Precision-improved heavy-load robot track and configuration partition collaborative planning method and system

Abstract

The invention relates to a precision-improved heavy-load robot track and configuration partition collaborative planning method and system, and relates to the technical field of robot control, and the method comprises the steps: carrying out the constraint of a machining point of a machining region, building a high-precision operation domain, further combining the clustering fitting, obtaining the optimal machining region segmentation, and carrying out the clustering fitting on the machining region segmentation, a plurality of processing relations are analyzed, conformal mapping and a model are utilized to carry out efficient error simulation, and track planning with uniform error distribution and continuous parameters is realized in combination with adaptive track planning. Simulated machining is carried out through trajectory planning, and robot configuration and workpiece layout are combined and modeled into a nested combination for collaborative optimization of configuration and layout. And finally, integral precision iterative optimization is adopted, region segmentation, trajectory planning and configuration optimization are brought into a unified framework for optimization, and a collaborative planning optimization result with improved machining precision is obtained, so that the integral precision and stability of the heavy-load robot in complex component machining are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot control, and in particular to a heavy-load robot trajectory and pose partition cooperative planning method and system with improved precision. BACKGROUND

[0002] At present, large and complex parts are widely used in the core components of high-end equipment such as aerospace, shipbuilding, and new energy, such as water turbine impellers, large wind power equipment blades, and oil tank shells. Complex parts usually have large size, narrow machining space, complex internal cavity, and various geometric features, making the machining extremely difficult. At the same time, such components often work in extreme conditions such as high temperature, high pressure, or high speed, which puts extremely strict requirements on the overall machining precision and performance.

[0003] Heavy-load industrial robots are gradually introduced into the manufacturing field of complex parts due to their high flexibility, large workspace, and convenient configuration. They can be used to realize complex machining and can be combined with devices such as variable positioners to achieve all-around machining of parts. At the same time, they are also easy to improve efficiency through multi-machine cooperative partition operation, and have obvious advantages compared to numerical control machine tools used in traditional technical solutions. However, the structure and characteristics of robots determine that they face significant challenges in high-precision machining.

[0004] Specifically, the error sources in the robot machining process are complex, including positioning errors, force deformation errors, and other geometric errors, as well as non-geometric errors such as wheel clearance and assembly deviation. Furthermore, during the machining of complex parts, the robot configuration and end posture change dramatically, making it difficult to predict and suppress machining errors, which is a major obstacle to the promotion of robots to high-precision process links. Among them, trajectory planning is the core problem in the field of numerical control machining and is also a key factor in determining machining efficiency and precision. Robot trajectory planning not only needs to meet the process requirements of the task space (such as error suppression, tool accessibility, and process parameter optimization), but also must consider the constraints of the configuration space (such as accessibility, stiffness, operability, and fairness). However, traditional trajectory planning methods have problems such as uneven trajectory, overlapping trajectory, and posture mutation in the machining of large and complex parts, making it difficult to control machining errors.

[0005] In the prior art, there are corresponding solutions for trajectory planning based on curvature optimization and pose optimization based on stiffness. However, most existing technologies are limited to single optimization of trajectory or pose, and lack cooperative planning mechanisms for trajectory and pose, so they cannot solve the problem of insufficient robot machining precision as a whole. SUMMARY

[0006] The application provides a heavy-load robot trajectory and configuration partitioned cooperative planning method and system with improved precision, which segments a processing area through high-precision operation domain segmentation, performs adaptive parameter continuity trajectory planning in each processing area through error simulation, realizes configuration optimization through nested combination and cooperative optimization of configuration and workpieces, finally considers overall cooperation, integrates area segmentation, trajectory planning and configuration optimization into a unified framework, realizes overall precision iterative optimization, and solves the problem of insufficient overall precision caused by the fragmentation of trajectory planning and configuration optimization and uneven processing error distribution in the prior art.

[0007] In a first aspect, the application provides a heavy-load robot trajectory and configuration partitioned cooperative planning method with improved precision, comprising:

[0008] For each processing point on the processing member, an operation domain of target precision is constructed based on multi-task constraints when the robot processes the processing member, the multi-task constraints include performance constraints and process constraints, and the operation domain is an attainable domain represented by high-dimensional data;

[0009] Based on each operation domain, a neighborhood graph is established, and an initial segmentation result of optimal processing area segmentation is obtained through clustering fitting processing;

[0010] Based on the initial segmentation result, error simulation is performed using conformal mapping combined with a high-dimensional model based on the analyzed processing relationship of the robot, and an error simulation result is obtained;

[0011] Adaptive trajectory planning is performed according to the error simulation result, and a parameter continuity trajectory planning result is obtained;

[0012] When simulating processing based on the trajectory planning result, the robot configuration change sequence is optimized and modeled as shortest path analysis according to the neighborhood graph, and cooperative optimization is performed through nested combination of configuration and workpiece layout, and a configuration optimization result is obtained;

[0013] Overall precision iterative optimization is adopted, and overall iterative optimization is performed based on the initial segmentation result, the trajectory planning result and the configuration optimization result, and a cooperative planning optimization result with improved processing precision is obtained.

[0014] Optionally, for each processing point on the processing member, an operation domain of target precision is constructed based on multi-task constraints when the robot processes the processing member, comprising:

[0015] A virtual model of the processing member is established, and each processing point of the virtual model is identified;

[0016] For each processing point, the performance constraints and process constraints when the robot processes the processing member are analyzed;

[0017] Converting the tool posture constraint into a robot pose constraint through inverse kinematics based on the performance constraint and the process constraint, and constructing an operation domain of target accuracy in a pose space of the robot;

[0018] The tool posture constraint is a posture of a machining tool under the performance constraint and the process constraint when the robot machines the machining member.

