Curved surface interpolation method and device for automatic operation object of engineering machinery

Abstract

The invention provides a curved surface interpolation method and device for an automatic operation object of engineering machinery, and relates to the technical field of curved surface interpolation optimization, and the method comprises the steps: carrying out the preprocessing of point cloud data in an operation environment of the engineering machinery through a multi-sensor fusion semantic segmentation model, removing noise and interference terms, and guaranteeing the purity of the data. Furthermore, selection of control points is optimized by utilizing a spatial data structure and a geometric analysis method, so that the data volume is reduced, and key information is reserved. Precise modeling is carried out on the surface of an operation object through a radial basis function interpolation algorithm, and the interpolation precision is improved in combination with a dynamic parameter optimization technology. Finally, the generated curved surface model not only provides accurate geometric information for operation planning of engineering machinery, but also provides support for optimization of key operation parameters such as digging force and fullness rate in combination with internal structure characteristics of an operation object. And through effective connection of sensing and planning, the working efficiency and reliability of the engineering machinery under the dynamic working condition are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of surface interpolation optimization, and particularly relates to a surface interpolation method and device for an automatic operation object of construction machinery. Background Art

[0002] In modern infrastructure construction, construction machinery plays an indispensable role and is widely used in multiple fields such as water conservancy, transportation, and energy. However, with the intensification of the labor shortage problem, the demand for automatic operation of construction machinery is increasing day by day. To achieve this goal, construction machinery needs to have the ability to autonomously sense the environment and interact with it, so as to achieve efficient and safe automatic operation.

[0003] At present, although sensor technology can already provide rich environmental information for construction machinery, these raw data cannot be directly used for operation planning. How to effectively convert sensor data into operation instructions has always been a key problem in the field of construction machinery automation. In the prior art, some methods attempt to simulate the operation scenario by establishing a database. For example, the motion trajectory of an excavator is planned by constructing a rectangular dot matrix table and inverse-solving joint angle data. However, these methods have obvious limitations: they mainly rely on a preset database, lack the ability to perceive complex changes in the actual operation environment in real time, and cannot effectively cope with dynamic working conditions.

[0004] In addition, the complexity of the operation object of construction machinery also poses challenges to automatic operation. For example, the surface of the operation object may be covered with interference objects such as weeds and stones, which will affect the accurate perception of the surface characteristics of the operation object. At the same time, the inhomogeneity of the internal structure of the operation object (such as the distribution of substances with different densities) will also affect the calculation of the excavation force, resulting in unexpected impacts during the excavation process. The prior art often ignores these factors and cannot comprehensively consider the surface and internal information of the operation object, thus limiting the accuracy and reliability of automatic operation.

[0005] In terms of data processing, the existing methods also have deficiencies. For example, when traditional surface interpolation methods process sparse key information, they often cannot accurately extract key points, resulting in inaccurate interpolation results. In addition, when existing methods optimize parameters, they lack consideration of the actual characteristics of the operation object, resulting in low efficiency and insufficient stability in the optimization process.

[0006] In view of this, the present application is proposed. Summary of the Invention

[0007] The present invention provides a surface interpolation method and device for an automatic operation object of construction machinery, which can at least partially improve the above problems.

[0008] To achieve the above object, the present invention adopts the following technical solutions:

[0009] A surface interpolation method for an automatic operation object of construction machinery, comprising:

[0010] Obtain the environmental information of the operation object collected by a preset sensor component, distinguish the environmental information in detail according to a semantic segmentation model, and obtain the point cloud information of the operation object therefrom;

[0011] Preprocess the point cloud information of the operation object to generate interpolation points to be interpolated, and screen the interpolation points to be interpolated according to a quadratic surface formula to obtain control points;

[0012] Construct an RBF interpolation model based on the control points, and perform multi-objective parameter optimization on the approximation parameter of the RBF interpolation model and the number of control points according to preset optimization parameters to obtain an interpolation surface output by the RBF interpolation model;

[0013] Perform operation planning on the interpolation surface to obtain a planned excavation path with optimized indicators, and end the current excavation operation.

