Method for determining shielded target based on mapping laser

Abstract

The invention relates to the technical field of data processing, in particular to a method for determining an occluded target based on mapping laser, and the method comprises the steps: collecting point cloud data, carrying out the clustering of all point cloud data, obtaining each similar point set, and obtaining the sparsity and regularity of the point cloud data in each similar point set; according to the sparsity and the regularity, obtaining the distribution characteristics of each similar point set, and obtaining the cohesion and the separation degree of each similar point set; obtaining the noise level of each similar point set according to the cohesion degree and the separation degree, obtaining the filtering radius of each similar point set according to the distribution characteristics and the noise level of each similar point set, and carrying out the filtering processing of the point cloud data in each similar point set, and obtaining the filtered point cloud data; and determining a shielding target according to the filtered point cloud data. The method improves the accuracy of a filtering result.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and more specifically, to a method for determining an obscured target based on mapping laser. Background Art

[0002] With the rapid development of intelligent logistics and automated warehousing, logistics robots have become an indispensable part of modern logistics systems. Logistics robots are mainly used for tasks such as handling, sorting, stacking and distribution of goods. They usually need to navigate and perform tasks efficiently in complex warehousing environments. In this process, target detection and obstacle avoidance are key technologies for logistics robots to achieve efficient and safe operation. Traditional logistics robots rely on sensors (such as lidar, ultrasonic sensors, cameras, etc.) to perceive the surrounding environment. Among them, lidar, as a common sensor, helps robots understand their location and the location of the target by scanning the surrounding environment and generating a point cloud map. In actual applications, logistics robots often face the problem of target obstruction in complex environments. For example, in a warehouse, stacked goods, shelves and other objects may block the robot's view of the target, resulting in the logistics robot's inability to accurately identify the obscured target. This obstruction may cause task failure or delay, affecting logistics efficiency. Therefore, how to accurately detect and determine the obscured target is an important challenge in the application of logistics robots.

[0003] The current patent document with authorization announcement number CN109613552B discloses a method for detecting and analyzing vegetation-obstructed targets in TLS multi-echo point clouds. The invention uses a trained neural network to identify intermediate echoes, and finds the vegetation-obstructed canopy based on the identified intermediate echoes to determine the vegetation-obstructed area. The last echo is then used to filter out non-vegetation-obstructed targets in the vegetation-obstructed area to detect vegetation-obstructed targets. However, the step of preprocessing point cloud data in the patent document is only applicable to point cloud data processing of vegetation, and is not applicable to point cloud data of objects in the warehouse collected by the logistics robot. Therefore, it is necessary to preprocess the point cloud data of objects in the warehouse collected by the logistics robot and then use the method of detecting vegetation-obstructed targets in the patent document to determine the obscured targets of the logistics robot in the warehouse.

[0004] When collecting point cloud data of all objects in the warehouse, noise may be generated due to scanner errors, environmental interference and other factors. Therefore, the collected point cloud data in the warehouse needs to be filtered, and then the obstructed targets of the logistics robot in the warehouse are determined based on the filtered point cloud data; because when collecting point cloud data of all objects in the warehouse, the objects included may be shelves, aisles, ground, goods, etc., and the point cloud data of these different objects show different geometric shapes and distribution characteristics. Therefore, when using the radius filtering algorithm to filter all point cloud data, if a unified filtering radius is used, the filtering effect may be poor. If only a filtering radius is set for one point cloud data each time, although the effect is better, the amount of point cloud data is huge, which will consume a lot of computing resources. Summary of the invention

[0005] In order to solve the problem of poor filtering effect when the radius filtering algorithm uses a unified filtering radius to filter all point cloud data, the present invention proposes a method for determining an obstructed target based on mapping laser, which includes the following steps: Collect point cloud data; cluster all point cloud data to obtain each similar point set; use the ratio of the regularity of the point cloud data in each similar point set to the sparsity of the point cloud data in each similar point set as the distribution feature of the point cloud data in each similar point set; Get the noise level of each similar point set , Represents the noise level of the jth similar point set; represents the cohesion of the jth similarity point set; represents the separation degree of the jth similar point set; Get the filter radius of each similar point set , Represents the filter radius corresponding to the j-th similar point set; Represents the distribution characteristics of the jth similar point set; Represents the rounding symbol; The point cloud data in each similar point set is filtered according to the filtering radius of each similar point set to obtain filtered point cloud data; and the occluded target is determined according to the filtered point cloud data.