[0019] Optionally, based on the operation domains, a neighborhood graph is established, and an initial segmentation result of optimal machining region segmentation is obtained through clustering fitting processing, including:

[0020] According to the operation domains between the machining points, similarity is analyzed through probability distribution;

[0021] Taking each machining point as a vertex, a neighborhood graph is constructed based on the similarity;

[0022] Based on the neighborhood graph, a graph weight matrix is obtained by using graph relationship encoding of an attention mechanism, and a similarity matrix is constructed;

[0023] Based on the similarity matrix, a clustering algorithm is used for machining point clustering, and a projection boundary extraction algorithm is used for fitting of a partition contour, to obtain an initial segmentation result of optimal region segmentation.

[0024] Optionally, according to the operation domains between the machining points, similarity is analyzed through probability distribution, including:

[0025] Key features are extracted from the operation domains of the machining points through a t-SNE dimension reduction algorithm, and the key features are represented in the form of probability distribution;

[0026] An optimal transport distance theory is used to analyze the similarity between the machining points based on the key features.

[0027] Optionally, based on the initial segmentation result, an error simulation result is obtained by using conformal mapping combined with a high-dimensional model for error simulation based on the analyzed machining relationship of the robot, including:

[0028] Based on the initial segmentation result, an error distribution simulation model is established;

[0029] A first relationship between a feed direction of the tool and a stiffness direction of the robot during machining is analyzed, and an influence relationship of a geometric form of a machining trajectory on a material removal rate and a cutting force is analyzed;

[0030] Based on the first relationship and the influence relationship, a machining relationship is determined;

[0031] In the error distribution simulation model, based on the machining relationship, dimension reduction processing is performed by using conformal mapping, and error distribution simulation of a complex curved surface is performed in combination with a high-dimensional model to obtain an error simulation result.

[0032] Optionally, adaptive trajectory planning is performed according to the error simulation result to obtain a trajectory planning result with parameter continuity, including:

[0033] Based on the error simulation result, adaptive threshold segmentation and region contour extraction are adopted to identify a key machining area to obtain an error mutation area.

[0034] In different error mutation areas, error uniform distribution processing is performed based on a preset trajectory form to obtain trajectory planning information with error uniformity.

[0035] Based on the trajectory planning information, adaptive switching processing is performed based on a preset trajectory parameter switching rule to update the trajectory planning information to obtain a trajectory planning result with parameter continuity.

[0036] Optionally, collaborative optimization is performed through nested combination of the posture and the workpiece layout to establish a posture optimization result, including:

[0037] For the shortest path analysis, global optimization is performed by searching algorithm with error as the weight, and the optimal path of posture change sequence optimization is obtained through graph cropping and efficient search algorithm.

[0038] The workpiece layout posture of each machining component is one-to-one corresponding to the posture sequence of the robot, the workpiece layout posture is taken as an upper variable, and the workpiece layout posture optimization and the posture change sequence optimization are modeled as a nested combination model.

[0039] The nested combination model is analyzed by an improved whale algorithm to obtain a posture optimization result.

[0040] Optionally, overall precision iterative optimization is adopted, and overall iterative optimization is performed based on the initial segmentation result, the trajectory planning result and the posture optimization result to obtain a collaborative planning optimization result for improving machining precision, including:

[0041] Overall precision iterative optimization is adopted to establish a unified framework of the initial segmentation result, the trajectory planning result and the posture optimization result.

[0042] An expectation maximization algorithm is introduced, overall iterative optimization is performed based on the unified framework until convergence is achieved, and a collaborative planning optimization result for improving machining precision is obtained.

[0043] The expectation maximization algorithm performs overall iterative optimization through an estimation step and a maximization step.

[0044] Optionally, the expectation maximization algorithm is introduced for overall iterative optimization based on the unified framework, including:

[0045] Under the current region segmentation, trajectory planning and pose optimization are performed through the estimation step to obtain end trajectory and robot workpiece pose, and to calculate the prediction error distribution;

[0046] According to the prediction error distribution, the region is re-divided through the maximization step to update the initial segmentation result, and the trajectory planning result and the pose optimization result are updated based on the updated initial segmentation result.

[0047] In a second aspect, the application provides a heavy-load robot trajectory and pose partition collaborative planning system with improved accuracy, including:

[0048] A high-precision operation domain establishment module is configured to, for each machining point on a machining member, construct an operation domain of target accuracy based on multi-task constraints when a robot machines the machining member, the multi-task constraints including performance constraints and process constraints, and the operation domain being an attainable domain represented by high-dimensional data;

[0049] A region segmentation module is configured to, based on each operation domain, establish a neighborhood graph, and obtain an initial segmentation result of optimal machining region segmentation through clustering fitting processing;

[0050] An error simulation module is configured to, based on the initial segmentation result, utilize conformal mapping combined with a high-dimensional model to perform error simulation based on an analyzed machining relationship of the robot, and obtain an error simulation result;

[0051] A trajectory planning module is configured to perform adaptive trajectory planning according to the error simulation result, and obtain a trajectory planning result with parameter continuity;

[0052] A collaborative optimization module is configured to, when simulating machining based on the trajectory planning result, model a sequence of robot pose changes as shortest path analysis according to the neighborhood graph, and perform collaborative optimization through nested combination of pose and workpiece layout to obtain a pose optimization result;

[0053] An accuracy iterative optimization module is configured to perform overall accuracy iterative optimization based on the initial segmentation result, the trajectory planning result, and the pose optimization result to obtain a collaborative planning optimization result with improved machining accuracy.