[0014] The present invention also provides a surface interpolation device for an automatic operation object of construction machinery, comprising:

[0015] A data segmentation unit for obtaining the environmental information of the operation object collected by a preset sensor component, distinguishing the environmental information in detail according to a semantic segmentation model, and obtaining the point cloud information of the operation object therefrom;

[0016] A preprocessing and screening unit for preprocessing the point cloud information of the operation object to generate interpolation points to be interpolated, and screening the interpolation points to be interpolated according to a quadratic surface formula to obtain control points;

[0017] A surface output unit for constructing an RBF interpolation model based on the control points, and performing multi-objective parameter optimization on the approximation parameter of the RBF interpolation model and the number of control points according to preset optimization parameters to obtain an interpolation surface output by the RBF interpolation model;

[0018] An operation planning unit for performing operation planning on the interpolation surface to obtain a planned excavation path with optimized indicators, and ending the current excavation operation.

[0019] In summary, the surface interpolation method for the automatic operation object of construction machinery aims to solve problems such as the disconnection between perception and planning, sparse surface features of the operation object, and frequent working condition changes in the existing technology of automatic operation of construction machinery. This method conducts refined preprocessing on the point cloud data in the working environment of construction machinery, effectively removing noise points and interference items to ensure the accuracy and reliability of the interpolated data. Further, by optimizing the selection strategy of control points, while reducing the data volume, key information is retained, and the interpolation efficiency is improved. The radial basis function (RBF) surface interpolation algorithm is used to accurately model the target object, and the interpolation accuracy is improved through dynamic parameter optimization technology. Finally, the generated surface model not only provides accurate geometric information for the operation planning of construction machinery, but also combines the internal structural features of the operation object to support the optimization of key operation parameters such as excavation force and full bucket ratio.

[0020] Furthermore, the core of the surface interpolation method for the automatic operation object of construction machinery lies in optimizing the entire process from the perception front-end to the planning back-end in combination with the actual requirements of the construction machinery operation scenario. Through the fine processing in the preprocessing stage, noise and interference items in the point cloud data are effectively screened out; in the control point selection stage, the selection strategy of control points is optimized, reducing the calculation amount and improving the efficiency; in the optimization stage, the preprocessing information is used as the weight of the optimization parameters, improving the optimization rate and reducing the complexity; in the data interaction stage, the preprocessing information is combined to guide the optimization calculation of subsequent steps, providing strong support for the autonomous decision-making and operation of construction machinery. Through the present invention, construction machinery can achieve the effective connection between front-end perception and back-end planning, significantly improving the accuracy and efficiency of automatic operation and adapting to complex and changeable working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a schematic flowchart of a surface interpolation method for the automatic operation object of construction machinery provided by the first embodiment of the present invention;

[0022] Figure 2 is a schematic overall flowchart of a surface interpolation method for the automatic operation object of construction machinery provided by the first embodiment of the present invention;

[0023] Figure 3 is a schematic module diagram of a surface interpolation device for the automatic operation object of construction machinery provided by the second embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0025] ReferenceFigure 1 , Figure 2 As shown in Figure 2 , the first embodiment of the present invention discloses a surface interpolation method for an automatic operation object of construction machinery, which can be executed by a surface interpolation device for an automatic operation object of construction machinery (hereinafter referred to as the interpolation device), and particularly, by one or more processors in the interpolation device to implement the following method:

[0026] S1. Obtain the environmental information of the operation object collected by a preset sensor component, distinguish the environmental information in detail according to the semantic segmentation model, and obtain the point cloud information of the operation object from it;

[0027] Specifically, step S1 includes: obtaining the environmental information of the operation object collected by a preset sensor component, where the sensor component includes a lidar and a depth camera;

[0028] Perform semantic segmentation processing on the environmental information based on the PMF semantic segmentation model to generate the point cloud information of the operation object, where the point cloud information of the operation object includes soil heap point cloud and soil slope point cloud.

[0029] In this embodiment, in the actual operation scenario, construction machinery needs to accurately perceive and model the operation object (such as soil heap, soil slope, etc.) to achieve automatic operation. First, the environmental information of the operation object is collected through a preset sensor component, and the sensor component includes a lidar and a depth camera. The lidar can provide high-precision distance measurement data, while the depth camera can directly obtain the depth information of each pixel point in the scene. By fusing the data of these two sensors, richer and more accurate three-dimensional environmental information can be obtained. Next, the PMF (Perception-Aware Multi-Sensor Fusion) semantic segmentation model is used to perform semantic segmentation processing on the collected environmental information. The PMF model can effectively fuse data from different sensors and improve the robustness and accuracy of scene segmentation. Through this model, the operation object can be accurately separated from the complex background environment, and the point cloud information of the operation object is generated, including the soil heap point cloud and the soil slope point cloud. This process not only provides an accurate data basis for subsequent surface interpolation, but also reduces the interference of noise and irrelevant information through semantic segmentation, improving the quality and usability of the data.