[0006] The innovation of the present invention lies in first dividing similar data in the point cloud data collected in the warehouse into one class, thereby clustering all point cloud data to obtain each similar point set; obtaining the filter radius of each similar point set according to the distribution characteristics of the point cloud data in each similar point set, and performing unified filtering processing on the point cloud data in each similar point set, which can significantly reduce the processing time, avoiding the problem of poor filtering effect of some point cloud data caused by the global use of the same filter radius for filtering, and possible over-filtering of other point cloud data, and also avoiding the problem of high computational consumption caused by setting a filter radius for one point cloud data.

[0007] Preferably, clustering all point cloud data to obtain each similar point set includes: Use the DBSCAN clustering algorithm to cluster all point cloud data to obtain each cluster and the cluster center of each cluster, and record each cluster as each similar point set; the cluster center of each cluster is the cluster center of each similar point set.

[0008] The innovation of the present invention lies in first dividing similar data in the point cloud data collected in the warehouse into one class, thereby clustering all point cloud data to obtain each similar point set; obtaining the filter radius of each similar point set according to the distribution characteristics of the point cloud data in each similar point set, and performing unified filtering processing on the point cloud data in each similar point set, which can significantly reduce the processing time, avoiding the problem of poor filtering effect of some point cloud data caused by the global use of the same filter radius for filtering, and possible over-filtering of other point cloud data, and also avoiding the problem of large computational complexity caused by setting a filter radius for one point cloud data.

[0009] Preferably, obtaining the regularity of the point cloud data in each similar point set includes: The modulus of the normal vector of each point cloud data in the j-th similar point set is obtained, and the variance of the modulus of the normal vectors of all point cloud data in the j-th similar point set is used as the regularity of the point cloud data in the j-th similar point set.

[0010] This facilitates the subsequent acquisition of the distribution characteristics of each similar point set.

[0011] Preferably, obtaining the sparsity of the point cloud data in each similar point set includes: The mean of the position coordinates of all point cloud data in each similar point set is taken as the center position of each similar point set; ; In the formula, Represents the sparsity of point cloud data in the jth similar point set; Represents the number of point cloud data in the jth similar point set; Represents the distance between the center position of the j-th similar point set and the r-th point cloud data in the j-th similar point set.

[0012] This facilitates the subsequent acquisition of the filter radius of each similar point set.

[0013] Preferably, obtaining the cohesion of the similarity point set includes: ; In the formula, represents the cohesion of the jth similarity point set; Represents the number of point cloud data in the jth similar point set; Represents the distance between the pth point cloud data in the jth similar point set and the cluster center of the jth similar point set; exp() represents an exponential function with a natural constant as the base.

[0014] Preferably, obtaining the separation degree of the similar point set includes: Obtain the distance between the cluster center of the j-th similar point set and the cluster center of each similar point set except the cluster center of the j-th similar point set, and record the similar point set corresponding to the minimum distance as the nearest neighbor similar point set of the j-th similar point set; The distance between the cluster center of the j-th similar point set and the cluster center of the nearest neighbor similar point set of the j-th similar point set is taken as the separation degree of the j-th similar point set.

[0015] This facilitates the subsequent acquisition of the filter radius of each similar point set.

[0016] Preferably, filtering the point cloud data in each similar point set according to the filter radius of each similar point set to obtain filtered point cloud data includes: According to the filter radius corresponding to each similar point set, the point cloud data in each similar point set is filtered using a radius filtering algorithm to obtain all filtered point cloud data.

[0017] Improved filtering effect.

[0018] Preferably, the collecting point cloud data includes: A lidar is installed on the top of the logistics robot to collect point cloud data on the surfaces of all objects in the warehouse.

