Robot positioning method in storage environment

Abstract

The invention relates to the technical field of data processing, and discloses a robot positioning method in a storage environment, and the method comprises the steps: determining a voxel index of an initial point cloud, obtaining voxel depth information based on the voxel index and the number of buckets in a preset hash table, carrying out the layering processing of a voxelization point cloud map according to the voxel depth information, and carrying out the positioning of a robot in a storage environment. And obtaining a plurality of sub-voxel maps, and positioning the robot in the storage environment corresponding to the target point cloud according to the plurality of sub-voxel maps and the target point cloud voxels. The voxel index of the initial point cloud is calculated, the initial point cloud is accurately divided, the number of storage buckets of the preset hash table is set, frequent capacity expansion and shrinkage of the hash table in the registration process are avoided, the voxelization point cloud map is divided into the multiple sub-voxel maps through hierarchical processing, the conflict and search time of the hash table are reduced, and the registration accuracy of the hash table is improved. And the point cloud data can be processed more meticulously, so that the robot positioning process in the storage environment is more accurate.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly to a method for robot positioning in a warehousing environment. Background Art

[0002] The positioning of mobile robots in a warehousing environment is crucial. In scenarios of continuous and rapid movement, classical algorithms often struggle to balance accuracy and real-time performance simultaneously. However, in the actual transportation in a warehousing scenario, high precision and real-time capabilities are essential for robots. Existing technologies have introduced point cloud registration techniques in robot positioning in a warehousing environment, aiming to align point cloud data obtained from different perspectives or at different times into a unified coordinate system, thereby generating a complete and accurate map or 3D model. To improve the registration efficiency, a voxelized 3D robot positioning method has emerged, which introduces the concept of voxel grid into the registration process, improving the calculation speed.

[0003] Traditional voxelized 3D robot positioning methods usually use hash tables to store the entire voxelized map to speed up the search. However, in a continuously changing scenario, the performance of the hash table will seriously decline. Fluctuations in point cloud density will lead to frequent fluctuations in the number of voxels in the hash table, which in turn triggers frequent expansion and contraction of the hash table, seriously affecting the accuracy of robot positioning and the performance of the voxel map in a warehousing environment. Summary of the Invention

[0004] The main objective of this application is to provide a method for robot positioning in a warehousing environment, aiming to solve the technical problem that in traditional robot positioning methods in a warehousing environment, when the point cloud density fluctuates, the hash table for storing the voxelized point cloud map expands and contracts frequently, thus affecting the accuracy of robot positioning.

[0005] To achieve the above objective, this application proposes a method for robot positioning in a warehousing environment, and the method includes:

[0006] Determine the voxel index of the initial point cloud, and obtain voxel depth information based on the voxel index and the preset number of hash table storage buckets;

[0007] Perform hierarchical processing on the voxelized point cloud map according to the voxel depth information to obtain multiple sub-voxel maps;

[0008] Locate the robot in the warehousing environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxel.

[0009] In one embodiment, the step of determining the voxel index of the initial point cloud and obtaining voxel depth information based on the voxel index and the preset number of hash table storage buckets includes:

[0010] By calculating the three-dimensional spatial point coordinates of the initial point cloud and the preset voxel resolution, the voxel index of each three-dimensional spatial point coordinate is obtained;

[0011] Based on the voxel index, a voxelized point cloud map is constructed, and based on the preset number of hash table buckets, the voxelized point cloud map is stored using a hash table;

[0012] According to the three-dimensional spatial point coordinates and the number of points inside the voxel, the voxel position information is calculated;

[0013] Based on the voxel position information, the distance between each voxel and the camera is calculated, and the distance is used as the voxel depth information.

[0014] In one embodiment, the step of constructing a voxelized point cloud map based on the voxel index and storing the voxelized point cloud map using a hash table based on the preset number of hash table buckets includes:

[0015] Based on the voxel index, a voxelized point cloud map is obtained;

[0016] According to the voxelized point cloud map and the initial point cloud, the number of voxels in the voxelized point cloud map is calculated;

[0017] By calculating the ratio of the number of elements in the storage bucket of the hash table to the initial number of buckets, the hash table quality factor is obtained;

[0018] According to the number of voxels and the hash table quality factor, the preset number of hash table buckets is calculated, and based on the preset number of hash table buckets, the voxelized point cloud map is stored using a hash table.

[0019] In one embodiment, the step of calculating the preset number of hash table buckets according to the number of voxels and the hash table quality factor and storing the voxelized point cloud map using a hash table based on the preset number of hash table buckets includes:

[0020] Calculate the preset number of hash table buckets according to the number of voxels and the hash table quality factor, and detect whether the hash table quality factor exceeds the preset expansion threshold;

[0021] If the hash table quality factor exceeds the preset expansion threshold, increase the preset number of hash table buckets and update the preset hash table;

[0022] If the hash table quality factor is lower than the preset shrinkage threshold, reduce the preset number of hash table buckets and update the preset hash table;

[0023] Based on the preset number of hash table buckets, the voxelized point cloud map is stored using a hash table.

[0024] In one embodiment, the step of positioning a robot in a warehousing environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxel includes:

[0025] Construct a voxel balanced binary tree based on the multiple sub-voxel maps, and determine a target point cloud sub-voxel map according to the voxel balanced binary tree;

[0026] Based on the target point cloud sub-voxel map, obtain a target point cloud voxel index by voxelizing the target point cloud;

[0027] Search for the target point cloud voxel corresponding to the target point cloud voxel index in the hash table storing the multiple sub-voxel maps;

[0028] Pair the target point cloud with the target point cloud voxel to obtain the correspondence between the target point cloud and the target point cloud voxel;

[0029] Based on the correspondence, position the robot in the warehousing environment corresponding to the target point cloud by calculating a loss function.

