Self-adaptive deflashing method and system for molded part

Abstract

The invention provides a self-adaptive deflashing method and system for a molded part, and the method comprises the steps: collecting the point cloud data of the surface of a workpiece through an eye-in-hand point cloud camera, carrying out the splicing and downsampling processing through an ICP algorithm, and manually selecting the feature points of a parting surface to fit the parameters of a flash growth plane; performing coarse registration on the scanning point cloud and the CAD model by using an FPFH algorithm, extracting a potential flash region, calculating Euclidean distance and normal vector difference characteristics, and segmenting the flash point cloud through a decision function; constructing an energy function optimization displacement vector containing a data item, a regularization item and a tolerance constraint item, and updating a trajectory attitude by adopting K neighbor interpolation; projecting the flash point cloud to a growth plane to establish a grid system, generating an outer contour through iterative expansion to calculate the height of the flash, establishing a mapping relation between the height and the feeding speed, and smoothing the speed. According to the method, excessive deformation is inhibited through a TC-NICP registration algorithm of tolerance constraint; the feeding speed is dynamically adjusted based on gridding analysis; and closed-loop control of workpiece scanning, track generation and deflashing is realized.

Description

Technical Field

[0001] The present invention relates to the field of robot grinding processing equipment, in particular to an adaptive flash removal method and system for molded parts. Background Art

[0002] In molding processes such as casting and injection molding, flash is inevitably formed on the workpiece surface due to factors such as incomplete sealing of the mold parting surface and excessive injection pressure. The presence of flash not only affects the product appearance quality and assembly accuracy, but also causes stress concentration, significantly reducing the mechanical properties and service life of the parts. Therefore, deburring, as a key post-processing step in the production of molded parts, is of great significance to ensuring product quality. With the improvement of industrial automation level, robotic automated deburring technology has gradually become a research hotspot due to its advantages such as high flexibility and strong adaptability, but the existing technology still has technical bottlenecks that need to be solved urgently.

[0003] Currently, mainstream robotic deflashing methods primarily include offline programming based on CAD models, online adjustment based on vision guidance, and adaptive machining based on force control. However, these methods exhibit significant flaws in practical applications: First, offline trajectory planning methods based on CAD models struggle to address actual workpiece dimensional deformation. During the molding process, workpiece deformation caused by uneven cooling and shrinkage can cause deviations between the preset trajectory and the actual machining position, impacting deflashing accuracy. Second, existing trajectory optimization methods lack the ability to accurately perceive and dynamically adapt to the size distribution of flash. Flash height and morphology vary significantly at different locations, and the use of uniform machining parameters cannot guarantee effective removal and results in low machining efficiency. Third, current non-rigid registration methods are prone to over-adaptation when processing point cloud data. Sensor measurement errors and point cloud stitching errors can cause the registration results to exceed the actual tolerance range of the workpiece, thereby affecting the accuracy of subsequent trajectory generation. Summary of the Invention

[0004] In order to solve the technical problems in the existing technology that the offline trajectory planning method based on the CAD model cannot effectively deal with the trajectory deviation caused by the deformation of the workpiece, the trajectory tuning method lacks the accurate perception and adaptability of the flash size distribution, resulting in low processing efficiency, and the non-rigid registration method is prone to over-adaptation error exceeding the tolerance range when processing point cloud data, which affects the trajectory generation accuracy, the present invention provides an adaptive deburring method and system for molded parts.

[0005] The technical solutions provided by the present invention are as follows:

[0006] First aspect:

[0007] The present invention provides a method for adaptively deflashing a molded part, comprising:

[0008] S1. Data acquisition and preprocessing: An eye-on-hand point cloud camera is used to collect point cloud data of the workpiece surface. The ICP algorithm is used for multi-view point cloud stitching. The point cloud is processed by an octree-based voxel downsampling algorithm. The feature points of the parting surface are manually selected, and the RANSAC plane fitting algorithm is used to obtain the flash growth plane parameters.

[0009] S2. Flash point cloud segmentation: Use the FPFH algorithm to roughly align the scanned point cloud with the CAD model point cloud, extract the potential flash area based on the registration results, calculate the Euclidean distance feature and the normal vector difference feature, and segment the flash point cloud using the decision function and threshold;

[0010] S3, tolerance-constrained trajectory pose optimization: Construct an energy function containing data terms, regularization terms, and tolerance constraint terms, constrain the displacement vector, use K-nearest-neighbor inverse distance weighted interpolation to calculate the trajectory point deformation, and update the trajectory pose;

[0011] S4. Trajectory speed optimization based on grid analysis: The flash point cloud is projected onto the growth plane and a grid system is established. The contour points are sorted and iteratively expanded to generate an equidistant outer contour. The flash height is calculated based on the number of layers and layer height. The optimal feed speed is obtained through mapping and interpolation, and the speed is smoothed.