[0019] The present invention has the following beneficial effects: the purpose of the present invention is to first divide similar data in the point cloud data collected in the warehouse into a class, thereby clustering all point cloud data to obtain each similar point set; obtaining the filter radius of each similar point set according to the distribution characteristics of the point cloud data in each similar point set, and performing unified filtering processing on the point cloud data in each similar point set, which can significantly reduce the processing time, avoiding the problem of poor filtering effect of some point cloud data caused by the global use of the same filter radius for filtering, and the possible over-filtering of other point cloud data, and also avoiding the problem of high computational consumption caused by setting a filter radius for one point cloud data. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 The present invention is a flowchart of a method for determining an obstructed target based on laser mapping according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0022] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0023] See also Figure 1 , which shows a flowchart of a method for determining an obstructed target based on mapping laser provided by an embodiment of the present invention, the method comprising the following steps: S001. Collect point cloud data.

[0024] In an embodiment of the present invention, a laser radar is installed on the top of the logistics robot, and point cloud data of the surfaces of all objects in the warehouse are collected by the laser radar; a sequence composed of all point cloud data is recorded as a point cloud data sequence.

[0025] S002. Cluster all point cloud data to obtain each similar point set, and obtain the sparsity and regularity of the point cloud data in each similar point set; based on the sparsity and regularity of the point cloud data in each similar point set, obtain the distribution characteristics of each similar point set.

[0026] It should be noted that when filtering the point cloud data of all objects in the warehouse collected by the logistics robot, the radius filter can well adapt to the point cloud data to form a compact point cloud distribution in the local space. By adjusting the size of the filter radius, it can effectively remove noise and outliers. At the same time, the calculation process of the radius filter is relatively simple, does not require complex mathematical operations or model training, and has high efficiency. When collecting the point cloud data of all objects in the warehouse, the objects included may include shelves, channels, ground, goods, etc., and the point cloud data of these different objects present different geometric shapes and distribution characteristics. Therefore, when using the radius filter algorithm to filter all point cloud data, collecting a unified filter radius may lead to poor filtering effect. If only one filter radius is set for one point cloud data each time for filtering, although the effect is better, the amount of point cloud data is huge, which will consume a lot of computing resources. Since similar data in point cloud data often comes from the same object, for example, goods on the same shelf, the ground or wall in the same channel, etc., these objects have high similarity in terms of geometric shape, spatial distribution and noise characteristics in the point cloud data. In other words, if the same filter radius is used to filter similar data in the point cloud data, it can avoid the poor filtering effect caused by the global use of the same filter radius for filtering, and avoid the problem of large amount of calculation caused by calculating a filter radius for each cloud data. Therefore, the present invention first clusters the point cloud data collected by the logistics robot, and regards the point cloud data in each clustering cluster as a similar point set, and then obtains the filter radius of each similar point set to filter all the point cloud data in each similar point set.

[0027] In an embodiment of the present invention, the DBSCAN clustering algorithm is used to cluster all point cloud data to obtain each cluster and the cluster center of each cluster, and each cluster is recorded as each similar point set to obtain the cluster center of each similar point set. It should be noted that the DBSCAN clustering algorithm is a prior art and will not be described in detail in the embodiment of the present invention.

[0028] It should be noted that in the point cloud data collected by the logistics robot, similar point sets usually correspond to the same object in the warehouse, such as shelves, goods, ground, etc., and these objects present specific geometric shapes and structures in the point cloud data. Therefore, by analyzing the distribution characteristics of the point cloud data in each similar point set, it is possible to provide an important basis for determining the filtering radius; if the position distribution of the point cloud data in any similar point set is more regular and the sparsity is lower, the possibility that the point cloud data in the similar point set belongs to the same object is greater. If the point cloud data in the similar point set belongs to the same object, the filtering radius of the similar point set needs to be smaller to provide fine filtering.

