time is the time required for the clustering identification calculation based on the method provided in the embodiments of the present application, the less the clustering identification time, the higher the calculation efficiency of the clustering identification. The cluster number identification is used to represent the accuracy of the clustering identification, 9 / 9 means that 9 obstacles are identified from the point cloud data actually having 9 obstacles, and 7 / 9 means that 7 obstacles are identified from the actual 9 obstacles. As can be seen from the comparison of examples 1, 2 and 3, the clustering identification accuracy and the calculation efficiency of the clustering identification of example 3 are the highest.

[0056] Table 1

[0057] Step 200, obtaining the target parameters Eps and MinPts based on the genetic algorithm combined with the clustering algorithm and the fitness function.

[0058] In the present embodiment, the clustering algorithm is DBSCAN clustering algorithm. The genetic algorithm (Genetic Algorithm, GA) is used in the embodiments of the present application. The fitness function used in the embodiments of the present application is the silhouette coefficient.

[0059] Eps is the neighborhood radius. The neighborhood is a circular region with a point cloud point in the point cloud data as the center. The radius of the circular region is the neighborhood radius.

[0060] MinPts is the minimum number of points. The minimum number of points (MinPts) represents the minimum number of neighborhood points required for a point cloud point in the point cloud data to be determined as a core point cloud point in the Eps neighborhood radius. When the neighborhood radius is Eps, the neighborhood points of a point, that is, the number of points within the neighborhood radius Eps around the point.

[0061] In this embodiment, point cloud data is processed based on a genetic algorithm combined with a clustering algorithm and a fitness function. This enables the automatic generation of the neighborhood radius (Eps) parameter and the minimum number of points (MinPts) parameter with high accuracy. This improves both the calculation speed and accuracy of clustering recognition, and further facilitates rapid and accurate obstacle detection.

[0062] Figure 3 This is a flowchart illustrating step 200 provided in an embodiment of this application. In some embodiments, please refer to... Figure 3 Step 200, which uses a genetic algorithm combined with a clustering algorithm and a fitness function to obtain the target parameters Eps and MinPts, includes: Step 210: Encode Eps and MinPts into chromosomes (Eps, MinPts), where Eps is the neighborhood radius and MinPts is the minimum number of points, and randomly generate n sets of parameter combinations as the initial population. Step 220: Calculate the offspring population based on the genetic algorithm; Step 230: Based on the clustering algorithm and fitness function calculation, obtain the fitness values ​​of the first generation population and the next generation population; Step 240: When the fitness function converges, the target parameters Eps and MinPts are obtained.

[0063] In this embodiment, the initial population is generated by randomly generating n sets of parameter combinations, which can be exemplarily represented as { , ), , ), ..., , The offspring population is generated from the initial population using a genetic algorithm. Each offspring population also represents n sets of parameter combinations, but the parameter values ​​are different. Based on the initial population, the offspring population is obtained through crossover, mutation, and iteration inherent in the genetic algorithm.

[0064] It is understandable that by encoding Eps and MinPts into chromosomes (Eps, MinPts), the final target parameters Eps and MinPts are the optimal parameters, which are the optimal neighborhood radius and the optimal minimum number of points.

[0065] Figure 4 This is a flowchart illustrating step 230 provided in the embodiments of this application.

[0066] In some embodiments, please refer to Figure 4 Step 230, which obtains the fitness values ​​of the first and second generation populations based on a clustering algorithm and a fitness function, includes: Step 231, based on the clustering algorithm, the calculation of each generation population in the initial population and the total population of offspring, the clustering results of each generation population are obtained; Step 232, based on the fitness function, the clustering results of each generation population are calculated respectively, and the fitness values of each generation population are obtained.

[0067] In this embodiment, the clustering results of each generation population include the first clustering result and the second clustering result. The first clustering result of each generation population is the average distance between the target point cloud points and other point cloud points in the point cloud cluster formed by the clustering of the generation population, and the second clustering result of each generation population is the minimum value of the average distance between the target point cloud points and other clusters (not this cluster) in the point cloud cluster formed by the clustering of the generation population, which is used to represent the separation degree of the target point cloud cluster and other point cloud clusters.

[0068] By bringing the first clustering result and the second clustering result of each generation population into the fitness function for calculation respectively, the fitness values of each generation population are obtained.

[0069] Figure 5 The flowchart of step 240 provided in the embodiments of the present application is shown.

