A gate machine multi-source point cloud fusion method based on grab positioning and a grab direct control method

Abstract

This invention discloses a multi-source point cloud fusion method for gantry cranes based on grab bucket positioning and a direct control method for the grab bucket, belonging to the field of gantry crane automation control technology. The multi-source point cloud includes a first laser point cloud for modeling the grab bucket and materials, and a second laser point cloud for modeling the working area. The fusion method includes: transforming the first laser point cloud to a ground coordinate system using a ground point cloud detection algorithm; obtaining the relationship between the ground coordinate system and the gantry crane grab bucket positioning system coordinate system based on real-time collected grab bucket position information, and transforming the first laser point cloud in the ground coordinate system to the gantry crane grab bucket positioning system coordinate system; and registering and fusing the second laser point cloud and the first laser point cloud transformed to the gantry crane grab bucket positioning system coordinate system. This invention can align the point cloud coordinate system with the gantry crane grab bucket positioning system coordinate system in real time, solving the coordinate system deviation problem caused by the reliance on static structural parameters in existing technologies.

Description

Technical Field

[0001] This invention belongs to the field of gantry crane automation control technology, specifically relating to a gantry crane multi-source point cloud fusion method based on grab bucket positioning and a grab bucket direct control method. Background Technology

[0002] In the intelligent grasping operations of gantry cranes, to achieve intelligent operations (such as automatic grasping and obstacle avoidance), it is necessary to acquire real-time 3D data of the working environment (such as ship holds and cargo stacks) using point cloud scanners. Currently, gantry cranes typically install laser point cloud scanners at the top of the grab bucket and below the operator's cab: the scanner at the top of the grab bucket acquires high-precision point clouds, mainly used to detect the shape and height of the point cloud of materials within the ship hold; the scanner below the operator's cab covers a large working area, but its accuracy is lower due to the distance from the target. Multiple gantry cranes are usually used in combination for scanning, primarily to scan the entire ship and obtain ship information.

[0003] Chinese patent CN114581606A discloses a method for modeling the hold and materials inside a bulk cargo gantry crane based on 3D LiDAR. It involves installing 3D LiDARs on both the gantry crane's rotating body and the "elephant trunk" head to acquire global and local point clouds, respectively. The method transforms the coordinate systems of each LiDAR: based on the gantry crane's mechanical design parameters, a static rotation and translation matrix is ​​preset to transform the radar point cloud of the rotating body to the gantry crane's base coordinate system. A rotary encoder is used to read the rotation mechanism angle θ, transforming the base coordinate system point cloud to the theoretical rotating coordinate system. Similarly, based on the design parameters, the static transformation matrix of the "elephant trunk" head radar to the base coordinate system is estimated and transformed to the theoretical rotating coordinate system. Finally, the ICP algorithm is used to perform coarse and fine registration of the point clouds of the two LiDARs in the rotating coordinate system, merging the "elephant trunk" head point cloud into the coordinate system of the rotating body point cloud, achieving point cloud fusion and unification.

[0004] However, the aforementioned existing technologies have the following shortcomings:

[0005] (1) The coordinate system relies on static structural design parameters to realize the transformation from point cloud coordinates to the theoretical rotary coordinate system. Dynamic errors caused by mechanical deformation and sensor accuracy are ignored, making the theoretical rotary coordinate system independent of the coordinate system of the gantry crane grab positioning system used for positioning and path planning.

[0006] (2) Point cloud data needs to undergo multi-level static matrix transformation (radar-base-rotation). Each transformation step carries the risk of error accumulation, affecting the accuracy of coordinate system transformation. Summary of the Invention

[0007] This invention addresses the shortcomings of existing technologies by providing a multi-source point cloud fusion method for gantry cranes based on grab bucket positioning and a direct grab bucket control method. This method can directly and accurately convert point cloud data collected by multiple scanners into the gantry crane grab bucket positioning coordinate system and dynamically align it with the real-time pose of the grab bucket. This enables direct and rapid mapping from the target position detected by the point cloud to the motion control parameters of the gantry crane, solving the coordinate system deviation problem caused by the reliance on static structural parameters in existing technologies.

