A method and device for real-time positioning and map construction of a scrap yard grabber

Abstract

The application relates to the technical field of computers, and discloses a scrap steel yard steel grabber instant positioning and map construction method and device, which comprises the following steps: acquiring first point cloud data uploaded by a first data acquisition device on a scrap steel yard steel grabber and an indoor environment initial map of the scrap steel yard; performing feature matching on the first point cloud data and the indoor environment initial map to obtain initial pose estimation information of the scrap steel yard steel grabber and position information in the indoor environment initial map; taking the initial pose estimation information as a radar odometer initial value, and obtaining interframe pose estimation information through nonlinear optimization; matching the interframe pose estimation information with the indoor environment initial map to obtain accurate pose estimation information, updating the indoor environment initial map according to the accurate pose estimation information, and obtaining a global map of the scrap steel yard. The method can accelerate the matching speed under the premise of ensuring the accuracy, can realize real-time positioning of the position of the scrap steel yard steel grabber in an indoor scene, and can obtain an indoor map.

Description

Technical Field

[0001] This invention relates to the field of computer technology, specifically to a method and apparatus for real-time positioning and map construction of a steel grabber in a scrap steel yard. Background Technology

[0002] Scrap steel is an energy-saving and green resource. In the process of recycling scrap steel, it is necessary to control the steel grabber to travel to the target location, load the scrap steel from the scrap steel pile into the transport vehicle, or unload the scrap steel from the transport vehicle into the scrap steel pile, so as to realize the outbound and inbound of scrap steel.

[0003] In indoor buildings with weak GPS signals, diverse functional areas, and numerous and changing obstacles, steel-grabbing machines find it difficult to accurately locate themselves without relying on satellite positioning systems such as GPS. Summary of the Invention

[0004] To address the aforementioned technical problems, embodiments of this application provide a method and apparatus for real-time positioning and map construction of a scrap steel grabber in a scrap steel yard, thereby solving the problem in the prior art that it is difficult for a scrap steel grabber to be accurately positioned in an indoor building without relying on satellite positioning systems such as GPS.

[0005] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part by practice of this application.

[0006] In one embodiment of this application, a method for real-time positioning and map construction of a scrap metal grabber in a scrap metal yard is provided, the method comprising:

[0007] Acquire the first point cloud data uploaded by the first data acquisition device on the steel grabber, as well as the initial map of the indoor environment of the scrap steel yard;

[0008] The first point cloud data is matched with the initial map of the indoor environment to obtain the initial pose estimation information of the steel grabber and its position information in the initial map of the indoor environment.

[0009] The initial pose estimation information is used as the initial value of the radar odometry, and the inter-frame pose estimation information is obtained by nonlinear optimization.

[0010] The inter-frame pose estimation information is matched with the initial indoor environment map to obtain accurate pose estimation information, and the initial indoor environment map is updated based on the accurate pose estimation information to obtain a global map of the scrap yard.

[0011] In one embodiment of this application, acquiring the first point cloud data uploaded by the first data acquisition device on the steel grabber includes:

[0012] Acquire the initial point cloud data uploaded by the first data acquisition device on the steel grabber;

[0013] Using multiple preset radar coordinate system information, and employing an iterative nearest point algorithm with the distance from a point to a surface as the error function, point cloud fusion is performed one by one based on the radar with the most laser lines to obtain fused point cloud data.

[0014] The fused point cloud data is subjected to ground segmentation and point cloud filtering, and the filtered non-ground point cloud data is used as the first point cloud data.

[0015] In one embodiment of this application, ground segmentation is performed on the fused point cloud data, including:

[0016] The parameters of the pre-built ground segmentation model are calculated by randomly sampling three points from the fused point cloud data.

[0017] Calculate the distance from the remaining points to the model. If the distance from the remaining points to the model is less than a preset threshold, then set the corresponding points as interior points and count the total number of interior points.

[0018] By iterating through random sampling steps and calculating the distance from a point to the model, the model parameters with the most interior points are obtained.

[0019] The pre-built ground segmentation model is updated based on the model parameter with the most interior points, and the fused point cloud data is transmitted to the updated ground segmentation model for ground segmentation.