[0070] In some embodiments, referring to Figure 5 When the fitness function converges, the step 240 of obtaining the target parameters Eps and MinPts includes: Step 241, calculating the absolute value of the difference between the fitness values of the adjacent two generation populations; Step 242, if the absolute value of the difference is greater than the preset convergence precision, continue to perform the genetic algorithm calculation step, that is, continue to perform step 220; if the absolute value of the difference is not greater than the preset convergence precision, end the genetic algorithm calculation step, that is, end step 220, and obtain the last generation population; Step 243, taking the last generation population as the optimal population, and taking the parameters of the optimal population as the target parameters Eps and MinPts.

[0071] In this embodiment, the absolute value of the difference between the fitness values of the adjacent two generation populations is compared with the preset convergence precision to determine whether to end the genetic algorithm calculation step, that is, to determine whether the optimal population is found. Finding the optimal population means finding the optimal parameters, that is, the target parameters Eps and MinPts.

[0072] In some embodiments, the expression of the fitness function is:

[0073] Wherein: is the fitness function; The average distance between the target point cloud points and other point cloud points in the point cloud cluster formed by clustering under the parameters corresponding to the i th generation population, used to represent the closeness of the target point cloud points and other point cloud points in the same point cloud cluster; The minimum value of the average distance between the target point cloud points and other clusters in the point cloud cluster formed by clustering under the parameters corresponding to the i th generation population, used to represent the separation degree of the target point cloud cluster and other point cloud clusters; The average distance between the k th point cloud point and other point cloud points in the point cloud cluster formed by clustering under the parameters corresponding to the i th generation population, used to represent the closeness of the k th point cloud point and other point cloud points in the same point cloud cluster; The minimum value of the average distance between the k th point cloud point and other clusters (not the current cluster) in the point cloud cluster formed by clustering under the parameters corresponding to the i th generation population, used to represent the separation degree of the k th point cloud cluster and other point cloud clusters; k is any point cloud point in the point cloud cluster, a natural number from 1 to p; p is the number of point clouds.

[0074] It can be understood that the parameters corresponding to the i th generation population, the parameter combination corresponding to the 1 st generation population { , ), , ), …, , } are obtained through the genetic algorithm.

[0075] It can be understood that the fitness function is the core link connecting parameter selection and clustering quality, used to quantify the advantages and disadvantages of clustering effect, and guide the search direction of the genetic algorithm. The number of point clouds refers to the total number of all points constituting the point cloud. After successful point cloud clustering, the point cloud points of the same obstacle are gathered together to form a point cloud cluster.

[0076] Figure 6 Another flowchart of step 200 provided in the embodiments of the present application.

[0077] In some embodiments, referring to Figure 6 , the step 200 of obtaining the target parameters Eps and MinPts based on the genetic algorithm combined with the clustering algorithm and the fitness function can include: Step 201, encode Eps and MinPts into chromosomes (Eps, MinPts), Eps is the neighborhood radius, MinPts is the minimum number of points, randomly generate n groups of parameter combinations as the 1 st generation parameters; Step 202, based on the clustering algorithm, calculate the n groups of parameter combinations, that is, the 1 st generation parameters, to obtain the clustering results and ; Step 203, based on the fitness function, calculate the clustering results and Step 204, based on the i-th generation parameters, the i+1-th generation parameters are obtained through the crossover, mutation and iteration of the genetic algorithm, i is a natural number greater than or equal to 1; ; Step 205, based on the clustering algorithm, the i+1-th generation parameters are calculated to obtain the clustering result and ; ; Step 206, based on the fitness function, the calculation result x i+1 is calculated to obtain the fitness value of the i+1-th generation parameters ; Step 207, when , return to step 204; when , enter step 208; Step 208, the i+1-th generation parameters are taken as the target parameters Eps and MinPts.

[0078] Wherein, indicates the fitness value of the i-th generation parameters; ε indicates the preset convergence precision, which can be set to 10 -3 or smaller value. , The meaning of is described above and will not be repeated here.

[0079] Step 300, based on the point cloud density distribution characteristics, the target parameters Eps and MinPts are optimized.

[0080] In some embodiments, the step of optimizing the target parameters Eps and MinPts based on the point cloud density distribution characteristics can include: Based on the distance between the target point cloud point and the beam transport vehicle, the point cloud data is divided into multiple sub-regions; The target parameters corresponding to each sub-region are calculated respectively; The calculation formula is:

[0081]

[0082] Wherein: is the Eps parameter generated after dynamic adjustment according to the distance d; MinPts is the MinPts parameter generated after dynamic adjustment according to the distance d; d is the distance between the target point cloud point and the beam transport vehicle; the target point cloud point is any point in each sub-region; indicates the basic distance threshold, that is, the target parameter Eps; This represents the base distance threshold, which is the target parameter MinPts; This represents the distance growth factor of Eps, used to characterize the rate at which Eps increases with distance d.