[0008] This invention provides the following technical solution:

[0009] Firstly, a method for fusing multi-source point clouds of gantry cranes based on grab bucket positioning is provided. The multi-source point cloud includes a first laser point cloud for modeling the grab bucket and materials, and a second laser point cloud for modeling the work area. The method for fusing multi-source point clouds of gantry cranes based on grab bucket positioning includes:

[0010] The first laser point cloud is transformed to a ground coordinate system using a ground point cloud detection algorithm;

[0011] Based on the real-time collected grab bucket position information, the relationship between the ground coordinate system and the gantry crane grab bucket positioning system coordinate system is obtained, and the first laser point cloud in the ground coordinate system is transformed to the gantry crane grab bucket positioning system coordinate system.

[0012] The second laser point cloud and the first laser point cloud, converted to the coordinate system of the gantry crane grab positioning system, are registered and fused.

[0013] Optionally, the scanner for acquiring the first laser point cloud is installed on the head of the elephant trunk and directly above the grab bucket, while the scanner for acquiring the second laser point cloud is installed below the driver's cab.

[0014] Optionally, the step of transforming the first laser point cloud to a ground coordinate system using a ground point cloud detection algorithm specifically involves:

[0015] The largest plane is segmented from the first laser point cloud using the RANSAC algorithm and used as the ground for detection;

[0016] Obtain the rotation matrix R from the detected ground normal vector to the actual ground normal vector. ground ;

[0017] Based on the positions of the wire rope and grab bucket in the first laser point cloud, a preset installation position offset matrix t is established. offset Then, the first laser point cloud P transformed to the ground coordinate system is obtained. ground ;

[0018] P ground =R ground ·P lidar +t offset ;

[0019] Among them, P lidar This is the original first laser point cloud.

[0020] Optionally, the rotation matrix R is obtained. ground The specific method is as follows:

[0021] Obtain the angle α between the detected ground normal vector and the actual ground normal vector. If the angle α lies within a defined interval of the same direction, then the rotation matrix R... ground The identity matrix is ​​used; if the included angle α is within the set opposite direction interval, then a 180° rotation is performed using a preset reference axis to obtain the rotation matrix R. ground If the included angle α is outside the defined intervals in the same and opposite directions, the rotation axis is calculated using the cross product, and the rotation angle is calculated using the dot product to obtain the rotation matrix R. ground .

[0022] Optionally, the step of obtaining the relationship between the ground coordinate system and the gantry crane grab positioning system coordinate system based on the real-time collected grab position information, and transforming the first laser point cloud in the ground coordinate system to the gantry crane grab positioning system coordinate system, specifically involves:

[0023] Real-time acquisition of the position information of the gantry crane grab bucket, including rotation angle β, trunk length L, and lifting height h;

[0024] Based on the positional relationship between the scanner collecting the first laser point cloud and the grab bucket, the vertical distance h between the ground coordinate system and the gantry crane grab bucket positioning system coordinate system is obtained. The first laser point cloud in the ground coordinate system is then transformed to the gantry crane grab bucket positioning system coordinate system, specifically as follows:

[0025]

[0026] Among them, P m The first laser point cloud in the coordinate system of the gantry crane grab positioning system, (X m ,Y m Z m Let R be the three-dimensional coordinates of the midpoint m of the first laser point cloud in the coordinate system of the gantry crane grab positioning system. z (β) is the rotation matrix about the z-axis corresponding to the gantry crane rotation angle β, (X ground ,Y ground Z ground P is the first laser point cloud transformed to the ground coordinate system. ground The three-dimensional coordinates of the midpoint.