[0020] In one embodiment of this application, point cloud filtering is performed on the fused point cloud data, including:

[0021] The point cloud filtering includes one of the following: cube filtering, voxel filtering, self-filtering, and clustering filtering;

[0022] The clustering filter is used to cluster non-ground point clouds and filter out clusters with fewer than 30 point clouds.

[0023] The self-filtering method calculates the shape of the steel grabbing machine's robotic arm in real time based on the tilt angle information of the robotic arm, and filters out the point cloud of the robotic arm.

[0024] In one embodiment of this application, the first point cloud data includes three-dimensional coordinates and reflection intensity. After using filtered non-ground point cloud data as the first point cloud data, the method further includes:

[0025] Input the first point cloud data into the pre-trained target detection model;

[0026] The pre-trained target detection model is used to detect targets in the first point cloud data, identify the steel grabber in the first point cloud data, and obtain the reflection intensity and location information of the steel grabber.

[0027] In one embodiment of this application, before inputting the first point cloud data into a pre-trained target detection model, the method further includes:

[0028] Obtain a target detection training dataset for specific targets in a scrap steel yard, including scrap steel piles, trucks, support columns, and steel grabbers.

[0029] The object detection training dataset is input into a pre-built initial model for training, thereby updating the parameters of the objective function of the initial model and obtaining the trained object detection model.

[0030] In one embodiment of this application, obtaining an initial map of the indoor environment of a scrap yard includes:

[0031] Acquire second-point cloud data uploaded by a second data acquisition device in an indoor scene;

[0032] The initial map of the indoor environment is obtained using LEGO-LOAM based on the second point cloud data.

[0033] In one embodiment of this application, the first point cloud data is matched with the initial indoor environment map to obtain the initial pose estimation information of the steel grabber and its position information in the initial indoor environment map, including:

[0034] The reflection intensity of the steel grabber in the initial map of the indoor environment is compared, and the target area with the same reflection intensity is taken as the region of interest for point cloud matching.

[0035] By using the keyframes of the initial indoor environment map, feature matching of planar points and edge points is performed on the region of interest to obtain successfully matched keyframes.

[0036] Based on the successfully matched keyframes, the initial pose estimation information and the position information in the initial map of the indoor environment are determined.

[0037] In one embodiment of this application, the initial pose estimation information is used as the initial value for radar odometry, and inter-frame pose estimation information is obtained through nonlinear optimization, including:

[0038] Extract and match line feature points and surface feature points from two adjacent frames of the point cloud of the steel grabber;

[0039] The error from point to line is determined based on the line feature points, the error from point to surface is determined based on the surface feature points, and the initial pose estimation information is used as the initial value for radar odometry.

[0040] Based on the point-to-line error, the point-to-surface error, and the initial value of the radar odometer, the inter-frame pose estimation information is obtained through nonlinear optimization.

[0041] In one embodiment of this application, a device for real-time positioning and mapping of a scrap metal grabber in a scrap metal yard is also provided, the device comprising:

[0042] The data acquisition module is used to acquire the first point cloud data uploaded by the first data acquisition device on the steel grabber, as well as the initial map of the indoor environment of the scrap steel yard;

[0043] The feature matching module is used to perform feature matching between the first point cloud data and the initial map of the indoor environment to obtain the initial pose estimation information of the steel grabber and its position information in the initial map of the indoor environment.

[0044] The odometry module is used to take the initial pose estimation information as the initial value of the radar odometry and obtain the inter-frame pose estimation information through nonlinear optimization.

[0045] The mapping module is used to update the pre-built ground segmentation model based on the model parameter with the most interior points, and to transmit the fused point cloud data to the updated ground segmentation model for ground segmentation.