[0054] In summary, the embodiment of the present application performs multi-job constraints on the machining point, establishes a high-precision operation domain, and constructs a neighborhood graph based on the operation domain, combined with clustering fitting, to obtain the optimal segmentation of the machining area. Based on the segmentation result of the area, a model is established to analyze various machining relationships, and conformal mapping and high-dimensional models are used for efficient error simulation, combined with adaptive trajectory planning to achieve trajectory planning with parameter continuity that evenly distributes errors. Through simulated machining of the planned trajectory, the robot pose and workpiece layout are combined to model a nested combination for collaborative optimization of the pose and layout. Finally, the overall precision is iteratively optimized, and the area segmentation, trajectory planning, and pose optimization are integrated into a unified framework for overall iterative optimization until convergence, resulting in collaborative planning and optimization results that improve machining precision, significantly improving the overall precision and stability of heavy-duty robots in complex component machining, and solving the problems of fragmented trajectory planning and pose optimization, uneven error distribution, and insufficient overall precision in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0055] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and serve to explain the principles of the present application together with the specification.

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, those skilled in the art can obtain other drawings from these drawings without any creative effort.

[0057] Figure 1 is a flowchart of a precision-improved heavy-duty robot trajectory and pose partition collaborative planning method provided by an embodiment of the present application;

[0058] Figure 2 is a flowchart of a precision-improved heavy-duty robot trajectory and pose partition collaborative planning method provided by an embodiment of the present application;

[0059] Figure 3 is a machining area segmentation schematic diagram provided by an optional example of the present application;

[0060] Figure 4 is a robot partition pose optimization schematic diagram provided by an optional example of the present application;

[0061] Figure 5 is a robot trajectory and pose collaborative optimization schematic diagram provided by an optional example of the present application;

[0062] Figure 6 is a block diagram of a precision-improved heavy-duty robot trajectory and pose partition collaborative planning system provided by an embodiment of the present application. DETAILED DESCRIPTION

[0063] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work belong to the scope of protection of the present application.

[0064] In order to facilitate the understanding of the embodiments of the present application, the following will be further explained and described with reference to the drawings and specific embodiments, and the embodiments do not constitute limitations on the embodiments of the present application.

[0065] Figure 1 A flowchart of a precision-improved heavy-load robot trajectory and configuration partition collaborative planning method provided by the embodiments of the present application is shown in FIG. 1. As shown in FIG. 1, the precision-improved heavy-load robot trajectory and configuration partition collaborative planning method provided by the embodiments of the present application can specifically include the following steps: Figure 1

[0066] In step 110, for each machining point on a machining member, an operation domain of target precision is constructed based on multi-task constraints when a robot machines the machining member.

[0067] The multi-task constraints include performance constraints and process constraints, and the operation domain is an attainable domain represented by high-dimensional data.

[0068] In the present embodiment, for a complex machining member, each machining point of the machining member is analyzed. For each machining point, the performance constraints and the process constraints when a heavy-load robot (hereinafter referred to as a robot) machines a machining member by using a tool such as a milling cutter are comprehensively considered, and the multi-task constraints are analyzed. Then, the multi-task constraints are used to construct a high-precision operation domain (abbreviated as high-precision operation domain, high-precision feasible domain or high-precision attainable domain) in a robot-workpiece pose space as an operation domain of target precision.

[0069] The robot-workpiece pose space mainly includes but is not limited to robot joint coordinates and workpiece layout poses, and the highest dimension is 12; the high-precision operation domain refers to a high-precision attainable domain, which is mainly represented by high-dimensional data, and the highest dimension is also 12.

[0070] In step 120, a neighborhood graph is established based on each operation domain, and an initial segmentation result of optimal machining area segmentation is obtained through clustering fitting processing.

[0071] ​In this embodiment, the neighborhood graph mainly includes but is not limited to: machining points, edges connecting machining points, and edge weights of each edge, etc. The neighborhood graph can be understood as a weighted undirected graph. The initial segmentation result refers to the initial machining region segmentation result, mainly including each machining region in the optimal segmentation in the initial segmentation.

[0072] In the related art, the prior art still has deficiencies in the segmentation of machining regions in robot machining of complex parts, specifically manifested as that the machining region division still depends on artificial experience, and lacks a reasonable partition optimization method.

[0073] To solve the defects of the prior art in machining region segmentation, this embodiment introduces a high-precision operation domain, realizes optimal segmentation of machining regions based on operation domain constraints, and provides a planning basis for subsequent trajectory planning.

[0074] In specific implementation, for each high-precision operation domain, a neighborhood graph is constructed with machining points as vertices, paths between machining points as edges, and edge weights. Then, the machining points in the neighborhood graph are clustered by a clustering algorithm, and different partition contours are fitted, and finally the initial machining region segmentation result of the machining component is obtained.

[0075] Step 130, on the basis of the initial segmentation result, based on the analyzed machining relationship of the robot, error simulation is performed using conformal mapping combined with a high-dimensional model to obtain an error simulation result.

[0076] Step 140, adaptive trajectory planning is performed according to the error simulation result to obtain a parameter continuity trajectory planning result.

[0077] The steps 130-140 are uniformly described as follows:

[0078] In specific implementation, on the basis of machining region segmentation, the machining relationship of the robot when machining the machining component is analyzed, including trajectory form, feed direction, and error distribution relationship, etc. Conformal mapping combined with a high-dimensional model (such as Dexel model) is introduced to analyze complex surfaces and perform error distribution simulation of complex surfaces to obtain an error simulation result.

[0079] Then, based on the error simulation result, adaptive trajectory planning is performed in a manner such as adaptive switching (such as adaptive threshold segmentation) to ensure parameter continuity between different trajectory forms, and a trajectory planning result is obtained.

[0080] Thus, this embodiment realizes error-uniform-distribution-oriented adaptive trajectory planning, so that the trajectory planning can evenly distribute errors, avoid local error concentration, and improve overall accuracy.