[0029] In the embodiment of the present invention, the mean value of the position coordinates of all point cloud data in each similar point set is taken as the center position of each similar point set; Get the sparsity of the point cloud data in each similarity point set: ; In the formula, Represents the sparsity of point cloud data in the jth similar point set; Represents the number of point cloud data in the jth similar point set; Represents the distance between the center position of the j-th similar point set and the r-th point cloud data in the j-th similar point set; The larger the value of is, the greater the sparsity of the point cloud data in the jth similar point set is.

[0030] Obtain the modulus of the normal vector of each point cloud data in the j-th similar point set, and use the variance of the modulus of the normal vectors of all point cloud data in the j-th similar point set as the regularity of the point cloud data in the j-th similar point set; The ratio of the regularity and sparsity of the point cloud data in each similar point set is used as the distribution feature of the point cloud data in each similar point set.

[0031] It should be noted that the higher the regularity and the smaller the sparsity of the point cloud data in the similar point set, the greater the possibility that the point cloud data in the similar point set belong to the same object, and the more obvious its distribution characteristics are.

[0032] S003, obtaining the cohesion and separation of each similar point set; and obtaining the noise level of each similar point set according to the cohesion and separation of each similar point set.

[0033] It should be noted that when collecting point cloud data of all objects in the warehouse, noise may be generated due to factors such as scanner errors and environmental interference. Therefore, there may be noise in the point cloud data of each similar point set obtained, and the noise level of each similar point set can directly affect the selection of the filter radius. In similar point sets with higher noise levels, a larger filter radius needs to be selected to remove noise; while in regional similar point sets with lower noise levels, a smaller filter radius can be selected to retain more point cloud information. Therefore, it is necessary to obtain the noise level of each similar point set, so as to facilitate the subsequent acquisition of the filter radius of each similar point set according to the noise level of each similar point set. It is known that similar point sets are obtained by clustering all point cloud data. Therefore, the noise level of each similar point set can be evaluated by the clustering effect of each similar point set. If the separation of each similar point set is higher and the cohesion is higher, the clustering effect of the similar point set is better and the noise level of the similar point set is lower.

[0034] In this embodiment of the present invention, the cohesion of each similar point set is obtained: ; In the formula, represents the cohesion of the jth similarity point set; Represents the number of point cloud data in the jth similar point set; Represents the distance between the pth point cloud data in the jth similar point set and the cluster center of the jth similar point set; exp() represents an exponential function with a natural constant as the base; the smaller the mean distance between all point cloud data in the jth similar point set and its cluster center, the higher the cohesion of the point cloud data in the jth similar point set.

[0035] Obtain the distance between the cluster center of the j-th similar point set and the cluster center of each similar point set except the cluster center of the j-th similar point set, and record the similar point set corresponding to the minimum distance as the nearest neighbor similar point set of the j-th similar point set; The distance between the cluster center of the j-th similar point set and the cluster center of the nearest neighbor similar point set of the j-th similar point set is taken as the separation degree of the j-th similar point set; Get the noise level for each set of similar points: ; In the formula, Represents the noise level of the jth similar point set; represents the cohesion of the jth similarity point set; Represents the separation degree of the jth similar point set; the higher the separation degree and the higher the cohesion degree of the jth similar point set, the better the clustering effect of the similar point set and the lower the noise level of the jth similar point set.

[0036] S004. Obtain a filter radius of each similar point set according to the distribution characteristics and noise level of each similar point set, filter the point cloud data in each similar point set according to the filter radius of each similar point set to obtain filtered point cloud data; determine the occluded target according to the filtered point cloud data.

[0037] It should be noted that, the higher the noise level of the known similar point set, the larger the filter radius needs to be selected to remove the noise. At the same time, the more obvious the distribution characteristics of the point cloud data in the similar point set, the greater the possibility that the point cloud data in the similar point set belong to the same object, and a smaller filter radius needs to be selected for fine processing. Therefore, the filter radius of the similar point set is comprehensively obtained based on the noise level of the similar point set and the distribution characteristics of the similar point set to avoid over-filtering or under-filtering caused by too large or too small a filter radius, so as to improve the filtering effect.