[0046] In the technical solution provided by the embodiments of this application, firstly, the first point cloud data uploaded by the first data acquisition device on the steel grabber and the initial indoor environment map of the scrap steel yard are acquired; then, the first point cloud data and the initial indoor environment map are matched for features to obtain the initial pose estimation information of the steel grabber and its position information in the initial indoor environment map; next, the initial pose estimation information is used as the initial value of the radar odometry, and the inter-frame pose estimation information is obtained through nonlinear optimization; finally, the inter-frame pose estimation information is matched with the initial indoor environment map to obtain accurate pose estimation information, and the initial indoor environment map is updated according to the accurate pose estimation information to obtain a global map of the scrap steel yard. The method in this invention combines the point cloud data of the steel grabber with the initial indoor environment map to locate the position of the steel grabber in the indoor scene in real time and obtain a global map of the scrap steel yard, which speeds up the matching speed while ensuring accuracy.

[0047] It should be understood that the above general description and the following detailed description are merely exemplary and do not limit this application. Attached Figure Description

[0048] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings from these drawings without any inventive effort. In the drawings:

[0049] Figure 1 This is a flowchart illustrating an exemplary embodiment of the present application of a method for real-time positioning and map construction of a steel grabber in a scrap yard;

[0050] Figure 2 This is a flowchart illustrating another exemplary embodiment of the present application of a method for real-time positioning and map construction of a scrap steel yard grabber;

[0051] Figure 3 This is a schematic diagram of the structure of a real-time positioning and mapping device for a scrap steel yard grabber, as shown in an exemplary embodiment of this application. Detailed Implementation

[0052] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0053] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0054] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0055] In this application, "multiple" refers to two or more; "and / or" describes the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship.

[0056] Here, we introduce the definitions of abbreviations and key terms used in this application:

[0057] SLAM (simultaneous localization and mapping) addresses the challenge of placing a robot in an unknown location within an unknown environment. The question is whether a method can enable the robot to gradually create a complete map of this environment while simultaneously determining its direction of movement. A complete map (a consistent map) means the robot can navigate unimpeded to every accessible corner of a room.

[0058] Currently, SLAM technology involves many sensors, such as range and angle sensors, visual cameras, LiDAR, and inertial navigation. Depending on the sensor used, it is divided into Lidar SLAM and Visual SLAM. Among them, Lidar SLAM has an early start, high reliability, high accuracy, intuitive mapping, and is not affected by changes in indoor lighting, so it is widely used in the field of intelligent robots.

[0059] The present application provides a method and apparatus for real-time positioning and mapping of a scrap steel grabber in a scrap steel yard, which involves the SLAM technology described above. Considering the indoor environment of the actual operation of the scrap steel grabber, these embodiments will be described in detail below.

[0060] Figure 1 This is a flowchart illustrating an exemplary embodiment of this application regarding a method for real-time positioning and map construction of a scrap metal grabber in a scrap metal yard. The method provided in this application can be executed by any electronic device with computing power; for example, it can be executed by a server or a terminal device, or by both. In the following embodiments, a terminal device is used as the execution subject for illustrative purposes, but this disclosure is not limited thereto.

[0061] Reference Figure 1 The real-time positioning and map construction method for scrap steel yard grabbers provided in this application embodiment may include the following steps.

[0062] In step S110, initial data is obtained.

[0063] For example, the first point cloud data uploaded by the first data acquisition device on the steel grabber, and the initial map of the indoor environment of the scrap steel yard are obtained.

[0064] In step S120, feature matching is performed on the point cloud data and the map.

[0065] For example, the first point cloud data is matched with the initial map of the indoor environment to obtain the initial pose estimation information of the steel grabber and its position information in the initial map of the indoor environment.

[0066] In step S130, inter-frame pose estimation information is determined.

[0067] For example, the initial pose estimation information is used as the initial value of the radar odometry, and the inter-frame pose estimation information is obtained by nonlinear optimization.

[0068] In step S140, precise pose estimation information is determined and a global map is constructed.

[0069] For example, the inter-frame pose estimation information is matched with the initial indoor environment map using Scan-To-Map to obtain accurate pose estimation information, and the initial indoor environment map is updated based on the accurate pose estimation information to obtain a global map of the scrap yard.

[0070] Through the above steps S110 to S140, the point cloud data of the steel grabber is combined with the initial map of the indoor environment to locate the position of the steel grabber in the indoor scene in real time and obtain the global map of the scrap steel yard. While ensuring accuracy, the matching speed is also accelerated.