[0081] Step 150, when simulating machining based on the trajectory planning result, the robot pose sequence is optimized and modeled as a shortest path analysis according to the neighborhood graph, and is cooperatively optimized through the nested combination of the pose and the workpiece layout to obtain a pose optimization result.

[0082] In the related art, the existing technology usually only performs single optimization of a trajectory or a pose when machining a complex part by a robot, and the pose optimization and the trajectory planning are separated, which easily leads to mismatching between the trajectory and the pose and poor machining stability.

[0083] To this end, the embodiment introduces cooperative optimization of the robot pose and the workpiece layout based on the precision traction when simulating machining based on the trajectory planning result. Based on the established neighborhood graph, the robot pose optimization is modeled as a shortest path analysis, i.e., a shortest path problem of the neighborhood graph is solved. The robot pose sequence is globally optimized.

[0084] In a specific implementation, the embodiment cooperatively optimizes the robot pose and the workpiece layout, combines the pose sequence optimization and the workpiece layout sequence optimization, and uses an improved intelligent algorithm to cooperatively optimize the workpiece layout and the robot pose to obtain a pose optimization result.

[0085] Step 160, the overall precision iterative optimization is used to perform overall iterative optimization based on the initial segmentation result, the trajectory planning result, and the pose optimization result to obtain a cooperative planning optimization result with improved machining precision.

[0086] In the related art, the existing technology lacks a global iterative optimization mechanism for the region, the trajectory, and the pose when machining a complex part by a robot, and cannot realize overall precision improvement, which leads to overall precision decline.

[0087] To realize overall precision improvement, the embodiment uses an overall precision iterative optimization strategy, introduces an expectation maximization (EM) algorithm, and integrates the region segmentation, the trajectory planning, and the pose optimization into a unified framework. Through overall iterative optimization, an error distribution is predicted, and the partition (i.e., the region segmentation) is adjusted. Based on the adjusted partition, the trajectory planning and the pose adjustment are performed again until the error distribution converges or the maximum number of iterations is reached, the overall iterative optimization process is completed, and finally a cooperative planning optimization result with improved machining precision is obtained. The cooperative planning optimization result mainly includes a finally planned trajectory, a pose, and a region division, i.e., cooperative matching of the trajectory, the pose, and the region division is finally realized.

[0088] It can be seen that the embodiment establishes a high-precision operation domain by constraining the machining points of the machining area, and further combines clustering fitting to obtain optimal machining area segmentation. On the machining area segmentation, various machining relationships are analyzed, and conformal mapping and models are used for efficient error simulation. Combined with adaptive trajectory planning, error uniform distribution and parameter continuity trajectory planning are realized. Through trajectory planning simulation, the robot pose and workpiece layout are combined to model the nested combination for collaborative optimization of pose and layout. Finally, the overall precision iterative optimization is adopted to integrate the area segmentation, trajectory planning and pose optimization into a unified framework for optimization, and the collaborative planning optimization result of machining precision improvement is obtained, thereby significantly improving the overall precision and stability of the heavy-duty robot in complex component machining, and solving the problems of trajectory planning and pose optimization fragmentation, uneven machining error distribution and insufficient overall precision in the prior art.

[0089] Referring to Figure 2 , a step flowchart of a precision-improved heavy-duty robot trajectory and pose partition collaborative planning method provided by an optional embodiment of the application is shown. The method can specifically include the following steps:

[0090] In step 210, for each machining point on the machining component, an operation domain of target precision is constructed based on the multi-operation constraints when the robot machines the machining component.

[0091] The multi-operation constraints include performance constraints and process constraints, and the operation domain is an achievable domain represented by high-dimensional data.

[0092] In an optional embodiment, the above-mentioned operation domain of target precision constructed for each machining point on the machining component based on the multi-operation constraints when the robot machines the machining component can specifically include: establishing a virtual model of the machining component and identifying each machining point of the virtual model; for each machining point, analyzing the performance constraints and process constraints when the robot machines the machining component; based on the performance constraints and the process constraints, converting the tool posture constraint into the robot pose constraint through inverse kinematics to construct the operation domain of target precision in the pose space of the robot; wherein the tool posture constraint is the posture of the machining tool under the performance constraints and the process constraints when the robot machines the machining component.

[0093] The pose space of the robot refers to the robot-workpiece pose space.

[0094] In a specific implementation, when the robot processes the integrally formed component, the robot performance constraints, including but not limited to positioning error, stability, operability, joint stiffness, etc., and the process constraints, including but not limited to non-interference, tool inclination angle range, cutting posture requirements, etc., can be met simultaneously. The embodiment comprehensively considers the above-mentioned multiple task constraints, analyzes the tool posture constraints, converts the tool posture constraints into robot pose constraints through inverse kinematics, and constructs a high-precision operation domain.

[0095] In step 220, based on each operation domain, a neighborhood graph is established, and an initial segmentation result of the optimal machining area segmentation is obtained through clustering fitting processing.

[0096] Optionally, the above-mentioned based on each operation domain, a neighborhood graph is established, and an initial segmentation result of the optimal machining area segmentation is obtained through clustering fitting processing, can include the following sub-steps:

[0097] In sub-step 2201, the similarity between the machining points is analyzed according to the operation domain between the machining points.

[0098] Since the operation domain is high-dimensional data, the embodiment can reduce the dimension of the operation domain to extract key features as key features. These features can be represented in the form of probability distribution, that is, the probability distribution corresponding to each machining point is obtained. Then, the probability distribution of two machining points is used to analyze the similarity between the machining points, that is, the similarity between the high-precision operation domains is analyzed.