[0027] Optionally, the registration and fusion of the second laser point cloud and the first laser point cloud, converted to the coordinate system of the gantry crane grab positioning system, specifically involves:

[0028] The first laser point cloud of the gantry crane grab positioning system coordinate system is used as the target point cloud, and the second laser point cloud is used as the source point cloud. The FastICP algorithm is used to perform three-level registration of the target point cloud and the source point cloud in a coarse-to-fine order. During the three-level registration process, the distance threshold in the FastICP algorithm is gradually tightened, and the transformation result of the previous level is automatically inherited as the initial pose during each level of iterative registration.

[0029] Secondly, a method for direct control of a grab bucket is provided, based on the grab bucket positioning-based multi-source point cloud fusion method for gantry cranes as described in any one of the first aspects, including:

[0030] After the second laser point cloud and the first laser point cloud in the coordinate system of the gantry crane grab positioning system are registered and fused, the fused point cloud P is obtained. combine ;

[0031] For the fused point cloud P combine The target position coordinates (X) of the grab bucket movement combine ,Y combine Z combine Perform polar coordinate decomposition to obtain the gantry crane's rotation angle θ = arctan(Y). combine ,X combine ) and radius of gyration The slewing angle of the gantry crane is the target rotation angle;

[0032] Based on the kinematic mapping relationship of the gantry crane mechanism, the luffing amplitude L of the gantry crane is the turning radius r, and the lifting height h of the grab bucket is the vertical height Z of the target position of the grab bucket. combine ;

[0033] Based on the amplitude of the change, the lifting height, and the target rotation angle, control commands for the grab bucket's movement are generated to drive the gantry crane's actuators.

[0034] Thirdly, a computer device is provided, including a processor and a memory; wherein, when the processor executes a computer program stored in the memory, it implements the steps of the gantry crane multi-source point cloud fusion method based on grab bucket positioning as described in any one of the first aspects.

[0035] Fourthly, a computer-readable storage medium is provided for storing a computer program; when the computer program is executed by a processor, it implements the steps of the gantry crane multi-source point cloud fusion method based on grab bucket positioning as described in any one of the first aspects.

[0036] Compared with the prior art, the beneficial effects of the present invention are:

[0037] (1) By acquiring the current position parameters of the grab bucket, the point cloud of the elephant trunk head is accurately converted from the point cloud coordinate system to the coordinate system of the gantry crane grab bucket positioning system, which solves the problem of coordinate system deviation caused by the reliance on static structural parameters in the existing technology, and realizes real-time alignment between the point cloud coordinate system and the coordinate system of the gantry crane grab bucket positioning system.

[0038] (2) A multi-scale registration framework (three-level threshold tightening and hierarchical iteration strategy) is adopted to remove the FPFH feature calculation and RANSAC global registration links in the traditional ICP. Through dynamic threshold adjustment and iterative parameter optimization, the point cloud registration efficiency is improved by 3-5 times while ensuring registration accuracy. This solves the problems of slow registration speed and high computing power requirements in the existing technology, and realizes the rapid fusion and unification of point clouds from multiple sensors.

[0039] (3) By decomposing the three-dimensional coordinates of the target position into rotation angle θ and rotation distance r through polar coordinate decomposition, combined with the kinematic mapping relationship of the gantry crane, the rotation distance r is the amplitude parameter L, and the vertical height of the target position of the grab bucket is the lifting height h. The amplitude, rotation and lifting control commands are directly generated, realizing the conversion of target coordinates to control commands with fewer levels and higher precision, solving the problem of multiple conversion levels and large cumulative error in the existing technology. Attached Figure Description

[0040] Figure 1 This is an overall flowchart of the gantry crane multi-source point cloud fusion method based on grab bucket positioning of the present invention;

[0041] Figure 2 This is a schematic diagram of the installation position of the 3D laser scanner for acquiring the second laser point cloud according to the present invention;

[0042] Figure 3 This is a schematic diagram of the installation position of the 3D laser scanner for acquiring the first laser point cloud according to the present invention;

[0043] Figure 4 This is a schematic diagram of the registration process between the first laser point cloud and the second laser point cloud of the present invention;

[0044] Figure 5 This is a flowchart of the grab bucket direct control method of the present invention. Detailed Implementation

[0045] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be used to limit the scope of protection of the present invention. It should be noted that the term "comprising" and any variations thereof in the specification, claims and the above-mentioned drawings of the present invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or devices.