[0030] S2. Preprocess the point cloud information of the operation object to generate interpolation points to be interpolated, and screen the interpolation points to be interpolated according to the quadratic surface formula to obtain control points;

[0031] Specifically, step S2 includes: constructing a KD-Tree data structure according to the point cloud information of the operation object, and calculating the KD-Tree data structure to calculate the local reachable distance and the local reachable density;

[0032] Calculate based on the local reachability distance and local reachability density to generate a local outlier factor LOF, and judge the local outlier factor LOF;

[0033] Filter out the local outlier factor LOF greater than the preset value. Based on the KD-Tree data structure and the PCA principal component analysis method, calculate the normal vector of the remaining local outlier factor LOF after filtering, and standardize the coordinates of the point cloud corresponding to the remaining local outlier factor LOF, and calculate the covariance matrix of the standardized data matrix, where the preset value is 1;

[0034] Solve the covariance matrix to obtain eigenvalues and eigenvectors, calculate the change rate of the normal vector according to the eigenvalues and eigenvectors, and screen out the point cloud with a change rate higher than the preset change value to obtain the points to be interpolated.

[0035] Substitute the coordinates of each point in the points to be interpolated into the preset quadratic surface formula to construct a design matrix and a target vector, and solve the interpolation coefficient process for the design matrix and the target vector according to the least squares method to obtain a quadratic surface equation;

[0036] Take the derivative of the quadratic surface equation to obtain the elements of the Hessian matrix, and calculate the elements of the Hessian matrix to generate the principal curvature k1 and the principal curvature k2;

[0037] Calculate the mean curvature according to the principal curvature k1 and the principal curvature k2, and classify the points to be interpolated into edge points and internal points based on the mean curvature;

[0038] Screen the internal points through a grid, select the internal point with the highest mean curvature, and use the selected internal points and the edge points as control points.

[0039] Preferably, edge points are selected from the points to be interpolated through a preset equal interval, where one nth of the length of each side of the interpolation surface is used as the equal interval for each side, and n is a parameter determined according to the size of the interpolation object.

[0040] In this embodiment, the point cloud information of the operation object (such as a soil pile, a soil slope, etc.) is collected by a high-precision sensor. These point cloud data contain rich environmental information, but may also be mixed with noise and outliers. These interference factors will seriously affect subsequent surface modeling and operation planning. Therefore, preprocessing the point cloud data is a crucial step. The purpose of preprocessing is to remove noise and interference terms, and at the same time screen out the points that play a key role in surface modeling, that is, control points.

[0041] First, use the KD-Tree data structure to partition the spatial information of the job object point cloud. The KD-Tree is an efficient data structure that can quickly perform spatial queries and neighborhood searches. By constructing the KD-Tree, the local reachability distance and local reachability density of each point can be calculated. These two parameters are the basis for subsequent calculation of the local outlier factor (LOF). The local outlier factor is an effective method for detecting outliers in a dataset. It determines whether a point is an outlier by comparing the density of a point with that of its neighboring points. In this method, by calculating the local reachability distance and local reachability density, the local outlier factor LOF of each point is further generated. For points with LOF values greater than the preset value of 1, they are considered outliers and should be filtered out. This process effectively removes the noise points caused by sensor errors or environmental disturbances (such as dust, etc.), improving the quality of the data.

[0042] After filtering out the outliers, based on the KD-Tree data structure and the PCA principal component analysis method, the normal vectors of the remaining points are calculated. PCA is a commonly used method for dimensionality reduction and feature extraction, which can effectively analyze the principal component directions of the data. By using PCA, the normal vector of each point is calculated, and the coordinates of these points are standardized. The standardized data matrix can be used to calculate the covariance matrix. The covariance matrix reflects the correlation of the data in each dimension. By solving the covariance matrix, eigenvalues and eigenvectors are obtained. These eigenvalues and eigenvectors are further used to calculate the change rate of the normal vector. The change rate of the normal vector can reflect the geometric change of the point cloud data in the local area. In this embodiment, the point cloud with a change rate higher than the preset change value is screened out. These points are usually caused by surface irregularities (such as weeds, stones, etc.). Through this screening process, the points to be interpolated are obtained, and these points will be used for subsequent surface modeling.