[0099] In an optional embodiment, the embodiment can analyze the similarity between the machining points according to the operation domain between the machining points through probability distribution, which can specifically include: extracting key features from the operation domain of each machining point through a t-SNE dimension reduction algorithm, and representing the key features in the form of probability distribution; based on the key features, the similarity between the machining points is analyzed using the optimal transport distance theory.

[0100] In a specific implementation, a dimension reduction algorithm (such as a t-SNE dimension reduction algorithm) can be introduced to reduce the dimension of high-dimensional data and extract main features. The optimal transport distance theory (such as Wasserstein distance) is used to evaluate the similarity of two high-precision feasible regions. The similarity can also be understood as the similarity.

[0101] In actual implementation, the optimal transport distance refers to the minimum transport quality (probability density in the discrete case, called probability mass) of converting one probability distribution (assuming ) in two probability distributions into another probability distribution (assuming ). The Wasserstein distance is the most common optimal transport distance and is widely used to measure the distance between two probability distributions.

[0102] Exemplarily, the optimal transmission distance theory is used to analyze the similarity between the machining points based on the key features, which can specifically include: taking the key features as probability distributions; taking the probability distributions as inputs, using bulldozer distance measurement, and analyzing the distance between two probability distributions to obtain the similarity between the machining points; wherein, , the probability distributions of two machining points are represented as , .

[0103] Thus, the embodiment realizes obtaining the similarity of the high-precision operation domain between each two machining points on the component.

[0104] Sub-step 2202: Taking each of the machining points as a vertex, a neighborhood graph is constructed based on the similarity.

[0105] Sub-step 2203: Based on the neighborhood graph, a graph weight matrix is obtained using graph relationship encoding of the attention mechanism, and a similarity matrix is constructed.

[0106] Sub-step 2204: Based on the similarity matrix, a clustering algorithm is used for machining point clustering, and a projection boundary extraction algorithm is used for fitting of the partition contour, to obtain an initial segmentation result of the optimal region segmentation.

[0107] The sub-step 2202 to sub-step 2204 are uniformly described as follows:

[0108] Referring to Figure 3 , in a specific implementation, a weighted undirected graph G (i.e., a neighborhood graph) is constructed with the machining points as vertices and the similarity as edge weights. Based on the neighborhood graph, graph relationship encoding of the attention mechanism (such as Query-Key mechanism) is used to obtain a graph weight matrix of the neighborhood graph, i.e., to obtain the edge weight of the edge between two machining points. Through the neighborhood graph and the graph weight matrix, a similarity matrix such as a Laplacian matrix is constructed.

[0109] Then, based on the similarity matrix, a clustering algorithm (such as a spectral clustering algorithm) is used for clustering of the machining points, and then a projection boundary extraction algorithm (alpha-shape) is used for fitting of the partition contour, to finally obtain the optimal segmentation of the machining region of the component, i.e., the initial segmentation result. Exemplarily, referring to Figure 3 , which is an exemplary machining region segmentation of one machining component.

[0110] Step 230: Based on the initial segmentation result, an error distribution simulation model is established.

[0111] Step 240: analyzing the first relationship between the feed direction of the tool during machining and the stiffness direction of the robot, and analyzing the influence relationship of the geometric form of the machining trajectory on the material removal rate and the cutting force.

[0112] Step 250: Determine the processing relationship based on the first relationship and the influence relationship.

[0113] Step 260: In the error distribution simulation model, based on the processing relationship, dimensionality reduction is performed using conformal mapping, and error distribution simulation of complex surfaces is performed in combination with a high-dimensional model to obtain error simulation results.

[0114] A unified explanation of steps 230-260 is provided below:

[0115] To address the shortcomings of existing technologies, such as the inability to evenly distribute errors in trajectory planning and the resulting decrease in overall accuracy due to localized error concentration, this embodiment achieves machining trajectory planning with uniform error distribution. (Refer to...) Figure 3 As shown, in the specific implementation, the error simulation model is first modeled. Then, based on the established error simulation model, the machining relationship between the machining trajectory and the robot error is analyzed, such as the relationship between the trajectory form, feed direction and error distribution.

[0116] Specifically, since the machining trajectory directly affects the cutting force distribution, it affects the robot error. Therefore, in trajectory planning, this embodiment comprehensively considers the following machining relationships: ① the relationship between the feed direction and the robot stiffness direction, i.e., the first relationship; ② the influence of the trajectory geometry on the material removal rate and cutting force, i.e., the influence relationship.

[0117] Then, conformal mapping and a high-dimensional Dexel model are introduced to simulate the error distribution of complex surfaces, and the error simulation results are obtained. Conformal mapping is used for dimensionality reduction to improve the efficiency of geometric operations.

[0118] Step 270: Perform adaptive trajectory planning based on the error simulation results to obtain trajectory planning results with parameter continuity.

[0119] In one optional embodiment, this embodiment performs adaptive trajectory planning based on the error simulation results to obtain a trajectory planning result with parameter continuity. Specifically, it may include: based on the error simulation results, using adaptive threshold segmentation and region contour extraction, identifying key processing areas to obtain error mutation regions; within different error mutation regions, performing error uniform distribution processing based on a preset trajectory form to obtain error uniform trajectory planning information; and based on the trajectory planning information, performing adaptive switching processing based on preset trajectory parameter switching rules to update the trajectory planning information and obtain a trajectory planning result with parameter continuity.

[0120] In practical implementation, based on the error simulation results, adaptive threshold segmentation can be used to analyze error mutation regions, identify key processing regions, and extract key processing regions as error mutation regions through region contours. Adaptive threshold segmentation includes error threshold segmentation.

[0121] Then, for different error mutation regions, error is evenly distributed based on the trajectory form, which includes, but is not limited to: circumferential trajectory (ensuring full coverage); spiral trajectory (constant curvature, reducing local error accumulation); and cycloidal trajectory (constant load, ensuring uniform cutting force). The trajectory planning information with uniform error is processed using trajectory parameter switching rules to ensure continuity and matching between different trajectory forms, ultimately yielding the trajectory planning result.