[0046] Before describing the embodiments of the present invention, the multi-source point cloud of this application will be further described:

[0047] The multi-source point cloud includes a first laser point cloud for modeling the grab and material, and a second laser point cloud for modeling the work area. The first and second laser point clouds are generated using two 3D laser scanners.

[0048] like Figure 2 As shown, the 3D laser scanner for acquiring the second laser point cloud is installed below the driver's cab, covering a large operating area. With the origin of the current coordinate system as the center and the z-axis as the rotation axis of the pan-tilt unit, the x-axis of the scanner is defined as the 0° reference direction for the scanning azimuth angle. It is stipulated that when the scanner rotates clockwise around the z-axis from the x-axis to the y-axis (i.e., rotates towards the rear of the driver's cab), the direction of the angle is positive, and when it rotates counterclockwise (i.e., rotates towards the front of the driver's cab), the direction of the angle is negative.

[0049] like Figure 3 As shown, the 3D laser scanner for collecting the first laser point cloud is installed near the head of the elephant trunk near the grab bucket. Its installation height is the distance from the head of the elephant trunk to the ground. It is mainly used to detect the shape and height of the material point cloud in the cabin. The 3D laser scanner forms the scanning field of view with the radar as the symmetrical conical scanning area. With the center of the equipment as the origin, the positive z-axis is perpendicular to the ground and pointing downwards, the y-axis is parallel to the gantry crane boom and points to the land side, and the right-hand rule is used to determine the x-axis direction.

[0050] Example 1

[0051] like Figure 1 As shown, a method for multi-source point cloud fusion of gantry cranes based on grab bucket positioning is provided, including the following steps:

[0052] Step S1: Transform the first laser point cloud to the ground coordinate system using a ground point cloud detection algorithm.

[0053] By using a ground point cloud detection algorithm, the origin coordinates of the point cloud data are aligned from the top of the elephant trunk to the ground directly below the elephant trunk. Then, the x-axis and y-axis are flipped to ensure that the direction of the coordinate system is consistent with the direction of the base coordinate system (with the rotation center of the gantry crane as the origin, the z-axis perpendicular to the horizontal plane upwards, the y-axis perpendicular to the wharf shoreline pointing towards the sea, and the x-axis direction determined by the right-hand rule).

[0054] In this embodiment, step S1 specifically includes:

[0055] Step S11: Use the RANSAC algorithm to segment the largest plane from the first laser point cloud as the detection ground.

[0056] The RANSAC algorithm can refer to existing technologies. Specifically, algorithm parameters, including distance threshold, number of sampling points, and number of iterations, can be preset based on expert experience.

[0057] Step S12: Obtain the rotation matrix R from the detected ground normal vector to the actual ground normal vector. ground .

[0058] Calculate the normal vector of the detected plane and normalize it. Ensure that the point cloud orientation is upward (Z component > 0), which conforms to the physical assumption that the ground is upward.

[0059] Step S12 specifically involves obtaining the angle α between the detected ground normal vector and the actual ground normal vector.

[0060] n1 and n2 are the detected ground normal vector and the actual ground normal vector, respectively.

[0061] 1) If the included angle α lies within the defined interval in the same direction, then the rotation matrix R ground It is an identity matrix.

[0062] 2) If the included angle α is within the set opposite direction range, then rotate 180° using the preset reference axis to obtain the rotation matrix R. ground .