[0071] In one embodiment of this application, acquiring the first point cloud data uploaded by the first data acquisition device on the steel grabber includes the following steps:

[0072] Acquire the initial point cloud data uploaded by the first data acquisition device on the steel grabber;

[0073] Using multiple preset radar coordinate system information, and employing an iterative nearest point algorithm with the distance from a point to a surface as the error function, point cloud fusion is performed one by one based on the radar with the most laser lines to obtain fused point cloud data.

[0074] The fused point cloud data is subjected to ground segmentation and point cloud filtering, and the filtered non-ground point cloud data is used as the first point cloud data.

[0075] In one embodiment of this application, ground segmentation is performed on the fused point cloud data, including the following steps:

[0076] The parameters of the pre-built ground segmentation model are calculated by randomly sampling three points from the fused point cloud data.

[0077] Calculate the distance from the remaining points to the model. If the distance from the remaining points to the model is less than a preset threshold, then set the corresponding points as interior points and count the total number of interior points.

[0078] By iterating through random sampling steps and calculating the distance from a point to the model, the model parameters with the most interior points are obtained.

[0079] The pre-built ground segmentation model is updated based on the model parameter with the most interior points, and the fused point cloud data is transmitted to the updated ground segmentation model for ground segmentation.

[0080] It should be noted that in this embodiment, the RANSAC algorithm is used to randomly sample the fused point cloud data and calculate the number of interior points. Then, the model parameter with the most interior points is used to update the pre-built ground segmentation model.

[0081] In one embodiment of this application, point cloud filtering is performed on the fused point cloud data, including the following steps:

[0082] The point cloud filtering includes one of the following: cube filtering, voxel filtering, self-filtering, and clustering filtering;

[0083] The clustering filter is used to cluster non-ground point clouds and filter out clusters with fewer than 30 point clouds.

[0084] The self-filtering method calculates the shape of the steel grabbing machine's robotic arm in real time based on the tilt angle information of the robotic arm, and filters out the point cloud of the robotic arm.

[0085] In one embodiment of this application, the first point cloud data includes three-dimensional coordinates and reflection intensity. After using filtered non-ground point cloud data as the first point cloud data, the following steps are further included:

[0086] Input the first point cloud data into the pre-trained target detection model;

[0087] The pre-trained target detection model is used to detect targets in the first point cloud data, identify the steel grabber in the first point cloud data, and obtain the reflection intensity and location information of the steel grabber.

[0088] In one embodiment of this application, before inputting the first point cloud data into a pre-trained target detection model, the following steps are further included:

[0089] Obtain a target detection training dataset for specific targets in a scrap steel yard, including scrap steel piles, trucks, support columns, and steel grabbers.

[0090] The object detection training dataset is input into a pre-built initial model for training, thereby updating the parameters of the objective function of the initial model and obtaining the trained object detection model.

[0091] In one embodiment of this application, obtaining an initial map of the indoor environment of a scrap yard includes:

[0092] Acquire second-point cloud data uploaded by a second data acquisition device in an indoor scene;

[0093] The initial map of the indoor environment is obtained using LEGO-LOAM based on the second point cloud data.

[0094] In one embodiment of this application, feature matching is performed between the first point cloud data and the initial indoor environment map to obtain the initial pose estimation information of the steel grabber and its position information in the initial indoor environment map, including:

[0095] The reflection intensity of the steel grabber in the initial map of the indoor environment is compared, and the target area with the same reflection intensity is taken as the region of interest for point cloud matching.

[0096] By using the keyframes of the initial indoor environment map, feature matching of planar points and edge points is performed on the region of interest to obtain successfully matched keyframes.

[0097] Based on the successfully matched keyframes, the initial pose estimation information and the position information in the initial map of the indoor environment are determined.

[0098] In one embodiment of this application, the initial pose estimation information is used as the initial value for radar odometry, and inter-frame pose estimation information is obtained through nonlinear optimization, including:

[0099] Extract and match line feature points and surface feature points from two adjacent frames of the point cloud of the steel grabber;

[0100] The error from point to line is determined based on the line feature points, the error from point to surface is determined based on the surface feature points, and the initial pose estimation information is used as the initial value for radar odometry.