[0030] In one embodiment, the step of constructing a voxel balanced binary tree based on the multiple sub-voxel maps and determining a target point cloud sub-voxel map according to the voxel balanced binary tree includes:

[0031] Construct a voxel balanced binary tree according to the voxel indexes of the multiple sub-voxel maps and the red-black tree mechanism;

[0032] Obtain the depth information of the target point cloud by transforming and traversing the target point cloud, and search for the target point cloud voxel index corresponding to the depth information of the target point cloud in the voxel balanced binary tree;

[0033] Determine the target point cloud sub-voxel map according to the target point cloud voxel index.

[0034] In addition, to achieve the above object, the present application also proposes a robot positioning device in a warehousing environment, and the robot positioning device in the warehousing environment includes:

[0035] An information calculation module, configured to determine the voxel index of the initial point cloud, and obtain voxel depth information based on the voxel index and the preset number of hash table buckets;

[0036] A map layering module, configured to perform layering processing on the voxelized point cloud map according to the voxel depth information to obtain multiple sub-voxel maps;

[0037] A point cloud registration module, configured to position a robot in a warehousing environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxel.

[0038] In addition, to achieve the above object, the present application further provides a robot positioning device in a warehousing environment, the device including: a memory, a processor, and a computer program stored on the memory and executable on the processor, the computer program being configured to implement the steps of the robot positioning method in the warehousing environment as described above.

[0039] In addition, to achieve the above object, the present application further provides a storage medium, the storage medium being a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the robot positioning method in the warehousing environment as described above.

[0040] In addition, to achieve the above object, the present application further provides a computer program product, the computer program product including a computer program, and when the computer program is executed by a processor, it implements the steps of the robot positioning method in the warehousing environment as described above.

[0041] The technical solution proposed by the present application calculates the voxel index of the initial point cloud, and based on the voxel index and the preset number of hash table storage buckets, obtains the voxel depth information, performs hierarchical processing on the voxelized point cloud map according to the voxel depth information to obtain multiple sub-voxel maps, and locates the robot in the warehousing environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxel. By calculating the voxel index of the initial point cloud, the present application accurately divides the initial point cloud, sets the preset number of hash table storage buckets to avoid frequent expansion and contraction of the hash table during the registration process, and hierarchical processing divides the voxelized point cloud map into multiple sub-voxel maps, which helps to reduce the conflicts and search time of the hash table, and can also process the point cloud data more carefully, making the robot positioning process in the warehousing environment more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings herein are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0043] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 It is a schematic flowchart provided for Embodiment 1 of the robot positioning method in the warehousing environment of the present application;

[0045] Figure 2 It is a schematic diagram of the algorithm for constructing the hierarchical structure of the voxelized point cloud map of the robot positioning method in the warehousing environment of the present application;

[0046] Figure 3 It is a schematic flowchart provided for the second embodiment of the robot positioning method in the storage environment of this application;

[0047] Figure 4 It is a schematic diagram of voxelization processing for the robot positioning method in the storage environment of this application;

[0048] Figure 5 It is a schematic diagram of hash mapping;

[0049] Figure 6 It is a schematic flowchart provided for the third embodiment of the robot positioning method in the storage environment of this application;

[0050] Figure 7 It is a schematic diagram of balanced binary tree management voxel index for this application;

[0051] Figure 8 It is a schematic diagram of the point cloud registration process for this application;

[0052] Figure 9 It is a schematic diagram of the module structure of the robot positioning device in the storage environment for the embodiment of this application;

[0053] Figure 10 It is a schematic diagram of the device structure of the hardware operating environment involved in the robot positioning method in the storage environment for the embodiment of this application.

[0054] The realization of the purpose, functional features and advantages of this application will be further described with reference to the embodiments and the accompanying drawings. Detailed implementation manners

[0055] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of this application and are not used to limit this application.

[0056] In order to better understand the technical solutions of this application, the following will be described in detail in combination with the accompanying drawings of the specification and specific implementation manners.

[0057] In the traditional voxelized three-dimensional storage environment, the robot positioning method usually uses a hash table to store the entire voxelized map to speed up the search speed. However, in a continuously changing scenario, the performance of the hash table will seriously decline. The fluctuation of the point cloud density will cause frequent fluctuations in the number of voxels in the hash table, which will in turn cause frequent expansion and contraction of the hash table, seriously affecting the accuracy of robot positioning and the performance of the voxel map in the storage environment.

[0058] Therefore, in order to overcome the above-mentioned deficiencies, the present application provides a solution. By calculating the voxel indices of the initial point cloud, the initial point cloud is accurately divided. Setting the number of buckets in the preset hash table can avoid frequent expansion and contraction of the hash table during the registration process. The hierarchical processing divides the voxelized point cloud map into multiple sub-voxel maps, which helps to reduce the conflicts and search time of the hash table and can also process the point cloud data more meticulously, making the robot positioning process in the warehousing environment more accurate.

[0059] It should be noted that the execution subject of each embodiment of the present application can be a computing service device or system with data processing, network communication, and program running functions, such as an electronic device capable of implementing the above functions, a robot positioning system in a warehousing environment, etc. Hereinafter, taking the robot positioning system in a warehousing environment (hereinafter referred to as "system") as an example, the following embodiments will be described.