[0083] Understandable The value needs to be adjusted according to the specific scenario, for example The value ranges from 0.001 to 0.05. The larger the value, the larger the neighborhood radius.

[0084] It is understandable to divide point cloud data into multiple sub-regions. For example, for a beam transport vehicle that is 7.5 meters wide, when the target direction is forward, the target space area can be set as the area in front of the beam transport vehicle that is 15 meters long, 7.6 meters wide, and the same height as the beam transport vehicle. For example, the target space area can be divided into two sub-regions, 0-9m and 9-15m, and the point cloud data can be divided into two corresponding regions. Alternatively, the target space area can be divided into three sub-regions, 0-7m, 7-11m, and 11-15m, and the point cloud data can be divided into three corresponding regions.

[0085] Step 400: Based on the optimized target parameters Eps and MinPts, a clustering algorithm is used to perform clustering recognition calculations to obtain obstacle information within the target spatial region.

[0086] Figure 7 This is a schematic diagram illustrating the result of clustering and identifying the point cloud data of the target spatial region of the beam transport vehicle in an embodiment of this application.

[0087] Understandably, please refer to Figure 7 Point cloud clustering recognition refers to the process of segmenting unordered 3D point cloud data into multiple independent subsets, or point cloud clusters, where each subset represents a potential object or structure. The core objective of point cloud clustering recognition is to distinguish different objects in a scene, such as pedestrians, vehicles, and trees, providing basic information for subsequent obstacle recognition, classification, and obstacle avoidance.

[0088] Figure 8 This is a schematic diagram of the obstacle detection system for a beam transport vehicle based on lidar point cloud clustering provided in this embodiment of the application. Please refer to... Figure 8 The obstacle detection system for beam transport vehicles based on lidar point cloud clustering may include a point cloud data acquisition module 801, a target parameter calculation module 802, a target parameter optimization module 803, and a clustering recognition module 804. Among them: The point cloud data acquisition module 801 is used to acquire point cloud data of the target spatial area based on the target direction of the beam transport vehicle scanned by lidar.

[0089] The target parameter determination module 802 is configured to obtain the target parameters Eps and MinPts based on a genetic algorithm in combination with a clustering algorithm and a fitness function.

[0090] The target parameter optimization module 803 is configured to perform optimization processing on the target parameters Eps and MinPts based on the density distribution characteristics of the point cloud.

[0091] The clustering identification module 804 is configured to perform clustering identification calculation based on the target parameters Eps and MinPts after the optimization processing, and obtain the obstacle information in the target space region.

[0092] In actual applications, the system can be a terminal device or a chip applied to a terminal device. In the present application, the system can realize the functions of multiple units in the manner of software, hardware, or a combination of software and hardware, so that the system can perform the steps of the beam truck obstacle detection method based on laser radar point cloud clustering provided in any of the above embodiments. The technical effects of each technical solution of the system can refer to the technical effects of the corresponding technical solutions in the beam truck obstacle detection method based on laser radar point cloud clustering, which will not be described one by one herein.

[0093] Figure 9 FIG. 1 is a structural schematic diagram of an electronic device provided in an embodiment of the present application.

[0094] Based on the hardware implementation of each unit in the above system, the present embodiment further provides an electronic device, as shown in FIG. 1, which comprises a memory 910 and a processor 920. The memory 910 stores a computer program, and the processor 920 implements the steps of the beam truck obstacle detection method based on laser radar point cloud clustering provided in any of the above embodiments when executing the computer program. Figure 9

[0095] Of course, in actual applications, as shown in FIG. 1, each component in the electronic device 900 is coupled together through a bus system 930. It can be understood that the bus system 930 is used to realize the connection and communication between the components. The bus system 930 includes not only a data bus, but also a power bus, a control bus, and a status signal bus. However, for the purpose of clear illustration, all kinds of buses are marked as the bus system 930 in the figure. Figure 9

[0096] ​​In practical applications, the processor can be at least one of an application specific integrated circuit (ASIC), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, and a microprocessor. It can be understood that, for different devices, the electronic device used to implement the functions of the processor can also be other devices, and the embodiments of the present application do not make specific limitations.