[0043] Next, substitute the coordinates of each point in the points to be interpolated into the preset quadratic surface formula. The quadratic surface formula is a commonly used mathematical model that can interpolate complex surface shapes well. By constructing a design matrix and a target vector, and using the least squares method to solve these matrices and vectors, the interpolation coefficients of the quadratic surface equation can be obtained. The least squares method is a commonly used optimization method that can minimize the interpolation error and thus obtain the optimal interpolation coefficients. After obtaining the quadratic surface equation, take its derivative to obtain the elements of the Hessian matrix. The Hessian matrix contains the second-order derivative information of the surface and can reflect the curvature change of the surface. By calculating the elements of the Hessian matrix, the principal curvature k1 and the principal curvature k2 can be obtained. The principal curvature is the maximum and minimum curvature values at a certain point on the surface and can well describe the geometric characteristics of the surface. According to the principal curvatures k1 and k2, calculate the mean curvature. The mean curvature is the average of the principal curvatures and can more intuitively reflect the degree of bending of the surface. Based on the mean curvature, distinguish the points to be interpolated into edge points and interior points. Edge points usually have a higher curvature, while the curvature of interior points is relatively low. Edge points are crucial for describing the contour of the object to be operated on, so special attention needs to be paid. Interior points reflect the internal structure of the object to be operated on.

[0044] To further optimize the selection of control points, a grid screening method is used to process the interior points. Divide the surface of the object to be operated on into multiple grids, and select the interior point with the highest mean curvature in each grid. These interior points can better reflect the internal characteristics of the object to be operated on. At the same time, select edge points from the points to be interpolated at preset equal intervals. The equal-interval selection method can ensure that the edge points are evenly distributed on the contour of the object to be operated on, thus completely reflecting the shape of the object to be operated on. Take one nth of the length of each side of the interpolation surface as the equal interval for each side, where n is a parameter determined according to the size of the interpolation object. By adjusting the value of n, the density of the edge points can be controlled, so as to achieve a balance between accuracy and computational efficiency. Finally, the selected interior points and edge points are used as control points for subsequent surface interpolation.

[0045] S3. Construct an RBF interpolation model based on the control points, and perform multi-objective parameter optimization on the approximation parameter of the RBF interpolation model and the number of control points according to preset optimization parameters to obtain an interpolation surface output by the RBF interpolation model;

[0046] Specifically, step S3 includes: screening a basis function model, inputting the control points into the basis function model, constructing an RBF surface interpolation equation, and solving the weight coefficients of the equation to obtain an RBF interpolation model;

[0047] Set the running time, residual standard deviation, mean square error, coefficient of determination, and average smoothness according to preset parameters and use them as evaluation indicators;

[0048] Obtain optimization parameters, take the step size and smoothing rate in the optimization parameters as the approximation parameters of the RBF interpolation model, and perform multi-objective parameter optimization on the approximation parameters and the number of the control points to obtain the interpolation surface output by the RBF interpolation model.

[0049] Preferably, the approximation parameter of the RBF interpolation model has a linear relationship with the proportion of the interference point clouds filtered out in the preprocessing stage. Among them, the larger the proportion, the more jittery the surface of the RBF interpolation model, the smaller the step size in the optimization parameters, and the smaller the proportion, the larger the smoothing rate in the optimization parameters.

[0050] In this embodiment, after the preprocessing of the point cloud data of the operation object and the screening of the control points are completed, the crucial surface modeling stage is entered. The core of this stage is to construct an RBF interpolation model using the control points, and adjust the approximation parameters and the number of the control points of the model through multi-objective parameter optimization, so as to obtain the optimal interpolation surface.

[0051] First, select the most suitable model for the current operation object from multiple basis function models. Input the control points obtained in the preprocessing stage into the selected basis function model to construct an RBF surface interpolation equation. RBF is a powerful function approximation tool, which can generate a smooth surface model according to the distribution of the input points. By solving the weight coefficients of the equation, a specific RBF interpolation model can be obtained. This process can not only accurately reflect the geometric characteristics of the operation object, but also provide an important basis for subsequent operation planning.