[0060] Based on this, the embodiments of the present application provide a robot positioning method in a warehousing environment, referring to Figure 1 , Figure 1 is a schematic flowchart of the first embodiment of the robot positioning method in a warehousing environment of the present application.

[0061] In this embodiment, the robot positioning method in a warehousing environment includes steps S10 to S30:

[0062] Step S10, determine the voxel indices of the initial point cloud, and obtain voxel depth information based on the voxel indices and the number of buckets in the preset hash table.

[0063] It should be noted that the point cloud is not only a set of discrete three-dimensional points on the surface of an object or scene in three-dimensional space, but also plays a crucial role in robotics, realizing the transformation from the physical world to the digital world. The robot in the warehousing environment scans the surrounding environment through various sensors (such as lidar, depth camera, etc.) it carries, captures the information of these discrete three-dimensional points, and forms a point cloud. This process is essentially a digital reconstruction of the real world, enabling the robot to "see" and understand its operating environment. Through the processing and analysis of the point cloud, such as the voxelization process described in step S10, the robot can efficiently organize and manage these data and quickly extract key features such as voxel depth information.

[0064] In addition, it should be noted that a voxel is the smallest unit in three-dimensional space, similar to a pixel in two-dimensional space; a voxel index refers to the position identifier of the voxel to which each point belongs when dividing point cloud data into different voxels in three-dimensional space; a hash table is a data structure that uses a hash function to map key-value pairs to positions in the table. In a hash table, a bucket is a container for storing key-value pairs with the same hash value, and voxel depth information refers to the depth (i.e., distance) of a voxel relative to a certain reference plane or viewpoint.

[0065] It should be understood that a hash table can quickly access data using key-value pairs, providing an ideal time complexity of O(1) for lookup. However, when there are too many input elements, resulting in severe conflicts in the hash table, the performance will drop significantly, approaching the O(n) time complexity. This performance degradation usually stems from the mismatch between the number of hash table buckets and the number of input elements. Setting a preset number of hash table buckets aims to balance conflicts and space utilization, ensuring that conflicts are reduced within a reasonable range while avoiding excessive idleness of buckets. By setting a reasonable preset number of hash table buckets, the expansion and contraction operations of the hash table can be reduced.

[0066] Step S20: Perform hierarchical processing on the voxelized point cloud map according to the voxel depth information to obtain multiple sub-voxel maps.

[0067] It should be noted that a voxelized point cloud map is a map form that converts point cloud data into a voxel grid representation. During the voxelization process, the point cloud data is divided into regular voxel units, thus simplifying data processing and analysis; a sub-voxel map is a local or specific-level voxel grid segmented from the voxelized point cloud map. By performing hierarchical processing on the voxelized point cloud map, multiple sub-voxel maps with different depths (or different regions) can be obtained.

[0068] For ease of understanding, reference is made to Figure 2 for illustration, but it does not limit the robot positioning method in the warehousing environment of this application. Figure 2 This is a schematic diagram of the algorithm for constructing the hierarchical structure of the voxelized point cloud map for the robot positioning method in the warehousing environment of this application. The algorithm for the hierarchical structure of the voxelized point cloud map manages voxels through the depth information of each voxel. Based on the depth information of each voxel and the total number of voxels, the voxelized point cloud map is hierarchically divided according to the distance from the voxel to the camera. Voxels with similar depth information are regarded as being in the same layer, and all voxels in the same layer form a sub-voxel map. The number of sub-voxel maps is positively correlated with the total number of voxels, ensuring that the number of voxels in each sub-voxel map is within a reasonable range. Among them, the return value of the algorithm for the hierarchical structure of the voxelized point cloud map is the voxel index depth_index, which is used to uniquely identify each layer of sub-voxel maps.

[0069] It can be understood that in this application, the voxelized point cloud map is stored in a hash table, and the voxelized point cloud map is hierarchically processed according to the voxel depth information to obtain multiple sub-voxel maps, that is, the voxels are classified according to the voxel depth information, and the voxels with the same voxel depth information are grouped under the same voxel index and then put into the same hash table, that is, the hash table originally containing the entire voxelized point cloud map is split into multiple sub-hash tables according to the voxel depth information. During this process, the number of voxel indices is controlled by the number of voxels, so the number of voxels under each voxel index will be controlled within a suitable range, so that each sub-hash table can be quickly searched under the preset number of hash table storage buckets, and the performance of the voxel map will not be affected by frequent expansion and contraction due to fluctuations in the number of voxels.

[0070] Step S30: Locate the robot in the storage environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxel.

[0071] It can be understood that multiple sub-voxel maps and the target point cloud voxel are obtained, and the sub-voxel maps and the target point cloud voxel have a consistent coordinate system and resolution. Moreover, when performing point cloud registration, it is necessary to ensure that the correspondence between the multiple sub-voxel maps and the target point cloud voxel is accurate. If there are errors or uncertainties in the correspondence, it may lead to deviations or failures in the registration results. Therefore, a balanced binary tree, a loss function, etc. can be introduced during the registration process to improve the accuracy and stability of the registration.

[0072] In this embodiment, by calculating the voxel index of the initial point cloud, the initial point cloud is accurately divided, and the preset number of hash table storage buckets is set to avoid frequent expansion and contraction of the hash table during the registration process. The hierarchical processing divides the voxelized point cloud map into multiple sub-voxel maps, which helps to reduce hash table conflicts and search time, and can also process point cloud data more carefully, making the robot positioning process in the storage environment more accurate.