[0063] The preset reference axis can be a fixed axis (e.g., the Z-axis) or any axis u perpendicular to both normal vectors, rotated 180°, with the Z-axis selected as the reference axis. When axis u is selected as the reference axis, R is obtained. ground =2uu T -I.

[0064] 3) If the included angle α is outside the defined intervals in the same and opposite directions, the rotation axis is calculated by cross product, and the rotation angle is calculated by dot product to obtain the rotation matrix R. ground .

[0065] Specifically, calculate the axis of rotation:

[0066] Generate the rotation matrix using Rodriguez's formula: R ground =I + sinα·[u] x +(1-cosα)·[u] x 2 ;[u] x It is the cross product matrix of the rotation axis u.

[0067] Apply the above transformation to all points to achieve point cloud rotation. Calculate the centroid of the ground points after rotation, and translate the point cloud along the z-axis so that the ground centroid lies in the plane Z=0.

[0068] It is worth noting that: due to the fixed installation orientation, the ground coordinate system O may be... ground The x and y axes are opposite to the directions of the rotary coordinate system. The ground point cloud is then flipped to obtain a ground point cloud with the same orientation as the rotary coordinate system.

[0069] Step S13: Based on the positions of the wire rope and grab bucket in the first laser point cloud, preset the installation position offset matrix t. offset Then, the first laser point cloud P transformed to the ground coordinate system is obtained. ground ;

[0070] P ground =R ground ·P lidar +t offset ;

[0071] Among them, P lidar This is the original first laser point cloud.

[0072] Step S2: Based on the real-time collected grab position information, obtain the relationship between the ground coordinate system and the gantry crane grab positioning system coordinate system, and convert the first laser point cloud in the ground coordinate system to the gantry crane grab positioning system coordinate system.

[0073] Step S2 is as follows:

[0074] Step S21: Collect the position information of the gantry crane grab bucket in real time, including the rotation angle β, the length of the trunk beam L, and the lifting height h.

[0075] Step S22: Based on the positional relationship between the scanner collecting the first laser point cloud and the grab bucket, the vertical distance h between the ground coordinate system and the gantry crane grab bucket positioning system coordinate system is obtained. The first laser point cloud in the ground coordinate system is then transformed to the gantry crane grab bucket positioning system coordinate system, specifically as follows:

[0076]

[0077] Among them, P mThe first laser point cloud in the coordinate system of the gantry crane grab positioning system, (X m ,Y m Z m Let R be the three-dimensional coordinates of the midpoint m of the first laser point cloud in the coordinate system of the gantry crane grab positioning system. z (β) is the rotation matrix about the z-axis corresponding to the gantry crane rotation angle β, (X ground ,Y ground Z ground P is the first laser point cloud transformed to the ground coordinate system. ground The three-dimensional coordinates of the midpoint.

[0078] Since the 3D laser scanner on the elephant's trunk head is located directly above the grab bucket, the coordinate system O after transformation through the ground coordinate system... ground Since the position of the grab bucket differs from its location by only one lifting height h, the point cloud can be aligned with the precise positioning coordinate system of the gantry crane using the grab bucket's pose information, thus obtaining the point cloud P. m .

[0079] Step S3: Register and fuse the second laser point cloud with the first laser point cloud converted to the coordinate system of the gantry crane grab positioning system.

[0080] Specifically, the first laser point cloud of the gantry crane grab positioning system coordinate system is used as the target point cloud, and the second laser point cloud is used as the source point cloud. The FastICP algorithm is used to perform three-level registration of the target point cloud and the source point cloud in a coarse-to-fine order. During the three-level registration process, the distance threshold in the FastICP algorithm is gradually tightened, and the transformation result of the previous level is automatically inherited as the initial pose during each level of iterative registration.

[0081] After registration, the second laser point cloud and the first laser point cloud, which have been converted to the coordinate system of the gantry crane grab positioning system, will be merged and fused. The specific method can refer to existing technologies, which involve steps such as noise reduction, stitching, and optimization.