[0101] Based on the point-to-line error, the point-to-surface error, and the initial value of the radar odometer, the inter-frame pose estimation information is obtained through nonlinear optimization.

[0102] For ease of understanding, one embodiment of this application illustrates a method for real-time positioning and map construction of a scrap metal grabber in a scrap metal yard using a specific example. For instance... Figure 2 As shown, Figure 2 This is a flowchart illustrating another exemplary embodiment of the present application of a method for real-time positioning and map construction of a scrap metal yard grabber, which is described in detail below:

[0103] Point cloud data is obtained by the lidar on the steel grabber, and the tilt angle is obtained by the motion sensor on the robotic arm of the steel grabber.

[0104] By using multiple preset lidar coordinate system information, multiple lidar point cloud data are fused, followed by ground segmentation and point cloud filtering. The filtered non-ground point cloud is then used as the point cloud data of the steel grabbing machine.

[0105] Load the target detection model of the scrap steel yard for point cloud target detection: Use the target detection dataset created for specific targets in the scrap steel yard to train the target detection model, perform target detection on the point cloud data of the steel grabber, and obtain the reflection intensity and position information of the target.

[0106] Load the global map created by LEGO-LOAM;

[0107] Feature matching is performed on the same ROI (region of interest) region in the map keyframes and the current point cloud of the steel grabber: the target detection model of the scrap steel yard is used to detect targets in the point cloud currently acquired by the steel grabber and the keyframes in the global map. Target regions with the same reflection intensity are taken as ROI regions for point cloud matching. Feature matching of planar points and edge points is performed on the point cloud ROI regions.

[0108] The coarse pose estimate of the steel grabber is determined by the successfully matched frames, and the position of the steel grabber in the scrap steel yard is obtained.

[0109] Line feature points and surface feature points of the point cloud of two adjacent frames of the steel grabber are extracted and matched. Based on the position information of the steel grabber in the scrap steel yard as the initial value, the inter-frame pose estimation is obtained through nonlinear optimization.

[0110] Based on the inter-frame pose estimation of the steel grabber, Scan-To-Map matching is performed to obtain the precise pose estimation of the steel grabber, and then mapping and global map are built and updated.

[0111] It should be noted that although the steps of the method in this application are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0112] The following describes an embodiment of the apparatus described in this application, which can be used to execute the real-time positioning and mapping method for scrap metal yard grabbers in the above embodiments of this application. See also Figure 3 , Figure 3 This is a schematic diagram illustrating the structure of a real-time positioning and mapping device for a scrap metal yard grabber, as shown in an exemplary embodiment of this application. The device includes a data acquisition module 301, a feature matching module 302, an odometer module 303, and a mapping module 304.

[0113] Among them, the data acquisition module 301 is used to acquire the first point cloud data uploaded by the first data acquisition device on the steel grabber, as well as the initial map of the indoor environment of the scrap steel yard.

[0114] The feature matching module 302 is used to perform feature matching between the first point cloud data and the initial map of the indoor environment to obtain the initial pose estimation information of the steel grabber and its position information in the initial map of the indoor environment.

[0115] Odometer module 303 is used to take the initial pose estimation information as the initial value of the radar odometer and obtain the inter-frame pose estimation information through nonlinear optimization solution.

[0116] The mapping module 304 is used to update the pre-built ground segmentation model according to the model parameter with the most interior points, and to transmit the fused point cloud data to the updated ground segmentation model for ground segmentation.

[0117] In another embodiment of this application, the structure of a real-time positioning and mapping device for a scrap metal yard grabber is provided in detail. The device includes:

[0118] The data acquisition module is used to acquire point cloud data through the lidar on the steel grabber and to acquire tilt angle through the motion sensor on the robotic arm of the steel grabber;

[0119] The preprocessing module is used to fuse multiple lidar point cloud data through multiple preset lidar coordinate system information, and then perform ground segmentation and point cloud filtering, using the filtered non-ground point cloud as the steel grabber.

[0120] The target detection module is used to load the target detection model of the scrap steel yard for point cloud target detection: the target detection model is trained using a target detection dataset created for specific targets in the scrap steel yard, and the target is detected on the point cloud data of the steel grabber to obtain the reflection intensity and position information of the target.