[0122] Step 280: When performing simulated processing based on the trajectory planning results, the robot configuration change sequence is optimized and modeled as the shortest path analysis according to the neighborhood graph. Collaborative optimization is performed through the nested combination of configuration and workpiece layout to obtain configuration optimization results.

[0123] In an optional embodiment, this embodiment achieves collaborative optimization through a nested combination of configuration and workpiece layout to establish configuration optimization results. Specifically, this may include: for the shortest path analysis, performing global optimization using a search algorithm with error as the weight, and processing it through graph clipping and efficient search algorithms to obtain the optimal path for configuration change sequence optimization; mapping the workpiece layout pose of each processing component to the robot's configuration sequence one-to-one, using the workpiece layout pose as the upper-level variable, and modeling the workpiece layout pose optimization and configuration change sequence optimization as a nested combination model; and analyzing the nested combination model using an improved whale algorithm to obtain configuration optimization results.

[0124] In this specific implementation, the configuration optimization mainly includes the collaborative optimization of configuration change sequence optimization and workpiece layout pose optimization. Specifically, during robot trajectory execution, the configuration change sequence directly affects machining accuracy. This embodiment models the optimization problem as a weighted directed graph shortest path problem, where machining points remain vertices, while edge weights are updated and determined primarily by a machining error evaluation function for global optimization. The optimal path is solved using search algorithms such as LKH (Lin-Kernighan-Helsgaun) or EET (Expansive Edge Tree), and graph pruning and efficient search algorithms are introduced to improve efficiency, thereby minimizing errors during configuration changes while improving the efficiency of optimal path solving.

[0125] In workpiece layout and pose optimization, each workpiece layout pose corresponds to a robot configuration sequence. In this embodiment, the workpiece layout pose is used as a higher-level variable, and the workpiece layout optimization and configuration optimization are modeled as a nested combined optimization problem. A nested combined optimization model is established, and an improved intelligent algorithm (such as the whale algorithm) is used to solve it, determining the minimum error of global machining and obtaining the configuration optimization result.

[0126] In practical implementation, the minimum global processing error is taken as the objective function, and the minimum processing error can be calculated using the objective function formula.

[0127] For example, based on the whale algorithm, The minimum machining error is calculated as the objective function. For the robot's end-effector posture, For the workpiece orientation, Joint angle, For robot inverse kinematics, As a stiffness evaluation index, This is an indicator of smoothness.

[0128] Step 290: Overall accuracy iterative optimization is adopted. Based on the initial segmentation result, the trajectory planning result and the configuration optimization result, overall iterative optimization is performed to obtain a collaborative planning optimization result that improves processing accuracy.

[0129] Optionally, an overall accuracy iterative optimization is adopted. Based on the initial segmentation result, the trajectory planning result, and the configuration optimization result, an overall iterative optimization is performed to obtain a collaborative planning optimization result that improves machining accuracy. This may include the following sub-steps:

[0130] Sub-step 2901 involves using overall accuracy iterative optimization to establish a unified framework for the initial segmentation results, the trajectory planning results, and the configuration optimization results.

[0131] Sub-step 2902 introduces the expectation maximization algorithm and performs overall iterative optimization based on the unified framework until the iteration converges, obtaining the collaborative planning optimization result with improved processing accuracy.

[0132] The expected value maximization algorithm is optimized through an overall iterative process involving estimation and maximization steps.

[0133] A unified explanation is provided for sub-steps 2901 and 2902:

[0134] In this implementation, to achieve the coupling of processing area segmentation, trajectory planning, and configuration optimization, this embodiment introduces the Expectation-Maximization (EM) algorithm to achieve overall iterative optimization. Region segmentation, trajectory planning, and configuration optimization are incorporated into a unified framework. Through the E-step (Expectation) and M-step (Maximization) of the Expectation-Maximization algorithm, the region segmentation (i.e., updating the initial region segmentation result), trajectory planning (i.e., updating the trajectory planning result), and configuration scheme (i.e., updating the configuration optimization result) are updated. Through iterative optimization, the overall accuracy of robot processing is gradually improved, ultimately achieving collaborative matching of trajectory, configuration, and region segmentation, resulting in a collaborative planning optimization result. This collaborative planning optimization result includes the final improved overall accuracy of the region segmentation, trajectory planning, and configuration scheme.

[0135] For example, refer to Figure 4 As shown, robot partition pose optimization is implemented; refer to Figure 5 As shown, this enables collaborative optimization of robot trajectory and posture.

[0136] In an optional embodiment, this embodiment introduces the expectation-maximization algorithm to perform overall iterative optimization based on the unified framework. Specifically, it may include: under the current region segmentation, performing trajectory planning and configuration optimization through the estimation step to obtain the end-effector trajectory and robot workpiece pose, and calculating the prediction error distribution; according to the prediction error distribution, performing region re-division through the maximization step to update the initial segmentation result, and updating the trajectory planning result and the configuration optimization result based on the updated initial segmentation result.

[0137] In the initial iteration of the algorithm, the current region segmentation is determined directly based on the initial segmentation result; during the algorithm iteration, the current region segmentation is determined by the initial segmentation result optimized in the previous iteration.

[0138] In the specific implementation, the iterative optimization of the E step includes: under the current region segmentation, performing trajectory planning and configuration optimization to obtain the end effector trajectory and robot-workpiece pose, and calculating the predicted error distribution; the iterative optimization of the M step includes: based on the error distribution results, re-dividing the region and updating the trajectory and configuration scheme. The EM step is repeated throughout the entire optimization iteration process until the error distribution converges or the maximum number of iterations is reached.