[0082] like Figure 4 As shown, the process begins with the preprocessing stage of the input point cloud, which requires simultaneous processing of point cloud P. m (As the target point cloud, its coordinate system is the gantry crane grab positioning system coordinate system O) location The point cloud from the scanner below the driver's cab and the point cloud from the scanner below the driver's cab (used as the source point cloud) are processed. Voxel downsampling is used to reduce the computational complexity of both (especially for potentially high-density redundant points in the scanner point cloud), and a fast normal estimation algorithm is employed to balance accuracy and speed in data preparation. The preprocessed point cloud then enters the multi-scale ICP registration stage, which proceeds sequentially at three scales. The core objective is to solve the point cloud P from the point cloud from the scanner below the driver's cab. m Transformation matrix of coordinate system:

[0083] The first stage performs coarse registration with a lenient threshold of 3 times the original distance, achieving initial pose alignment after 100 iterations. The second stage tightens the threshold by 60%, achieving medium-precision registration through 100 iterations. The third stage further tightens the threshold by 60% and performs 200 sufficient iterations to achieve high-precision registration. Each stage automatically inherits the transformation result from the previous stage as the initial pose, and the convergence condition is set to a relative tolerance of 1e-6 to avoid over-iteration. Finally, the registration process outputs a point cloud from the scanner below the driver's cab to a point cloud P. m The final transformation matrix of the coordinate system, and the coordinates after transformation with the point cloud P. m Precisely aligned scanner point cloud P n This allows for the fusion of the two within the same coordinate system.

[0084] FastICP employs multi-scale registration as its core strategy, gradually improving accuracy through a three-level registration process from coarse to fine. This is complemented by a dynamic threshold adjustment mechanism—the distance threshold is tightened progressively by 60% from lenient to strict. Simultaneously, the iteration count is optimized: the first two levels each use 100 iterations to quickly complete the initial alignment, while the final level uses 200 iterations to ensure sufficient accuracy. To overcome computational bottlenecks, the solution directly eliminates the traditional FPFH feature calculation and RANSAC global registration stages, significantly reducing time complexity.

[0085] The multi-scale registration framework effectively avoids local optima and improves convergence speed by progressively tightening the threshold; the iterative strategy reduces the overall time consumption while ensuring the final accuracy by allocating the number of iterations in a hierarchical manner; the complex FPFH feature extraction and RANSAC feature matching processes are eliminated, and the normal estimation is simplified to further reduce overhead and eliminate computational bottlenecks; in addition, the parameter dynamic adjustment mechanism of this application achieves efficient coordination of the entire process through threshold reduction, pose inheritance and adaptive convergence conditions.

[0086] In terms of performance improvement, this application reduces the time complexity from O(n) of the original algorithm. 2 The multi-scale ICP speedup is reduced from O(nlogn) + O(kn) to only O(kn), achieving a speedup of 3-5 times in actual tests. Memory usage is significantly reduced by eliminating the storage of the FPFH feature matrix and reducing the overhead of k-nearest neighbor search. Robustness is enhanced by the low sensitivity of the multi-scale strategy to the initial pose, the adaptability of the dynamic threshold to the degree of overlap, and the avoidance of local optima by the stepwise optimization, achieving a balanced improvement in speed and accuracy.

[0087] Example 2

[0088] like Figure 5 As shown, a direct control method for a grab bucket, based on the grab bucket positioning-based gantry crane multi-source point cloud fusion method in Example 1, includes the following steps:

[0089] Step D1: After registering and fusing the second laser point cloud with the first laser point cloud converted to the coordinate system of the gantry crane grab positioning system, the fused point cloud P is obtained. combine .

[0090] Step D2: For the fused point cloud P combine The target position coordinates (X) of the grab bucket movement combine ,Y combine Z combine Perform polar coordinate decomposition to obtain the gantry crane's rotation angle θ = arctan(Y). combine ,X combine ) and radius of gyration The slewing angle of the gantry crane is the target rotation angle.