[0052] Secondly, in order to ensure the accuracy and stability of the RBF interpolation model, a series of evaluation indexes need to be set. These indexes include running time, residual standard deviation, mean square error, coefficient of determination, and average smoothness. The running time reflects the calculation efficiency of the model; the residual standard deviation and the mean square error measure the accuracy of the model interpolation; the coefficient of determination is used to evaluate the data interpretation ability of the model; the average smoothness ensures that the generated surface has good smoothness. Through these evaluation indexes, the performance of the RBF interpolation model can be comprehensively evaluated, thus providing a clear direction for the optimization process.

[0053] Next, obtain the optimization parameters, and use the step size and smoothing rate in the optimization parameters as the proximity parameters of the RBF interpolation model. The setting of the proximity parameters is crucial for the interpolation effect of the model. In the present invention, the proximity parameters have a linear relationship with the proportion of the interference point cloud filtered out in the preprocessing stage. Specifically, when the proportion of the interference point cloud filtered out is larger, the surface of the RBF interpolation model is more likely to exhibit jitter. This is because filtering out too many interference point clouds may cause the model to be overly sensitive to local changes in the data. At this time, it is necessary to reduce the step size in the optimization parameters to improve the stability of the model. On the contrary, when the proportion of the interference point cloud filtered out is small, the smoothing rate in the optimization parameters can be increased accordingly. A larger smoothing rate helps to smooth the surface of the model, reduce unnecessary fluctuations, and thus obtain a smoother and more stable interpolation surface.

[0054] Finally, based on the above evaluation indicators and the setting of the proximity parameters, perform multi-objective parameter optimization on the proximity parameters and the number of control points. This optimization process aims to balance the accuracy, stability, and computational efficiency of the model. By adjusting the number of control points and the proximity parameters, it is possible to improve the stability and computational efficiency of the model while ensuring the accuracy of the model. Finally, obtain the interpolation surface output by the RBF interpolation model. This interpolation surface can not only accurately reflect the geometric shape of the operation object but also provide an important reference basis for the automated operation of construction machinery.

[0055] S4. Perform operation planning on the interpolation surface to obtain a planned excavation path with optimized indicators, and end this excavation operation.

[0056] Specifically, step S4 includes: using the planning terminal to perform operation planning on the output interpolation surface to obtain a planned excavation path with optimized indicators, where the optimized indicators include the full bucket rate and the excavation force.

[0057] Obtain the actual excavation force as feedback information, and correct the prior information in the operation planning according to the feedback information.

[0058] Preferably, use the proportion of the interference point cloud filtered out in the preprocessing stage as the prior information about the internal distribution of the operation object, and calculate the full bucket rate and the excavation force according to the prior information.

[0059] In this embodiment, it enters the operation planning stage. The core of this stage is to use the planning terminal to perform operation planning on the output interpolated surface to generate an optimized excavation path. Specifically, the planning terminal will formulate an optimal excavation path according to the geometric characteristics of the interpolated surface and in combination with the actual situation of the operation object. The optimization indexes of this path mainly include the full bucket rate and the excavation force. The full bucket rate refers to the loading efficiency of the bucket during the excavation process, reflecting the economy and efficiency of the operation; while the excavation force refers to the force required during the excavation process, which directly affects the safety and stability of the operation. By optimizing these two indexes, it can ensure that the construction machinery achieves the best performance during the operation process.

[0060] When performing operation planning, first, the proportion of the interference point cloud filtered out in the preprocessing stage is obtained and used as the prior information about the internal distribution of the operation object. This prior information provides an important reference basis for the planning, enabling the planning process to more accurately reflect the actual situation of the operation object. Based on this prior information, more accurate full bucket rate and excavation force can be calculated, thus providing a scientific basis for the subsequent operation planning. This data-driven optimization method not only improves the accuracy of the planning but also enhances the flexibility and adaptability of the operation.

[0061] After completing the operation planning, the construction machinery will perform actual operations according to the generated planned excavation path. During the operation process, the actual excavation force is monitored in real time and used as feedback information. This feedback mechanism can effectively compare the actual operation situation with the previous plan, so as to timely discover the deficiencies in the plan. According to the feedback information of the actual excavation force, the prior information in the operation planning is corrected. This ability of dynamic adjustment enables the construction machinery to maintain an efficient operation state in a complex and changeable operation environment, ensuring the safety and economy of the operation.