[0139] In summary, this application's embodiments impose multiple operational constraints on processing points, establish a high-precision operating domain, and construct a neighborhood graph based on the operating domain. Combined with clustering fitting, the optimal segmentation of the processing area is obtained. Based on the region segmentation results, a model is established to analyze various processing relationships. Conformal mapping and high-dimensional models are used for efficient error simulation, and adaptive trajectory planning is combined to achieve trajectory planning with uniform error distribution and parameter continuity. Processing is simulated using the planned trajectory, and the robot configuration and workpiece layout are modeled as a nested combination for collaborative optimization of configuration and layout. Finally, overall accuracy iterative optimization is adopted, incorporating region segmentation, trajectory planning, and configuration optimization into a unified framework for overall iterative optimization until iterative convergence. Ultimately, a collaborative planning optimization result with improved processing accuracy is obtained, thereby significantly improving the overall accuracy and stability of heavy-duty robots in the processing of complex components.

[0140] As can be seen, to address the technical problem of existing technologies lacking a global iterative optimization mechanism for region, trajectory, and configuration, thus hindering the overall accuracy improvement, the embodiments of this application mainly achieve: optimal segmentation of the processing area based on operational domain constraints; adaptive trajectory planning oriented towards uniform error distribution; collaborative optimization of robot configuration and workpiece layout pose; and iterative optimization coordination of region, trajectory, and configuration. The technical effect of the embodiments of this application is to significantly improve the overall accuracy and stability of heavy-duty robots in the processing of complex components, providing high-precision and flexible processing solutions for industries such as aerospace and energy equipment.

[0141] Therefore, compared with the prior art, the advantages of this application are:

[0142] ① More reasonable region segmentation: The global optimal segmentation of the processing area is achieved through high-precision operation domain and spectral clustering, avoiding the limitations of manual segmentation;

[0143] ② More uniform error distribution: The trajectory form and parameters are adaptively switched to avoid local error concentration and improve contour consistency;

[0144] ③ More efficient collaborative optimization: The configuration and workpiece layout are linked for optimization, ensuring that the trajectory matches the configuration and reducing local extreme value traps;

[0145] ④ More reliable overall accuracy: Through iterative optimization mechanism, global coordination of region, trajectory and configuration is achieved, and the overall accuracy is significantly improved.

[0146] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should know that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps may be performed in other orders or simultaneously.

[0147] like Figure 6As shown in the figure, this application embodiment also provides a heavy-duty robot trajectory and configuration partitioning collaborative planning system 600 with improved accuracy, including:

[0148] The high-precision operation domain establishment module 610 is used to construct an operation domain with target precision for each processing point on the processing component, based on the multiple operation constraints when the robot processes the processing component. The multiple operation constraints include performance constraints and process constraints, and the operation domain is an reachable domain represented by high-dimensional data.

[0149] The region segmentation module 620 is used to establish a neighborhood graph based on each of the operation domains and obtain the initial segmentation result of the optimal processing region segmentation through cluster fitting processing.

[0150] The error simulation module 630 is used to perform error simulation based on the initial segmentation results and the analyzed robot's processing relationship, using conformal mapping combined with a high-dimensional model to obtain error simulation results.

[0151] The trajectory planning module 640 is used to perform adaptive trajectory planning based on the error simulation results to obtain trajectory planning results with parameter continuity.

[0152] The collaborative optimization module 650 is used to optimize and model the robot configuration change sequence as the shortest path analysis based on the neighborhood graph when performing simulated processing based on the trajectory planning results, and to perform collaborative optimization through the nested combination of configuration and workpiece layout to obtain configuration optimization results.

[0153] The precision iterative optimization module 660 is used to perform overall precision iterative optimization based on the initial segmentation result, the trajectory planning result and the configuration optimization result to obtain a collaborative planning optimization result that improves machining accuracy.

[0154] It should be noted that the precision-enhanced heavy-duty robot trajectory and configuration partitioning collaborative planning system provided in the embodiments of this application can execute the precision-enhanced heavy-duty robot trajectory and configuration partitioning collaborative planning method provided in any embodiment of this application, and has the corresponding functions and beneficial effects of the execution method.

[0155] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0156] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for collaborative planning of trajectory and configuration partitioning for heavy-duty robots with improved accuracy, characterized in that, include: For each processing point on the processing component, based on the multiple operation constraints when the robot processes the processing component, an operation domain with target accuracy is constructed. The multiple operation constraints include performance constraints and process constraints. The operation domain is an reachable domain represented by high-dimensional data. Based on each of the aforementioned operational domains, a neighborhood graph is established, and the initial segmentation result of the optimal processing region segmentation is obtained through cluster fitting. Based on the initial segmentation results, and based on the analyzed robot's processing relationships, error simulation is performed using conformal mapping combined with a high-dimensional model to obtain the error simulation results. Adaptive trajectory planning is performed based on the error simulation results to obtain trajectory planning results with parameter continuity; When simulating processing based on the trajectory planning results, the robot configuration change sequence is optimized and modeled as the shortest path analysis according to the neighborhood graph. Collaborative optimization is performed through the nested combination of configuration and workpiece layout to obtain configuration optimization results. An overall accuracy iterative optimization is adopted, based on the initial segmentation result, the trajectory planning result, and the configuration optimization result, to obtain a collaborative planning optimization result that improves processing accuracy.

2. The method according to claim 1, characterized in that, For each machining point on the machining component, based on the multiple operation constraints when the robot is machining the component, an operational domain for the target accuracy is constructed, including: Establish a virtual model of the processed components and identify each processing point in the virtual model; For each of the aforementioned processing points, analyze the performance constraints and process constraints of the robot when processing the aforementioned components; Based on the performance constraints and the process constraints, the tool posture constraints are converted into robot configuration constraints through inverse kinematics, and the target accuracy operation domain is constructed in the robot's pose space. The tool posture constraint refers to the posture of the machining tool under the performance constraint and the process constraint when the robot is machining the machining component.