[0091] Step D3: Based on the kinematic mapping relationship of the gantry crane mechanism, the luffing amplitude L of the gantry crane is the turning radius r, and the lifting height h of the grab bucket is the vertical height Z of the target position of the grab bucket movement. combine .

[0092] The kinematic mapping relationship of the gantry crane mechanism is based on existing technology.

[0093] Step D4: Based on the luffing amplitude, lifting height, and target rotation angle, generate control commands for the grab bucket's movement to drive the gantry crane's actuator.

[0094] This application first uses the point cloud data from the lidar on the gantry crane's head to transform its coordinate system origin to the ground using a ground point cloud detection algorithm. Then, it acquires the real-time positioning data of the grab bucket (rotation angle θ, luffing distance L, and lifting height h as parameters). Based on the grab bucket's current position, a dynamic correction matrix is ​​generated to achieve the transformation from the gantry crane's head point cloud to the grab bucket's positioning coordinate system. This solves the coordinate system deviation problem caused by existing technologies relying on static structural parameters, achieving real-time alignment between the point cloud coordinate system and the gantry crane grab bucket positioning system coordinate system.

[0095] This application employs the FastICP algorithm to rapidly fuse and register the point cloud from the scanner beneath the driver's cab onto the calibrated coordinate system. This solves the problems of slow registration speed and high computational requirements in existing technologies, achieving rapid fusion and unification of point clouds from multiple sensors.

[0096] This application uses the target position detected by the fused point cloud to convert it into a grab bucket position control command through the gantry crane motion parameter mapping model. This achieves a low-level, high-precision conversion from target coordinates to control commands, solving the problem of multiple conversion levels and large cumulative errors in the prior art.

[0097] Example 3

[0098] The present invention provides a computer device, including a processor and a memory; wherein, when the processor executes the computer program stored in the memory, it implements the steps of the above-described gantry crane multi-source point cloud fusion method based on grab bucket positioning.

[0099] For more detailed information on the above methods, please refer to the relevant content disclosed in the foregoing embodiments, which will not be repeated here.

[0100] Example 4

[0101] The present invention provides a computer-readable storage medium for storing a computer program; when the computer program is executed by a processor, it implements the steps of the above-described gantry crane multi-source point cloud fusion method based on grab bucket positioning.

[0102] For more detailed information on the above methods, please refer to the relevant content disclosed in the foregoing embodiments, which will not be repeated here.

[0103] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. The devices and storage media disclosed in the embodiments are described simply because they correspond to the methods disclosed in the embodiments; relevant parts can be referred to the method section.

[0104] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments of the present invention.