[0062] To sum up, the surface interpolation method for the automatic operation object of construction machinery aims to solve the problems in the prior art such as the disconnection between perception and planning during the automatic operation process of construction machinery, the sparse surface features of the operation object, and the frequent changes in working conditions. Through a series of innovative steps, this method realizes the efficient connection from environmental perception to operation planning, significantly improving the automation operation accuracy and efficiency of construction machinery under complex working conditions.

[0063] Specifically, the surface interpolation method for the automatic operation object of construction machinery not only solves the problems in the prior art such as the disconnection between perception and planning, sparse surface features of the operation object, and frequent changes in working conditions during the automatic operation process of construction machinery, but also significantly improves the accuracy and efficiency of surface interpolation through refined preprocessing, optimized control point selection strategies, dynamic parameter optimization, and closed-loop operation planning, providing strong support for the autonomous decision-making and operation of construction machinery. This method provides an efficient and accurate technical solution for the automatic operation of construction machinery, can effectively cope with complex and changeable operation environments, and significantly improves operation efficiency and resource utilization rate.

[0064] Please refer to Figure 3 , the second embodiment of the present invention provides a surface interpolation device for the automatic operation object of construction machinery, which includes:

[0065] A data segmentation unit 201, configured to obtain the environmental information of the operation object collected by a preset sensor assembly, distinguish the environmental information in detail according to a semantic segmentation model, and obtain the point cloud information of the operation object therefrom;

[0066] A preprocessing and screening unit 202, configured to preprocess the point cloud information of the operation object to generate points to be interpolated, and screen the points to be interpolated according to a quadratic surface formula to obtain control points;

[0067] A surface output unit 203, configured to construct an RBF interpolation model based on the control points, and perform multi-objective parameter optimization on the approximation parameter of the RBF interpolation model and the number of control points according to preset optimization parameters to obtain an interpolated surface output by the RBF interpolation model;

[0068] An operation planning unit 204, configured to perform operation planning on the interpolated surface to obtain a planned excavation path with optimized indicators, and end the current excavation operation.

[0069] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art of the present technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A surface interpolation method for an automatic operation object of an engineering machinery, characterized in that: include: Obtain the environmental information of the work object collected by the preset sensor components, distinguish the environmental information in detail according to the semantic segmentation model, and obtain the point cloud information of the work object from it; Preprocessing the point cloud information of the work object to generate points to be interpolated, and screening the points to be interpolated according to the quadratic surface formula to obtain control points; Constructing an RBF interpolation model based on the control points, and performing multi-objective parameter optimization on the approach parameters and the number of control points of the RBF interpolation model according to preset optimization parameters to obtain an interpolation surface output by the RBF interpolation model; The interpolation surface is planned to obtain a planned excavation path with optimized indicators, and the excavation operation is terminated.

2. The surface interpolation method for an automatic operation object of an engineering machinery according to claim 1, characterized in that: Obtain the environmental information of the work object collected by the preset sensor component, distinguish the environmental information in detail according to the semantic segmentation model, and obtain the point cloud information of the work object from it, specifically: Acquire environmental information of the working object collected by a preset sensor component, wherein the sensor component includes a laser radar and a depth camera; The environmental information is semantically segmented based on the PMF semantic segmentation model to generate work object point cloud information, wherein the work object point cloud information includes soil pile point cloud and soil slope point cloud.

3. The surface interpolation method for an automatic operation object of an engineering machinery according to claim 1, characterized in that: Preprocess the point cloud information of the operation object to generate points to be interpolated, specifically: According to the point cloud information of the operation object, a KD-Tree data structure is constructed, and the KD-Tree data structure is calculated to obtain a local reachable distance and a local reachable density; Calculating according to the local reachable distance and the local reachable density to generate a local outlier factor LOF, and judging the local outlier factor LOF; Filter out the local outlier factor LOF that is greater than a preset value, calculate the normal vector of the remaining local outlier factor LOF after filtering out based on the KD-Tree data structure and the PCA principal component analysis method, and standardize the coordinates of the point cloud corresponding to the remaining local outlier factor LOF, and calculate the covariance matrix of the standardized data matrix, wherein the preset value is 1; The covariance matrix is ​​solved to obtain eigenvalues ​​and eigenvectors, the change rate of the normal vector is calculated according to the eigenvalues ​​and eigenvectors, and the point cloud with a change rate higher than a preset change value is screened out to obtain the points to be interpolated.