[0073] Based on the first embodiment of this application, in the second embodiment of this application, the same or similar content as the above-mentioned embodiment one can be referred to the above introduction and will not be elaborated hereinafter. On this basis, please refer to Figure 3 , the step S10 may include steps S101 to S104:

[0074] Step S101: Calculate the voxel index of each three-dimensional space point coordinate by calculating the three-dimensional space point coordinates of the initial point cloud and the preset voxel resolution.

[0075] It should be noted that in point cloud registration, the voxelization method improves efficiency by dividing the point cloud into voxel grids. However, the change in point cloud density in the moving scene affects the performance of the traditional fixed-resolution voxelization method.

[0076] Therefore, the voxel size can be flexibly adjusted according to the local density of the point cloud, taking into account both accuracy and efficiency. When voxelizing the initial point cloud, based on the three-dimensional spatial point coordinates of the initial point cloud and the preset voxel resolution, the voxel index of each three-dimensional point coordinate of the initial point cloud is calculated to determine the voxel grid to which it belongs.

[0077] For ease of understanding, reference is made to Figure 4 for illustration, but it does not limit the present application. Figure 4 This is a schematic diagram of the voxelization process for the robot positioning method in the warehousing environment of the present application. Taking the voxelization method of the classic Voxelized Generalized Iterative Closest Point (VGICP) algorithm as an example, divide the three-dimensional spatial point coordinates of the initial point cloud by the preset voxel resolution and then subtract 0.5, and finally round down to get the result. This structure is the index of the voxel where the point is located. Points with the same calculation result through the voxelization method are considered to be in the same voxel. Taking Figure 4 as the three-dimensional coordinates in, for example, point B with coordinates (1.5, 1.5, 1.5) in three-dimensional space. Divide point B by the voxel resolution of 1.0 and then subtract 0.5, and finally round down to get the voxel index (1, 1, 1). Similarly, points with the same above calculation result are considered to be points in the voxel corresponding to the voxel index (1, 1, 1), that is, Figure 4 the voxel represented by the second outer cube in.

[0078] Step S102: Construct a voxelized point cloud map based on the voxel index, and store the voxelized point cloud map using a hash table based on the number of buckets in the preset hash table.

[0079] It should be understood that since each voxel has a unique voxel index, this index can form a one-to-one correspondence with the voxel itself. Therefore, when constructing the voxelized point cloud map, this correspondence can be utilized, taking the voxel index as the key of the hash table and the point cloud data contained in the voxel or the statistical information of the voxel (such as centroid, covariance, etc.) as the value of the hash table. In this way, through the hash table, the information in the voxelized point cloud map can be efficiently stored and retrieved.

[0080] Step S103: Calculate the voxel position information according to the three-dimensional spatial point coordinates and the number of points in the voxel.

[0081] It can be understood that the voxel position information calculation can be to first sum up the three-dimensional spatial point coordinates of the initial point cloud and then divide by the number of all points within the voxel. Specifically, for each voxel, traverse all the points inside it, and sum up the three-dimensional spatial coordinates of these points respectively. Divide the sum in each direction by the total number of points within the voxel to calculate the average value in that direction. These three average values (corresponding to the X, Y, and Z directions respectively) constitute the centroid coordinates of the voxel, and use the calculated centroid coordinates as the voxel position information. The position of the voxel can be quickly accessed and queried by associating the voxel position information with the voxel index.

[0082] Step S104, calculate the distance between each voxel and the camera based on the voxel position information, and use the distance as the voxel depth information.

[0083] It should be understood that after obtaining the voxel position information, calculate the distance between each voxel and the camera, and this distance is the voxel depth information. Specifically, during the system initialization or camera calibration process, determine the camera position and represent it in the form of coordinates in three-dimensional space. For each voxel, use the distance formula in three-dimensional space (such as Euclidean distance) to calculate the distance between its centroid coordinates and the camera coordinates, and use the calculated distance as the voxel depth information.

[0084] In this embodiment, by calculating the three-dimensional spatial point coordinates of the initial point cloud and the preset voxel resolution, obtain the voxel index of each three-dimensional spatial point coordinate, construct a voxelized point cloud map based on the voxel index, compress a large amount of point cloud data into fewer voxel representations, each voxel contains the information of multiple points, reducing the complexity of the data. Based on the preset number of hash table storage buckets, use the hash table to store the voxelized point cloud map. According to the three-dimensional spatial point coordinates and the number of voxels, calculate the voxel position information, calculate the distance between each voxel and the camera based on the voxel position information, and use the distance as the voxel depth information, so that the corresponding relationship between the target point cloud and the initial point cloud can be quickly located to achieve accurate registration. When the point cloud data changes, the voxel information in the hash table can be efficiently updated to maintain the real-time and accuracy of the data.

[0085] As an implementation manner, step S102 in this embodiment may include: obtaining a voxelized point cloud map based on the voxel index; calculating the number of voxels in the voxelized point cloud map according to the voxelized point cloud map and the initial point cloud; obtaining the hash table prime factor by calculating the ratio of the number of elements in the storage bucket of the hash table to the initial number of storage buckets; calculating the preset number of hash table storage buckets according to the number of voxels and the hash table prime factor, and using the hash table to store the voxelized point cloud map based on the preset number of hash table storage buckets.

[0086] It should be understood that a voxelized point cloud map is constructed based on voxel indices, and the voxel indices are used to identify voxels, where each voxel contains information about the points located inside it (such as coordinates, colors, etc.). Calculate the number of voxels in the voxelized point cloud map according to the voxelized point cloud map and the initial point cloud. This step is to understand the distribution of voxels in the voxelized point cloud map for subsequent optimized storage of the hash table. The calculation of the number of voxels can be achieved by traversing the voxelized point cloud map and counting the number of different voxel indices.