[0012] Second aspect:

[0013] The present invention provides an adaptive deflashing system for molded parts, comprising:

[0014] Data acquisition system, execution system, control system and software system. The data acquisition system includes a point cloud camera and a hand-eye calibration system. The point cloud camera adopts the eye-on-hand installation method to acquire multi-view point cloud data as the robot moves. It is connected to the industrial control computer via Ethernet to realize point cloud data transmission. The hand-eye calibration system includes a calibration board and calibration algorithm software for establishing the conversion relationship between the camera coordinate system and the robot base coordinate system;

[0015] The execution system includes a flexible grinding unit, a rotary file, and an industrial robot body. The flexible grinding unit has a floating compensation function and is connected to the standard interface of the robot end through a flange. The rotary file uses a cylindrical carbide tool with a diameter selected according to the characteristics of the workpiece and maintains a 90° processing angle with the workpiece surface. The industrial robot body is a 6-degree-of-freedom articulated industrial robot with a payload of no less than 10 kg.

[0016] The control system includes the industrial control computer, an electrical control unit, a robot controller and a communication interface module. The industrial control computer is used for data processing and method implementation, and calculation of the processing trajectory. The electrical control unit is integrated with a PLC controller to achieve coordinated control of various execution components, and contains a safety protection circuit, an emergency stop function and a safety protection mechanism. The robot controller provides real-time trajectory planning and interpolation functions, and has an Ethernet communication interface for data exchange with the industrial control computer. The communication interface module uses industrial Ethernet as the main communication method and supports TCP / IP and Modbus communication protocols.

[0017] The software system includes a point cloud processing module developed based on Open3D and a host computer communication module based on open_ABB.

[0018] The beneficial effects brought about by the technical solution provided by the present invention include at least:

[0019] (1) In the present invention, the tolerance-constrained TC-NICP registration algorithm is used to achieve accurate recognition of workpiece surface deformation and dynamic adjustment of trajectory posture. The algorithm introduces process tolerance constraints in the point cloud registration process, effectively suppressing the excessive deformation problem of traditional non-rigid registration methods, and reducing the average error of trajectory points from 1.00 mm in traditional methods to within 0.71 mm. By compensating for the deformation of the workpiece caused by cooling shrinkage in real time, the fit between the trajectory and the actual processing position is significantly improved, ensuring that the tool path accurately fits the complex curved surface, thereby improving the accuracy and consistency of the deburring operation, and is particularly suitable for the post-processing of high-precision molded parts;

[0020] (2) In the present invention, a trajectory speed tuning method based on grid analysis realizes dynamic optimization of processing parameters by establishing an adaptive mapping relationship between burr height and feed speed. This technology projects the burr point cloud onto the growth plane and constructs a grid system. It combines the kd tree algorithm and iterative expansion profile generation technology to accurately measure the burr height distribution in different areas. According to the pre-calibrated material grinding parameters, it automatically matches the optimal feed speed and ensures that the speed change rate of adjacent trajectory points does not exceed 20% through speed smoothing. Compared with traditional uniform parameter processing, this method can reduce the processing time by 70%, significantly improving production efficiency while ensuring the quality of deburring.

[0021] (3) In the present invention, a complete vision-guided deburring system architecture is constructed, realizing full-process automation from data acquisition, trajectory planning to execution control. The system integrates three-dimensional vision sensors, industrial robots, flexible grinding units and industrial control systems. Through eye-in-hand point cloud acquisition, hand-eye calibration and multi-subsystem collaborative control, workpiece scanning, burr segmentation, trajectory generation and deburring operations can be completed without human intervention. The automated operation mode not only reduces the intensity of manual labor, but also eliminates human operation errors through standardized processes, so that the processing accuracy and consistency are significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 A schematic flow chart of a method for adaptively deflashing a molded part provided by an embodiment of the present invention;

[0024] Figure 2 A simple schematic diagram of an adaptive deflashing method for a molded part provided by an embodiment of the present invention;

[0025] Figure 3 A schematic structural diagram of an adaptive deflashing system for molded parts provided by an embodiment of the present invention;

[0026] Figure 4 An enlarged view of position A in the structural schematic diagram of an adaptive deflashing system for a molded part provided by an embodiment of the present invention;

[0027] Figure 5 A histogram showing the error distribution between the trajectory and the true value of an adaptive deflashing method for molded parts provided by an embodiment of the present invention and a traditional method;

[0028] Figure 6 A comparison chart of the trajectory and true value fitting effects of an adaptive deflashing method for molded parts provided by an embodiment of the present invention and other methods.