[0105] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A method for multi-source point cloud fusion based on grab positioning of a portal reclaimer, the multi-source point cloud comprising a first laser point cloud for modeling a grab and material and a second laser point cloud for modeling a work area, characterized in that, include: The first laser point cloud is transformed to a ground coordinate system using a ground point cloud detection algorithm; Based on the real-time collected grab bucket position information, the relationship between the ground coordinate system and the gantry crane grab bucket positioning system coordinate system is obtained, and the first laser point cloud in the ground coordinate system is transformed to the gantry crane grab bucket positioning system coordinate system. The second laser point cloud and the first laser point cloud, converted to the coordinate system of the gantry crane grab positioning system, are registered and fused. The scanner for collecting the first laser point cloud is installed on the head of the elephant trunk and directly above the grab bucket, while the scanner for collecting the second laser point cloud is installed below the driver's cab. The process of transforming the first laser point cloud to a ground coordinate system using a ground point cloud detection algorithm specifically involves: The largest plane is segmented from the first laser point cloud using the RANSAC algorithm and used as the ground for detection; Obtain the rotation matrix from the detected ground normal vector to the actual ground normal vector. ; Based on the positions of the wire rope and grab bucket in the first laser point cloud, a preset installation position offset matrix is ​​established. Then, the first laser point cloud transformed to the ground coordinate system is obtained. ; ； in, This represents the original first laser point cloud; The process involves obtaining the relationship between the ground coordinate system and the gantry crane grab positioning system coordinate system based on the real-time collected grab position information, and then transforming the first laser point cloud in the ground coordinate system to the gantry crane grab positioning system coordinate system. Specifically: Real-time acquisition of the position information of the gantry crane grab bucket, including the rotation angle. Elephant Trunk Length and lifting height ; Based on the positional relationship between the scanner collecting the first laser point cloud and the grab bucket, the vertical distance between the ground coordinate system and the gantry crane grab bucket positioning system coordinate system is determined by the lifting height. The first laser point cloud in the ground coordinate system is transformed to the coordinate system of the gantry crane grab positioning system, specifically as follows: in, This represents the first laser point cloud in the coordinate system of the gantry crane grab positioning system. The midpoint of the first laser point cloud in the coordinate system of the gantry crane grab positioning system m The three-dimensional coordinates To the angle of rotation of the gantry crane Corresponding winding z The rotation matrix of the axis. The first laser point cloud transformed to the ground coordinate system The three-dimensional coordinates of the midpoint.

2. The multi-source point cloud fusion method for gantry cranes based on grab bucket positioning according to claim 1, characterized in that, Obtain the rotation matrix The specific method is as follows: Obtain the angle between the detected ground normal vector and the actual ground normal vector. If the included angle If it is located within the defined interval in the same direction, then the rotation matrix It is the identity matrix; If the included angle If the value is located within the specified opposite direction range, then rotate 180° using the preset reference axis to obtain the rotation matrix. ; If the included angle If the rotation is located outside the defined intervals in the same and opposite directions, the rotation axis is calculated using the cross product, and the rotation angle is calculated using the dot product to obtain the rotation matrix. .

3. The multi-source point cloud fusion method for gantry cranes based on grab bucket positioning according to claim 1, characterized in that, The registration and fusion of the second laser point cloud and the first laser point cloud, converted to the coordinate system of the gantry crane grab positioning system, specifically involves: The first laser point cloud of the gantry crane grab positioning system coordinate system is used as the target point cloud, and the second laser point cloud is used as the source point cloud. The FastICP algorithm is used to perform three-level registration of the target point cloud and the source point cloud in a coarse-to-fine order. During the three-level registration process, the distance threshold in the FastICP algorithm is gradually tightened, and the transformation result of the previous level is automatically inherited as the initial pose during each level of iterative registration.

4. A direct control method for a grab bucket, based on the grab bucket positioning-based gantry crane multi-source point cloud fusion method according to any one of claims 1-3, characterized in that, include: After registering and fusing the second laser point cloud with the first laser point cloud in the coordinate system of the gantry crane grab positioning system, a fused point cloud is obtained. ; For fused point clouds Target position coordinates of the middle grab bucket movement Perform polar coordinate decomposition to obtain the gantry crane's rotation angle. and radius of gyration The slewing angle of the gantry crane is the target rotation angle; Based on the kinematic mapping relationship of the gantry crane mechanism, the amplitude of the gantry crane... Radius of gyration The lifting height of the grab bucket Vertical height of the target position for the grab bucket to move ; Based on the amplitude of the change, the lifting height, and the target rotation angle, control commands for the grab bucket's movement are generated to drive the gantry crane's actuators.

5. A computer device, characterized in that, It includes a processor and a memory; wherein, when the processor executes the computer program stored in the memory, it implements the steps of the gantry crane multi-source point cloud fusion method based on grab bucket positioning as described in any one of claims 1-3.

6. A computer-readable storage medium, characterized in that, Used to store computer programs; when the computer programs are executed by the processor, they implement the steps of the gantry crane multi-source point cloud fusion method based on grab bucket positioning as described in any one of claims 1-3.

Images