[0087] Then, obtain the hash table quality factor by calculating the ratio of the number of elements in the storage bucket of the hash table to the initial number of storage buckets. The hash table quality factor is an important indicator for measuring the current storage efficiency of the hash table. When the number of elements in the storage bucket is too large, the query performance of the hash table may decline. Therefore, it is necessary to calculate the quality factor to evaluate whether it is necessary to adjust the size of the hash table or remap the hash function. Further, calculate the preset number of hash table storage buckets according to the number of voxels and the hash table quality factor, and use the hash table to store the voxelized point cloud map based on this number. By reasonably setting the number of storage buckets, good performance of the hash table can be ensured when storing and querying the voxelized point cloud map.

[0088] As an implementation manner, the step of calculating the preset number of hash table storage buckets according to the number of voxels and the hash table quality factor and using the hash table to store the voxelized point cloud map based on the preset number of hash table storage buckets includes: calculating the preset number of hash table storage buckets according to the number of voxels and the hash table quality factor, and detecting whether the hash table quality factor exceeds a preset expansion threshold; if the hash table quality factor exceeds the preset expansion threshold, increase the preset number of hash table storage buckets and update the preset hash table; if the hash table quality factor is lower than the preset shrinkage threshold, reduce the preset number of hash table storage buckets and update the preset hash table; use the hash table to store the voxelized point cloud map based on the preset number of hash table storage buckets.

[0089] It should be understood that a suitable preset number of hash table storage buckets is calculated based on the number of voxels and the hash table prime factor. At the same time, the hash table prime factor is continuously monitored to evaluate the current storage state of the hash table. If the hash table prime factor exceeds the preset expansion threshold (indicating that there are too many elements in the current hash table storage buckets, which may lead to a decline in query performance), the preset number of hash table storage buckets is increased. After expansion, the hash table needs to be updated, including reallocating storage buckets and recalculating hash values, to ensure that all data in the voxelized point cloud map can be correctly mapped to the new hash table. If the hash table prime factor is lower than the preset contraction threshold (indicating that there are too few elements in the current hash table storage buckets and there is space waste), the preset number of hash table storage buckets is reduced. After contraction, the hash table also needs to be updated to optimize the storage space and improve query efficiency. Based on the adjusted preset number of hash table storage buckets, a hash table is used to store the voxelized point cloud map. At this time, the hash table has been optimized and can efficiently store and query the data of the voxelized point cloud map.

[0090] It should be noted that the thresholds need to be reasonably set according to the actual application scenario and performance requirements. The expansion threshold is usually set before the performance of the hash table starts to decline significantly, while the contraction threshold is set when the space utilization rate of the hash table is too low. After expansion or contraction, the hash function needs to be recalculated to ensure that the data can be correctly mapped to the new hash table, which usually involves reconstructing the hash table or reallocating storage buckets.

[0091] For the sake of easy understanding, reference is made to Figure 5 for illustration, but it does not limit this application. Figure 5 As shown in the hash mapping schematic diagram, the hash table calculates the address of the storage bucket for the input element through the hash function, and then mounts the element to the corresponding storage bucket. Since the number of storage buckets is preset, generally the time complexity of hash table lookup is only O(1). However, if the number of storage buckets is much less than the number of input elements and too many elements are mounted on one storage bucket, serious conflicts will occur. In the case of serious hash conflicts, the time complexity of O(1) is likely to degenerate into O(n). Usually, the hash conflict prime factor and the expansion and contraction strategies are used to solve the conflicts of the hash table. Expansion means expanding the number of storage buckets of the hash table and using more storage buckets to mount elements to reduce conflicts. Contraction means reducing the number of storage buckets, which means reducing the number of storage buckets correspondingly when the number of input elements is small to avoid waste. Among them, the hash table prime factor is the ratio of the number of elements in the storage bucket to the number of storage buckets (that is, the ratio of the number of elements in the storage bucket to the initial number of storage buckets), which is used to describe the intensity of hash conflicts. Whether to expand or contract is determined according to the prime factor.

[0092] In an `unordered_map` (an associative container in the C++ standard library that implements key-value pair storage based on a hash table), the default load factor for resizing is usually 0.7. This means that if the current number of buckets is 10, each bucket can store at most 7 elements. Once this limit is exceeded, resizing is required. The default load factor for shrinkage is usually 0.1. After resizing, the elements in the old buckets need to be re-migrated to the newly resized buckets. This process is called rehashing. If the number of input elements fluctuates significantly, the hash table needs to be resized repeatedly, resulting in repeated rehashing. In such cases, the performance of the hash table will decrease significantly. Based on this situation, the voxelized point cloud map hierarchical structure algorithm in this application controls the size of each sub-voxel map and presets the number of buckets in the `unordered_map` according to the number of voxels in the sub-voxel map to avoid collisions. Controlling the number of input voxels within a reasonable range can prevent the hash table from being resized frequently, thereby improving performance.

[0093] In this embodiment, by calculating the number of voxels in the voxelized point cloud map based on the voxelized point cloud map and the initial point cloud, obtaining the load factor of the hash table by calculating the ratio of the number of elements in the hash table bucket to the initial number of buckets, calculating the appropriate preset number of hash table buckets based on the number of voxels and the load factor of the hash table, avoiding frequent resizing of the hash table during the registration process, and dynamically adjusting the preset number of hash table buckets by monitoring the load factor of the hash table and according to the preset expansion threshold and preset shrinkage threshold, it can ensure that the hash table always maintains high performance and a low collision rate when storing the voxelized point cloud map, helps to adapt to point cloud data of different scales and distributions, improves the flexibility and scalability of the hash table. Compared with the traditional resizing method, the steps of this embodiment are more accurate, flexible and smooth.