[0121] The data acquisition module is used to load the global map created by LEGO-LOAM;

[0122] The feature matching module is used to perform feature matching on the same target ROI region in the map keyframes and the current point cloud of the steel grabber: the scrap steel yard target detection model is used to perform target detection on the point cloud currently acquired by the steel grabber and the keyframes in the global map, and the target region with the same reflection intensity is taken as the ROI region for point cloud matching. The feature matching of planar points and edge points is performed on the point cloud ROI region.

[0123] The positioning module is used to determine the coarse pose estimate of the steel grabber by the successfully matched frames, and to obtain the position of the steel grabber in the scrap yard;

[0124] The odometer module is used to extract and match the line and surface feature points of the point cloud between two adjacent frames of the steel grabber. Based on the position information of the steel grabber in the scrap yard as the initial value, the inter-frame pose estimation is obtained through nonlinear optimization.

[0125] The mapping module is used to perform Scan-To-Map matching based on the inter-frame pose estimation of the steel grabber, obtain the fine pose estimation of the steel grabber, and then perform mapping and update the global map.

[0126] It should be noted that the apparatus provided in the above embodiments and the method provided in the above embodiments belong to the same concept, and the specific way in which each module and unit performs operations has been described in detail in the method embodiments, and will not be repeated here.

[0127] The above description is merely a preferred exemplary embodiment of this application and is not intended to limit the implementation of this application. Those skilled in the art can easily make corresponding modifications or alterations based on the main concept and spirit of this application. Therefore, the scope of protection of this application should be determined by the scope of protection claimed in the claims.

Claims

1. A method for real-time positioning and map construction of a steel grabber in a scrap steel yard, characterized in that, The method includes: Acquire the first point cloud data uploaded by the first data acquisition device on the steel grabber, as well as the initial map of the indoor environment of the scrap steel yard; The first point cloud data is matched with the initial map of the indoor environment to obtain the initial pose estimation information of the steel grabber and its position information in the initial map of the indoor environment. The initial pose estimation information is used as the initial value of the radar odometry, and the inter-frame pose estimation information is obtained by nonlinear optimization. The inter-frame pose estimation information is matched with the initial indoor environment map to obtain accurate pose estimation information, and the initial indoor environment map is updated based on the accurate pose estimation information to obtain a global map of the scrap steel yard. Among them, acquiring the first point cloud data uploaded by the first data acquisition device on the steel grabber includes: Acquire the initial point cloud data uploaded by the first data acquisition device on the steel grabber; Using multiple preset radar coordinate system information, and employing an iterative nearest point algorithm with the distance from a point to a surface as the error function, point cloud fusion is performed one by one based on the radar with the most laser lines to obtain fused point cloud data. The fused point cloud data is subjected to ground segmentation and point cloud filtering, and the filtered non-ground point cloud data is used as the first point cloud data. The point cloud filtering includes self-filtering, which calculates the shape of the steel grabbing machine arm in real time based on the tilt angle information of the steel grabbing machine arm and filters out the point cloud of the steel grabbing machine arm.

2. The method for real-time positioning and map construction of scrap steel grabbers in scrap steel yards according to claim 1, characterized in that, Ground segmentation is performed on the fused point cloud data, including: The parameters of the pre-built ground segmentation model are calculated by randomly sampling three points from the fused point cloud data. Calculate the distance from the remaining points to the model. If the distance from the remaining points to the model is less than a preset threshold, then set the corresponding points as interior points and count the total number of interior points. By iterating through random sampling steps and calculating the distance from a point to the model, the model parameters with the most interior points are obtained. The pre-built ground segmentation model is updated based on the model parameter with the most interior points, and the fused point cloud data is transmitted to the updated ground segmentation model for ground segmentation.

3. The method for real-time positioning and map construction of scrap steel grabbers in scrap steel yards according to claim 1, characterized in that, Point cloud filtering is performed on the fused point cloud data, including: The point cloud filtering includes one of the following: cube filtering, voxel filtering, self-filtering, and clustering filtering; The clustering filter is used to cluster non-ground point clouds and filter out clusters with fewer than 30 point clouds.