[0038] In this embodiment of the present invention, the filter radius corresponding to each similar point set is obtained: ; In the formula, Represents the filter radius corresponding to the j-th similar point set; Represents the noise level of the jth similar point set; Represents the distribution characteristics of the jth similar point set; Represents the rounding symbol; According to the filter radius corresponding to each similar point set, the point cloud data in each similar point set is filtered using the radius filtering algorithm to obtain all filtered point cloud data; The occluded target is determined based on the filtered point cloud data. The specific process of determining the occluded target is the public technology in the invention of "A detection and analysis method for vegetation occluded targets in TLS multi-echo point cloud" with announcement number "CN109613552B".

[0039] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for determining an obscured target based on mapping laser, characterized in that: include: Collect point cloud data; Cluster all point cloud data to obtain each similar point set; The ratio of the regularity of the point cloud data in each similar point set to the sparsity of the point cloud data in each similar point set is used as the distribution feature of the point cloud data in each similar point set; Get the noise level of each similar point set , Represents the noise level of the jth similar point set; represents the cohesion of the jth similarity point set; represents the separation degree of the jth similar point set; Get the filter radius of each similar point set , Represents the filter radius corresponding to the j-th similar point set; Represents the distribution characteristics of the jth similar point set; Represents the rounding symbol; Performing filtering processing on the point cloud data in each similar point set according to the filtering radius of each similar point set to obtain filtered point cloud data; The occluded target is determined based on the filtered point cloud data.

2. The method for determining an obscured target based on mapping laser according to claim 1, characterized in that: The above-mentioned clustering of all point cloud data to obtain each similar point set includes: Use the DBSCAN clustering algorithm to cluster all point cloud data to obtain each cluster and the cluster center of each cluster, and record each cluster as each similar point set; the cluster center of each cluster is the cluster center of each similar point set.

3. A method for determining an obscured target based on mapping laser according to claim 1, characterized in that: The acquisition of the regularity of the point cloud data in each similar point set includes: The modulus of the normal vector of each point cloud data in the j-th similar point set is obtained, and the variance of the modulus of the normal vectors of all point cloud data in the j-th similar point set is used as the regularity of the point cloud data in the j-th similar point set.

4. The method for determining an obscured target based on mapping laser according to claim 1, characterized in that: The acquisition of the sparsity of point cloud data in each similar point set includes: The mean of the position coordinates of all point cloud data in each similar point set is taken as the center position of each similar point set; ； In the formula, Represents the sparsity of point cloud data in the jth similar point set; Represents the number of point cloud data in the jth similar point set; Represents the distance between the center position of the j-th similar point set and the r-th point cloud data in the j-th similar point set.

5. The method for determining an obstructed target based on mapping laser according to claim 1, characterized in that: The acquisition of the cohesion of the similar point set includes: ； In the formula, represents the cohesion of the jth similarity point set; Represents the number of point cloud data in the jth similar point set; Represents the distance between the pth point cloud data in the jth similar point set and the cluster center of the jth similar point set; exp() represents an exponential function with a natural constant as the base.

6. The method for determining an obstructed target based on mapping laser according to claim 1, characterized in that: The acquisition of the separation degree of the similar point set includes: Obtain the distance between the cluster center of the j-th similar point set and the cluster center of each similar point set except the cluster center of the j-th similar point set, and record the similar point set corresponding to the minimum distance as the nearest neighbor similar point set of the j-th similar point set; The distance between the cluster center of the j-th similar point set and the cluster center of the nearest neighbor similar point set of the j-th similar point set is taken as the separation degree of the j-th similar point set.

7. The method for determining an obscured target based on mapping laser according to claim 1, characterized in that: The filtering process is performed on the point cloud data in each similar point set according to the filtering radius of each similar point set to obtain the filtered point cloud data, including: According to the filter radius corresponding to each similar point set, the point cloud data in each similar point set is filtered using a radius filtering algorithm to obtain all filtered point cloud data.

8. The method for determining an obscured target based on mapping laser according to claim 1, characterized in that: The collecting of point cloud data comprises: A lidar is installed on the top of the logistics robot to collect point cloud data on the surfaces of all objects in the warehouse.

Images