[0094] Based on the first embodiment of this application, in the third embodiment of this application, the same or similar content as in the above-mentioned first embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 6 , step S30 may include steps S301 to S305:

[0095] Step S301, constructing a voxel balanced binary tree based on the multiple sub-voxel maps, and determining the target point cloud sub-voxel map according to the voxel balanced binary tree.

[0096] It should be understood that in this step, a red - black tree mechanism (or other balanced binary tree algorithms) can be utilized to construct a voxel balanced binary tree according to the voxel indices of the sub - voxel maps. In this process, each node represents a voxel or a set of voxels, and contains pointers to its child nodes as well as additional information (such as the number of interior points in the voxel, average depth, etc.). Then, traverse each point in the target point cloud, compare the depth information of the target point cloud with the depth range of the nodes in the tree, and decide whether to move to the left subtree or the right subtree according to the comparison result to find the corresponding voxel index. Then, retrieve the corresponding sub - voxel map from the voxelized point cloud map according to the voxel index of the target point cloud.

[0097] Therefore, step S301 may include: constructing a voxel balanced binary tree according to the voxel indices of the multiple sub - voxel maps and the red - black tree mechanism; obtaining the depth information of the target point cloud through transformation and traversal of the target point cloud, and finding the voxel index of the target point cloud corresponding to the depth information of the target point cloud in the voxel balanced binary tree; determining the target point cloud sub - voxel map according to the voxel index of the target point cloud.

[0098] It should be noted that a red - black tree is a self - balancing binary search tree, which can complete insertion, deletion, and search operations in O(log n) time. It maintains the balance of the tree through color attributes (red or black) and operations such as rotation and recoloring. The concept of red - black tree is introduced in this application to efficiently manage the voxels in the map, improving the registration speed while taking into account the registration accuracy, and further improving the real - time performance of robot positioning in the warehouse environment.

[0099] Exemplarily, as Figure 7 shown, Figure 7 This is a schematic diagram of the balanced binary tree management of voxel indices in this application. The voxel index is essentially the depth information of different hierarchical voxels, so it is ordered. The increasing voxel index starting from the root node represents the distance from the camera to the voxels at different depth layers from near to far. Constructing the voxel index into a balanced binary tree enables efficient searching. Also, since both the voxelized point cloud map and the voxel index are gradually constructed during the process of traversing the entire point cloud, new nodes need to be frequently inserted and the heights of the left and right subtrees need to be balanced during the construction of the binary tree, so that the search time complexity of the entire tree remains at O(logn). Using the red - black node mechanism in the balanced binary tree can still maintain the balance of the tree during the continuous insertion of new voxel indices, ensuring the query performance of the binary tree.

[0100] Step S302: Based on the target point cloud sub - voxel map, obtain the voxel index of the target point cloud by performing voxelization processing on the target point cloud.

[0101] Step S303: Search for the voxel of the target point cloud corresponding to the voxel index of the target point cloud in the hash table storing the multiple sub - voxel maps.

[0102] Step S304: Pair the target point cloud with the target point cloud voxel to obtain the correspondence between the target point cloud and the target point cloud voxel.

[0103] It can be understood that in 3D point cloud processing, in order to associate the target point cloud with its corresponding voxel information, a series of steps need to be performed to ensure that each point cloud point can be accurately matched with the voxel it belongs to. Using the previously constructed hash table (which stores multiple sub-voxel maps and their related information), the voxel data corresponding to each point can be quickly found according to the voxel index, and then the correspondence between each point and its voxel index can be obtained, and the pairing operation can be performed, which means associating each point cloud point with the voxel information found through the hash table, so that the voxel to which each point belongs and all the points contained in the voxel can be quickly accessed.

[0104] Step S305: Based on the correspondence, calculate the loss function to locate the robot in the storage environment corresponding to the target point cloud.

[0105] It should be noted that based on the correspondence between the target point cloud and the target point cloud voxel, the target point cloud can be further registered by calculating the loss function. Among them, the loss function (or called the error function) is a measure of the alignment degree between the target point cloud and the reference point cloud. Based on the correspondence, the distance between each target point and its corresponding reference point (or voxel center, average position of points in the voxel, etc.) can be calculated, and the sum, average value or other statistics of these distances can be used as the loss function. Common loss functions include point-to-point distance, point-to-plane distance, and normal distance, etc.

[0106] For ease of understanding, reference is made to Figure 8 for illustration, but it does not limit this application. Figure 8 This is a schematic diagram of the point cloud registration process of this application. The initial point cloud calculates the voxel index through transformation, constructs a voxel balanced binary tree based on the voxel index, searches for the target point cloud voxel index corresponding to the depth information of the target point cloud in the voxel balanced binary tree, thereby determining the target point cloud sub-voxel map, performs voxelization processing on the target point cloud to obtain the target point cloud voxel index, searches for the target point cloud voxel corresponding to the target point cloud voxel index in the hash table storing multiple sub-voxel maps, pairs the target point cloud with the target point cloud voxel to obtain the correspondence between the target point cloud and the target point cloud voxel, and finally inputs the coordinates and covariance of the point and voxel (determined by this point and the 20 nearest surrounding points) to calculate the loss function.