[0097] The memory can be a volatile memory (such as a random access memory (RAM)), a non-volatile memory (such as a read-only memory (ROM), a flash memory, a hard disk (HDD) or a solid state disk (SSD)), or a combination of the above types of memories, and provides instructions and data to the processor.

[0098] The electronic device described in the embodiments of the present application can be a terminal device or a chip applied to a terminal device. The terminal device described in the embodiments of the present application can include a terminal device with a shooting function, such as a mobile phone, a tablet computer, a notebook computer, a palm computer, a personal digital assistant (PDA), a portable media player (PMP), a wearable device, a camera, and the like.

[0099] In the example embodiments, the embodiments of the present application also provide a computer readable storage medium, for example, a memory including a computer program, which can be executed by a processor of an electronic device to complete the steps of the foregoing method.

[0100] The embodiments of the present application also provide a computer program product including a computer program, which, when executed by a processor, implements the steps of any of the methods described in the embodiments of the present application.

[0101] Optionally, the computer program product can be applied to the electronic device in the embodiments of the present application, and the computer program instructions cause the computer to execute the corresponding processes realized by the electronic device in the various methods of the embodiments of the present application. For brevity, details are not repeated here.

[0102] The embodiments of the present application further provide a computer program.

[0103] Optionally, the computer program can be applied to the electronic device in the embodiments of the present application, and when the computer program runs on the computer, causes the computer to execute the corresponding processes realized by the electronic device in the various methods of the embodiments of the present application. For brevity, details are not repeated here.

[0104] It should be understood that in the embodiments of the present application, data related to user information and the like are involved, and when the embodiments of the present application are applied to specific products or technologies, the user's permission or consent needs to be obtained, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of countries and regions.

[0105] It should be understood that the terms used in the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "an" and "the" used in the present application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein means and includes any or all possible combinations of one or more associated listed items. In the present application, the expressions "have", "may have", "include" and "contain", or "may include" and "may contain" can be used herein to indicate the presence of the corresponding features (for example, elements such as numerical values, functions, operations or components), but do not exclude the presence of additional features.

[0106] It should be understood that although the terms first, second, third, etc. can be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other, and do not necessarily be used to describe a particular order or sequence. For example, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as the first information, without departing from the scope of the present application.

[0107] The technical solutions described in the embodiments of the present application can be combined arbitrarily without conflict.

[0108] In several embodiments provided in the present application, it should be understood that the disclosed methods, systems, apparatuses and devices can be implemented in other ways. The above-described embodiments are merely illustrative, for example, the division of units is merely a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed components can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0109] The units described above as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place or distributed on multiple network units; part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0110] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be realized in the form of hardware or in the form of hardware plus software functional unit.

[0111] The above is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for obstacle detection in a beam transport vehicle based on lidar point cloud clustering, characterized in that, include: The target direction of the beam transport vehicle is scanned by lidar to obtain point cloud data of the target spatial area; Based on the genetic algorithm combined with the clustering algorithm and fitness function, the target parameters Eps and MinPts are obtained; Based on the point cloud density distribution characteristics, the target parameters Eps and MinPts are optimized. Based on the optimized target parameters Eps and MinPts, a clustering algorithm is used to perform clustering recognition calculations to obtain obstacle information within the target spatial region.

2. The detection method according to claim 1, characterized in that, The steps for obtaining the target parameters Eps and MinPts based on a genetic algorithm combined with a clustering algorithm and a fitness function include: Eps and MinPts are encoded into chromosomes (Eps, MinPts), where Eps is the neighborhood radius and MinPts is the minimum number of points. n sets of parameter combinations are randomly generated as the initial population. The offspring population is obtained based on genetic algorithm calculations. The fitness values ​​of the first generation and the next generation population are obtained based on clustering algorithms and fitness function calculations. When the fitness function converges, the target parameters Eps and MinPts are obtained.

3. The obstacle detection method for beam transport vehicles based on lidar point cloud clustering according to claim 2, characterized in that, The step of obtaining the fitness values ​​of the initial and subsequent generations of the population based on clustering algorithms and fitness functions includes: Clustering algorithms are used to calculate the clustering results of each generation of the population in the first generation and the total population of descendants. Based on the fitness function, the clustering results of each generation of the population are calculated to obtain the fitness value of each generation of the population.