4. The surface interpolation method for an automatic operation object of an engineering machinery according to claim 1, characterized in that: The points to be interpolated are screened according to the quadratic surface formula to obtain control points, which are specifically: Substituting the coordinates of each point in the to-be-interpolated point into a preset quadratic surface formula, constructing a design matrix and a target vector, and solving the interpolation coefficients of the design matrix and the target vector according to the least squares method to obtain a quadratic surface equation; Derivative the quadratic surface equation to obtain Hessian matrix elements, and calculate the Hessian matrix elements to generate principal curvatures k1 and k2; Calculating an average curvature according to the principal curvature k1 and the principal curvature k2, and distinguishing the to-be-interpolated points into edge points and internal points based on the average curvature; The internal points are screened through a grid, the internal points with the highest average curvature are selected, and the screened internal points and the edge points are used as control points.

5. The surface interpolation method for an automatic operation object of an engineering machinery according to claim 4, characterized in that: The edge points are selected from the points to be interpolated at a preset equal spacing, wherein one nth of the length of each side of the interpolation surface is used as the equal spacing of each side, and n is a parameter determined according to the size of the interpolation object.

6. The surface interpolation method for an automatic operation object of an engineering machinery according to claim 1, characterized in that: An RBF interpolation model is constructed based on the control points, and multi-objective parameter optimization is performed on the approach parameters and the number of control points of the RBF interpolation model according to preset optimization parameters to obtain an interpolation surface output by the RBF interpolation model, specifically: Select a basis function model, input the control points into the basis function model, construct an RBF surface interpolation equation, and solve the weight coefficient of the equation to obtain the RBF interpolation model; The running time, residual standard deviation, mean square error, coefficient of determination, and average smoothness are set according to the preset parameters and used as evaluation indicators; The optimization parameters are obtained, the step size and smoothing rate in the optimization parameters are used as the approximation parameters of the RBF interpolation model, and multi-objective parameter optimization processing is performed on the approximation parameters and the number of control points to obtain the interpolation surface output by the RBF interpolation model.

7. The surface interpolation method for an automatic operation object of an engineering machinery according to claim 6, characterized in that: The proximity parameter of the RBF interpolation model is linearly related to the proportion of the interference point cloud filtered out in the preprocessing stage, wherein the larger the proportion, the more jittery the surface of the RBF interpolation model, the smaller the step size in the optimization parameter, the smaller the proportion, and the greater the smoothing rate in the optimization parameter.

8. The surface interpolation method for an automatic operation object of an engineering machinery according to claim 1, characterized in that: The interpolation surface is operated and planned to obtain a planned mining path with optimized indicators, and the mining operation is terminated. Specifically, the following steps are performed: Using the planning end to perform operation planning on the output interpolation surface, and obtain a planned excavation path with optimized indicators, wherein the optimized indicators include full bucket rate and excavation force; The actual excavation force is obtained as feedback information, and the prior information in the operation plan is corrected according to the feedback information.

9. The surface interpolation method for an automatic operation object of an engineering machinery according to claim 8, characterized in that: The proportion of the interference point cloud filtered out in the preprocessing stage is used as the prior information of the internal distribution of the work object, and the full bucket rate and the digging force are calculated based on the prior information.

10. A surface interpolation device for an automatic operation object of an engineering machinery, characterized in that: include: A data segmentation unit is used to obtain the environmental information of the working object collected by the preset sensor component, distinguish the environmental information in detail according to the semantic segmentation model, and obtain the point cloud information of the working object therefrom; A preprocessing and screening unit, used for preprocessing the point cloud information of the work object to generate points to be interpolated, and screening the points to be interpolated according to a quadratic surface formula to obtain control points; A surface output unit, used to construct an RBF interpolation model based on the control points, and perform multi-objective parameter optimization on the approach parameters and the number of control points of the RBF interpolation model according to preset optimization parameters to obtain an interpolation surface output by the RBF interpolation model; The operation planning unit is used to perform operation planning on the interpolation surface, obtain a planned excavation path with optimized indicators, and end the excavation operation.

Images