3. The method according to claim 1, characterized in that, Based on each of the aforementioned operational domains, a neighborhood graph is established, and the initial segmentation result for the optimal processing region segmentation is obtained through cluster fitting, including: Based on the operational domains between each processing point, similarity is analyzed using probability distribution. Using each of the aforementioned processing points as vertices, a neighborhood graph is constructed based on the aforementioned similarity; Based on the neighborhood graph, a graph weight matrix is ​​obtained by using graph relation encoding with an attention mechanism, and a similarity matrix is ​​constructed. Based on the similarity matrix, a clustering algorithm is used to cluster the processing points, and a projection boundary extraction algorithm is combined to fit the partition contours to obtain the initial segmentation result of the optimal region segmentation.

4. The method according to claim 3, characterized in that, Based on the operational domains between each processing point, similarity is analyzed through probability distribution, including: Key features are extracted from the operational domain of each of the processing points using the t-SNE dimensionality reduction algorithm, and the key features are represented in the form of a probability distribution. Using the optimal transmission distance theory, the similarity between processing points is analyzed based on the aforementioned key features.

5. The method according to claim 1, characterized in that, Based on the initial segmentation results, and based on the analyzed robot's processing relationships, error simulation is performed using conformal mapping combined with a high-dimensional model to obtain the error simulation results, including: Based on the initial segmentation results, an error distribution simulation model is established; The analysis examines the relationship between the tool feed direction and the robot stiffness direction during machining, as well as the influence of the geometry of the machining trajectory on the material removal rate and cutting force. Based on the first relationship and the influence relationship, the processing relationship is determined; In the error distribution simulation model, based on the processing relationship, dimensionality reduction is performed using conformal mapping, and error distribution simulation of complex surfaces is performed in combination with a high-dimensional model to obtain error simulation results.

6. The method according to claim 1, characterized in that, Based on the error simulation results, adaptive trajectory planning is performed to obtain trajectory planning results with parameter continuity, including: Based on the error simulation results, adaptive threshold segmentation and region contour extraction are used to identify key processing areas and obtain error mutation regions. Within different error mutation regions, error uniformity distribution processing is performed based on a preset trajectory form to obtain error uniform trajectory planning information. Based on the trajectory planning information, adaptive switching processing is performed according to preset trajectory parameter switching rules to update the trajectory planning information and obtain a trajectory planning result with parameter continuity.

7. The method according to claim 1, characterized in that, Collaborative optimization is achieved through nested combinations of configuration and workpiece layout to establish configuration optimization results, including: For the aforementioned shortest path analysis, a search algorithm is used to perform global optimization with error as the weight, and then graph clipping and efficient search algorithms are applied to obtain the optimal path for configuration change sequence optimization. The workpiece layout pose of each processing component is mapped one-to-one with the robot's configuration sequence. The workpiece layout pose is used as the upper-level variable, and the workpiece layout pose optimization and configuration change sequence optimization are modeled as a nested combination model. The nested combination model was analyzed using an improved whale algorithm to obtain configuration optimization results.

8. The method according to claim 1, characterized in that, An overall accuracy iterative optimization is adopted, based on the initial segmentation result, the trajectory planning result, and the configuration optimization result, to obtain a collaborative planning optimization result that improves machining accuracy, including: An iterative optimization approach is adopted to establish a unified framework for the initial segmentation results, the trajectory planning results, and the configuration optimization results; An expectation-maximization algorithm is introduced, and overall iterative optimization is performed based on the unified framework until the iteration converges, resulting in a collaborative planning optimization result that improves processing accuracy. The expected value maximization algorithm is optimized through an overall iterative process involving estimation and maximization steps.

9. The method according to claim 8, characterized in that, An expectation-maximization algorithm is introduced, and overall iterative optimization is performed based on the aforementioned unified framework, including: Under the current region segmentation, the estimated steps are used to perform trajectory planning and configuration optimization to obtain the end-effector trajectory and robot workpiece pose, and to calculate the prediction error distribution. Based on the predicted error distribution, the region is re-divided through a maximization step, the initial segmentation result is updated, and the trajectory planning result and the configuration optimization result are updated based on the updated initial segmentation result.

10. A collaborative planning system for trajectory and configuration partitioning of heavy-duty robots with improved accuracy, characterized in that, include: The high-precision operation domain establishment module is used to construct an operation domain with target precision for each processing point on the processing component, based on the multiple operation constraints when the robot processes the processing component. The multiple operation constraints include performance constraints and process constraints, and the operation domain is an reachable domain represented by high-dimensional data. The region segmentation module is used to establish a neighborhood graph based on each of the operation domains and obtain the initial segmentation result of the optimal processing region segmentation through cluster fitting. The error simulation module is used to perform error simulation based on the initial segmentation results and the analyzed robot's processing relationships, using conformal mapping combined with a high-dimensional model to obtain error simulation results. The trajectory planning module is used to perform adaptive trajectory planning based on the error simulation results, and obtain trajectory planning results with parameter continuity. The collaborative optimization module is used to optimize and model the robot configuration change sequence as the shortest path analysis based on the neighborhood graph when performing simulated processing based on the trajectory planning results. It performs collaborative optimization through the nested combination of configuration and workpiece layout to obtain configuration optimization results. The accuracy iterative optimization module is used to perform overall accuracy iterative optimization based on the initial segmentation result, the trajectory planning result, and the configuration optimization result to obtain a collaborative planning optimization result that improves processing accuracy.

Images