4. The obstacle detection method for beam transport vehicles based on lidar point cloud clustering according to claim 2, characterized in that, The step of obtaining the target parameters Eps and MinPts when the fitness function converges includes: Calculate the absolute value of the difference in fitness values ​​between two adjacent generations of the population; If the absolute value of the difference is greater than the preset convergence precision, the genetic algorithm calculation step continues; if the absolute value of the difference is not greater than the preset convergence precision, the genetic algorithm calculation step ends and the last generation population is obtained. The last generation of the population is taken as the optimal population, and the parameters of the optimal population are taken as the target parameters Eps and MinPts.

5. The obstacle detection method for beam transport vehicles based on lidar point cloud clustering according to claim 2, characterized in that, The expression for the fitness function is: in: The fitness function; is the average distance between target point cloud points and other point cloud points within the point cloud cluster formed by clustering under the parameters corresponding to the i-th generation population, used to characterize the closeness between target point cloud points and other point cloud points within the same point cloud cluster; The minimum average distance between the target point cloud points and other clusters within the point cloud cluster formed by clustering under the parameters corresponding to the i-th generation population is used to characterize the degree of separation between the target point cloud cluster and other point cloud clusters. is the average distance between the kth point cloud point and other point cloud points within the point cloud cluster formed by clustering under the parameters corresponding to the i-th generation population, used to characterize the closeness between the kth point cloud point and other point cloud points within the same point cloud cluster. The minimum average distance between the k-th point cloud point and other clusters within the point cloud cluster formed by clustering under the parameters corresponding to the i-th generation population is used to characterize the degree of separation between the k-th point cloud cluster and other point cloud clusters. k is any point cloud point within the point cloud cluster, a natural number from 1 to p; p represents the number of point clouds.

6. The obstacle detection method for beam transport vehicles based on lidar point cloud clustering according to claim 1, comprising: The step of optimizing the target parameters Eps and MinPts based on the point cloud density distribution characteristics includes: Based on the distance between the target point cloud points and the beam transport vehicle, the point cloud data is divided into multiple sub-regions; The target parameters for each sub-region are calculated separately. The calculation formula is: in: The Eps parameter is generated after dynamic adjustment based on distance d; MinPts These are the MinPts parameters generated after dynamic adjustment based on distance d; d represents the distance between the target point cloud point and the beam transport vehicle; the target point cloud point is any point within each sub-region. This represents the base distance threshold, which is the target parameter Eps; This represents the base distance threshold, which is the target parameter MinPts; This represents the distance growth factor of Eps, used to characterize the rate at which Eps increases with distance d.

7. The obstacle detection method for beam transport vehicles based on lidar point cloud clustering according to claim 1, characterized in that, After the step of obtaining point cloud data of the target spatial region based on scanning the target direction of the beam transport vehicle with lidar, the method further includes: For point cloud data of the target spatial region, invalid point cloud points are removed by statistical filtering and the amount of point cloud data is reduced by voxel filtering to obtain the first point cloud data. The RANSAC algorithm is used to remove ground point cloud data to obtain second point cloud data, which is the preprocessed point cloud data.

8. The obstacle detection method for beam transport vehicles based on lidar point cloud clustering according to claim 1, characterized in that, After the step of obtaining point cloud data of the target spatial region based on scanning the target direction of the beam transport vehicle with lidar, the method further includes: Based on the dimension of the point cloud data, a three-dimensional tree structure data is constructed using a spatial partitioning algorithm to obtain point cloud data with a three-dimensional tree structure.

9. The obstacle detection method for beam transport vehicles based on lidar point cloud clustering according to claim 1, characterized in that, After the step of obtaining point cloud data of the target spatial region based on scanning the target direction of the beam transport vehicle with lidar, the method further includes: Based on the density distribution characteristics of the point cloud data, the point cloud data is divided into multiple density regions; Samples are drawn proportionally from each density region to form a sample dataset; The sample dataset is as follows: in, This indicates taking the absolute value. For the sample dataset, For the j-th density region dataset, Let be the sampling ratio of the dataset for the j-th density region, where j is a natural number from 1 to m.

10. An obstacle detection system for a beam transport vehicle based on lidar point cloud clustering, characterized in that, include: The point cloud data acquisition module is used to acquire point cloud data of the target spatial area based on the target direction of the beam transport vehicle scanned by LiDAR; The target parameter determination module is used to obtain the target parameters Eps and MinPts based on a genetic algorithm combined with a clustering algorithm and a fitness function. The target parameter optimization module is used to optimize the target parameters Eps and MinPts based on the point cloud density distribution characteristics. The clustering and identification module is used to perform clustering and identification calculations based on the optimized target parameters Eps and MinPts, and to obtain obstacle information within the target spatial region.

Images