[0107] This embodiment constructs a voxel balanced binary tree based on multiple sub-voxel maps, providing higher search efficiency, and introducing a red-black tree mechanism to maintain the balance of the tree when inserting and deleting nodes. Determine the target point cloud sub-voxel map according to the voxel balanced binary tree. By voxelizing the target point cloud, obtain the voxel index of the target point cloud. Search for the target point cloud voxel corresponding to the voxel index of the target point cloud in the hash table storing the multiple sub-voxel maps. Pair the target point cloud with the target point cloud voxel to obtain the corresponding relationship between the target point cloud and the target point cloud voxel. By defining an appropriate loss function to measure the point cloud registration accuracy, improve the accuracy of the positioning result.

[0108] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the robot positioning method in the warehousing environment of this application. Based on this technical concept, more forms of simple transformations are within the protection scope of this application.

[0109] This application also provides a robot positioning device in a warehousing environment. Please refer to Figure 9 , the robot positioning device in the warehousing environment includes:

[0110] An information calculation module 10, configured to determine the voxel index of the initial point cloud, and obtain voxel depth information based on the voxel index and the preset number of hash table storage buckets;

[0111] A map layering module 20, configured to perform layering processing on the voxelized point cloud map according to the voxel depth information to obtain multiple sub-voxel maps;

[0112] A point cloud registration module 30, configured to position the robot in the warehousing environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxel.

[0113] The robot positioning device in the warehousing environment provided by this application adopts the robot positioning method in the above embodiment, which can solve the technical problem that in the traditional robot positioning method in the warehousing environment, when the point cloud density fluctuates, the hash table storing the voxelized point cloud map frequently expands and contracts, thus affecting the positioning accuracy of the robot. Compared with the prior art, the beneficial effects of the robot positioning device in the warehousing environment provided by this application are the same as those of the robot positioning method in the above embodiment, and other technical features in the robot positioning device in the warehousing environment are the same as those disclosed in the method of the above embodiment, and will not be elaborated here.

[0114] The present application provides a robot positioning device in a warehousing environment. The robot positioning device in the warehousing environment includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the robot positioning method in the warehousing environment in the first embodiment above.

[0115] Reference is made below to Figure 10 , which shows a schematic structural diagram of a robot positioning device in a warehousing environment suitable for implementing the embodiments of the present application. The robot positioning device in the warehousing environment in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Desctions), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 10 The shown robot positioning device in the warehousing environment is merely an example and should not impose any limitation on the functions and usage scope of the embodiments of the present application.

[0116] As Figure 10As shown in the figure, the robot positioning device in a warehousing environment may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM: Read Only Memory) 1002 or the program loaded from the storage device 1003 into the random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of the robot positioning device in the warehousing environment are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the robot positioning device in the warehousing environment to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows a robot positioning device in a warehousing environment with various systems, it should be understood that it is not required to implement or have all the shown systems. More or fewer systems can be alternatively implemented or had.

[0117] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above functions defined in the methods of the embodiments disclosed in the present application are executed.

[0118] The robot positioning device in the warehousing environment provided by this application adopts the robot positioning method in the warehousing environment in the above embodiment, which can solve the technical problem that in the traditional robot positioning method in the warehousing environment, when the point cloud density fluctuates, the hash table storing the voxelized point cloud map frequently expands and contracts, thus affecting the accuracy of robot positioning. Compared with the prior art, the beneficial effects of the robot positioning device in the warehousing environment provided by this application are the same as those of the robot positioning method in the warehousing environment provided by the above embodiment, and other technical features in this robot positioning device in the warehousing environment are the same as those disclosed in the method of the previous embodiment, which will not be elaborated here.

[0119] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0120] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claimed rights.

[0121] This application provides a computer-readable storage medium with computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the robot positioning method in the warehousing environment in the above embodiment.

[0122] The computer-readable storage medium provided by this application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or component. The program code contained on the computer-readable storage medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0123] The above computer-readable storage medium can be included in a robot positioning device in a warehousing environment; it can also exist independently without being assembled into a robot positioning device in a warehousing environment.

[0124] The above computer-readable storage medium carries one or more programs. When the one or more programs are executed by a robot positioning device in a warehousing environment, the robot positioning device in the warehousing environment is caused to: calculate the voxel index of the initial point cloud, obtain voxel depth information based on the voxel index and the preset number of hash table buckets, perform hierarchical processing on the voxelized point cloud map according to the voxel depth information to obtain multiple sub-voxel maps, and position the robot in the warehousing environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxel.

[0125] Computer program code for performing the operations of this application can be written in one or more programming languages or combinations thereof. The above-mentioned programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0126] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0127] The modules described in the embodiments of this application can be implemented in software or in hardware. Among them, the name of the module does not constitute a limitation to the unit itself in some cases.

[0128] The readable storage medium provided by this application is a computer-readable storage medium. The computer-readable storage medium stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned robot positioning method in a warehousing environment, and can solve the technical problem that in the traditional robot positioning method in a warehousing environment, when the point cloud density fluctuates, the hash table storing the voxelized point cloud map frequently expands and contracts, thus affecting the accuracy of robot positioning. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by this application are the same as those of the robot positioning method in a warehousing environment provided by the above embodiments, and will not be elaborated here.

[0129] This application also provides a computer program product, including a computer program, and when the computer program is executed by a processor, it realizes the steps of the above-mentioned robot positioning method in a warehousing environment.

[0130] The computer program product provided by this application can solve the technical problem that in the traditional robot positioning method in a warehousing environment, when the point cloud density fluctuates, the hash table storing the voxelized point cloud map frequently expands and contracts, thus affecting the accuracy of robot positioning. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as those of the robot positioning method in a warehousing environment provided by the above embodiments, and will not be elaborated here.