4. The method for real-time positioning and map construction of a scrap metal yard grabber according to claim 1, characterized in that, The first point cloud data includes three-dimensional coordinates and reflection intensity. After using filtered non-ground point cloud data as the first point cloud data, it also includes: Input the first point cloud data into the pre-trained target detection model; The pre-trained target detection model is used to detect targets in the first point cloud data, identify the steel grabber in the first point cloud data, and obtain the reflection intensity and location information of the steel grabber.

5. The method for real-time positioning and map construction of a scrap metal yard grabber according to claim 4, characterized in that, Before inputting the first point cloud data into the pre-trained object detection model, the following steps are also included: Obtain a target detection training dataset for specific targets in a scrap steel yard, including scrap steel piles, trucks, support columns, and steel grabbers. The object detection training dataset is input into a pre-built initial model for training, thereby updating the parameters of the objective function of the initial model and obtaining the trained object detection model.

6. The method for real-time positioning and map construction of a scrap metal yard grabber according to claim 1, characterized in that, Obtain an initial map of the indoor environment of the scrap yard, including: Acquire second-point cloud data uploaded by a second data acquisition device in an indoor scene; The initial map of the indoor environment is obtained using LEGO-LOAM based on the second point cloud data.

7. The method for real-time positioning and mapping of scrap steel grabbers in scrap steel yards according to claim 1 or 4, characterized in that, The first point cloud data is matched with the initial indoor environment map to obtain the initial pose estimation information of the steel grabber and its position information in the initial indoor environment map, including: The reflection intensity of the steel grabber in the initial map of the indoor environment is compared, and the target area with the same reflection intensity is taken as the region of interest for point cloud matching. By using the keyframes of the initial indoor environment map, feature matching of planar points and edge points is performed on the region of interest to obtain successfully matched keyframes. Based on the successfully matched keyframes, the initial pose estimation information and the position information in the initial map of the indoor environment are determined.

8. The method for real-time positioning and map construction of a scrap metal yard grabber according to claim 1, characterized in that, Using the initial pose estimation information as the initial value for radar odometry, inter-frame pose estimation information is obtained through nonlinear optimization, including: Extract and match line feature points and surface feature points from two adjacent frames of the point cloud of the steel grabber; The error from point to line is determined based on the line feature points, the error from point to surface is determined based on the surface feature points, and the initial pose estimation information is used as the initial value for radar odometry. Based on the point-to-line error, the point-to-surface error, and the initial value of the radar odometer, the inter-frame pose estimation information is obtained through nonlinear optimization.

9. A real-time positioning and mapping device for a scrap metal grabber in a scrap metal yard, characterized in that, The device includes: The data acquisition module is used to acquire the first point cloud data uploaded by the first data acquisition device on the steel grabber, and the initial map of the indoor environment of the scrap steel yard. Acquiring the first point cloud data uploaded by the first data acquisition device on the steel grabber includes: acquiring the initial point cloud data uploaded by the first data acquisition device on the steel grabber; using multiple preset radar coordinate system information, and employing an iterative nearest-point algorithm with the distance from a point to a surface as the error function, performing point cloud fusion one by one based on the radar with the most laser lines to obtain fused point cloud data; performing ground segmentation and point cloud filtering on the fused point cloud data, and using the filtered non-ground point cloud data as the first point cloud data; the point cloud filtering includes self-filtering, which calculates the shape of the steel grabber's robotic arm in real time based on the tilt angle information of the steel grabber's robotic arm, and filters out the point cloud data of the steel grabber's robotic arm. The feature matching module is used to perform feature matching between the first point cloud data and the initial map of the indoor environment to obtain the initial pose estimation information of the steel grabber and its position information in the initial map of the indoor environment. The odometry module is used to take the initial pose estimation information as the initial value of the radar odometry and obtain the inter-frame pose estimation information through nonlinear optimization. The mapping module is used to update the pre-built ground segmentation model based on the model parameter with the most interior points, and to transmit the fused point cloud data to the updated ground segmentation model for ground segmentation.

Images