[0131] The above are only some embodiments of this application, and do not limit the patent scope of this application. Any equivalent structural transformation made by using the specification and drawings of this application under the technical concept of this application, or any direct / indirect application in other related technical fields, is included in the patent protection scope of this application.

Claims

1. A robot positioning method in a warehouse environment, characterized in that: The method comprises the following steps: Determine a voxel index of an initial point cloud, and obtain voxel depth information based on the voxel index and a preset number of hash table buckets; Performing layered processing on the voxelized point cloud map according to the voxel depth information to obtain a plurality of sub-voxel maps; The robot is positioned in the storage environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxels.

2. The robot positioning method in a warehouse environment as claimed in claim 1, characterized in that: The step of determining the voxel index of the initial point cloud and obtaining the voxel depth information based on the voxel index and the preset number of hash table buckets includes: By calculating the three-dimensional space point coordinates of the initial point cloud and the preset voxel resolution, the voxel index of each three-dimensional space point coordinate is obtained; Constructing a voxelized point cloud map based on the voxel index, and storing the voxelized point cloud map using a hash table based on a preset number of hash table storage buckets; Calculating voxel position information according to the three-dimensional space point coordinates and the number of points in the voxel; The distance between each voxel and the camera is calculated based on the voxel position information, and the distance is used as the voxel depth information.

3. The robot positioning method in a warehouse environment as claimed in claim 2, characterized in that: The step of constructing a voxelized point cloud map based on the voxel index and storing the voxelized point cloud map using a hash table based on a preset number of hash table storage buckets includes: Obtain a voxelized point cloud map based on the voxel index; Calculating the number of voxels in the voxelized point cloud map according to the voxelized point cloud map and the initial point cloud; Obtaining a hash table prime factor by calculating a ratio of the number of elements in a storage bucket of the hash table to the number of initial storage buckets; According to the number of voxels and the hash table quality factor, a preset number of hash table storage buckets is calculated, and based on the preset number of hash table storage buckets, a hash table is used to store the voxelized point cloud map.

4. The robot positioning method in a warehouse environment as claimed in claim 3, characterized in that: The step of calculating the number of preset hash table storage buckets according to the number of voxels and the hash table quality factor, and using a hash table to store the voxelized point cloud map based on the preset number of hash table storage buckets includes: Calculating a preset number of hash table storage buckets according to the number of voxels and the hash table quality factor, and detecting whether the hash table quality factor exceeds a preset expansion threshold; If the hash table quality factor exceeds the preset expansion threshold, the number of preset hash table storage buckets is increased and the preset hash table is updated; If the hash table quality factor is lower than the preset shrinking threshold, reducing the number of preset hash table storage buckets and updating the preset hash table; Based on the preset number of hash table storage buckets, a hash table is used to store the voxelized point cloud map.

5. The robot positioning method in a storage environment according to any one of claims 1 to 4, characterized in that: The step of positioning the robot in the storage environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxels comprises: constructing a voxel-balanced binary tree based on the multiple sub-voxel maps, and determining a sub-voxel map of a target point cloud according to the voxel-balanced binary tree; Based on the target point cloud sub-voxel map, obtaining a target point cloud voxel index by voxelizing the target point cloud; Searching for a target point cloud voxel corresponding to the target point cloud voxel index in a hash table storing the plurality of sub-voxel maps; Pairing the target point cloud with the target point cloud voxel to obtain a corresponding relationship between the target point cloud and the target point cloud voxel; Based on the corresponding relationship, the robot is positioned in the storage environment corresponding to the target point cloud by calculating the loss function.

6. The robot positioning method in a warehouse environment as claimed in claim 5, characterized in that: The step of constructing a voxel balanced binary tree based on the multiple sub-voxel maps, and determining a sub-voxel map of a target point cloud according to the voxel balanced binary tree comprises: Constructing a voxel balanced binary tree according to the voxel indexes of the plurality of sub-voxel maps and a red-black tree mechanism; Obtaining target point cloud depth information by transforming and traversing the target point cloud, and searching the target point cloud voxel index corresponding to the target point cloud depth information in the voxel balanced binary tree; The target point cloud sub-voxel map is determined according to the target point cloud voxel index.

7. A robot positioning device in a warehouse environment, characterized in that: The robot positioning device in the storage environment includes: An information calculation module, used to determine the voxel index of the initial point cloud, and obtain voxel depth information based on the voxel index and a preset number of hash table buckets; A map layering module, used for performing layering processing on the voxelized point cloud map according to the voxel depth information to obtain a plurality of sub-voxel maps; The point cloud registration module is used to locate the robot in the storage environment corresponding to the target point cloud according to the multiple sub-voxel maps and the target point cloud voxels.

8. A robot positioning device in a warehouse environment, characterized in that: The robot positioning device in a warehousing environment comprises: a memory, a processor, and a robot positioning program in a warehousing environment stored in the memory and executable on the processor. When the robot positioning program in a warehousing environment is executed by the processor, the robot positioning method in a warehousing environment as described in any one of claims 1 to 6 is implemented.

9. A storage medium, characterized in that: The storage medium stores a robot positioning program in a warehouse environment, and when the robot positioning program in a warehouse environment is executed by the processor, the robot positioning method in a warehouse environment according to any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that The computer program product comprises a robot positioning program in a warehouse environment, and when the robot positioning program in a warehouse environment is executed by a processor, the robot positioning method in a warehouse environment as claimed in any one of claims 1 to 6 is implemented.

Images