[0029] In the figure: 11. Workbench; 12. Fixture; 13. Air pump; 14. Flexible grinding unit; 15. Point cloud camera; 16. Rotary file; 17. ABB robotic arm; 18. Electrical control cabinet; 19. ABB control cabinet; 21. Grinding unit controller; 22. System base. DETAILED DESCRIPTION

[0030] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0031] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0032] In the embodiments of the present invention, the terms "image" and "picture" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same. The terms "of," "corresponding," and "corresponding" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same.

[0033] In the embodiments of the present invention, sometimes a subscript such as W1 may be mistakenly written as a non-subscript form such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0034] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0035] Reference Manual Figure 1 , which shows a flow chart of an adaptive deflashing method for a molded part provided by an embodiment of the present invention.

[0036] An embodiment of the present invention provides an adaptive deflashing method for molded parts. The method can be implemented by an adaptive deflashing system for molded parts. The adaptive deflashing system for molded parts can be a terminal or a server. The processing flow of the adaptive deflashing method for molded parts can include the following steps:

[0037] S1. Data acquisition and preprocessing: An eye-on-hand point cloud camera is used to collect point cloud data of the workpiece surface. The ICP algorithm is used for multi-view point cloud stitching. The point cloud is processed by an octree-based voxel downsampling algorithm. The feature points of the parting surface are manually selected, and the RANSAC plane fitting algorithm is used to obtain the flash growth plane parameters.

[0038] In one possible implementation, S1 further includes:

[0039] The parameters of the fringe growth plane include the position p of the fringe growth plane plane and normal vector n plane .

[0040] It should be noted that step S1 is used for the collection and initial processing of point cloud data, providing a foundation for subsequent processes. The ICP (Iterative Closest Point) algorithm is an iterative closest point algorithm. Multi-view point cloud stitching using the ICP algorithm can be fused into a complete workpiece model. Point cloud processing using an octree-based voxel downsampling algorithm can reduce data volume while preserving geometric features.

[0041] S2. Fly point cloud segmentation: Use the FPFH algorithm to roughly align the scanned point cloud with the CAD model point cloud, extract the potential flash area based on the registration results, calculate the Euclidean distance feature and normal vector difference feature, and segment the flash point cloud using the judgment function and threshold.

[0042] It should be noted that step S2 is used to segment the scanned data and identify the flash data therein, which has two purposes:

[0043] 1) Used for speed tuning module to measure burr height;

[0044] 2) Remove the influence of burrs on surface deformation, making surface deformation recognition more accurate.

[0045] In one possible implementation, S2 further includes:

[0046] Extracting potential flash area includes extracting the distance from the flash growth plane p plane ,n plane The potential flash area p within a certain range on both sides potential , and calculate the feature quantity.

[0047] In a possible implementation, calculating the feature quantity specifically includes:

[0048] Suppose point p in the scanned point cloud i The corresponding point in the CAD model point cloud is q i , and calculate the following feature quantities:

[0049] Euclidean distance feature: f dist (p i )=|p i -q i |

[0050] Normal vector difference characteristics:

[0051] in, and Point p i and q i The unit normal vector at ;

[0052] Based on the above features, the judgment function for extracting the flash point cloud is defined as follows:

[0053] B(p i )=ω1f dist (p i )+ω2f normal (p i )

[0054] Among them, ω1 and ω2 are weight coefficients, which are determined through experimental optimization. The extraction rule of the flash point cloud is:

[0055] P burr =p i ∣B(p i )>B threshold ,p i ∈P potential

[0056] Among them, B threshold is the judgment threshold;

[0057] Finally, P burr Segmented from the scan point cloud, the pure scan data P with the flash removed is obtained no-burr .

[0058] S3. Tolerance-constrained trajectory pose optimization: Construct an energy function containing data terms, regularization terms, and tolerance constraint terms, constrain the displacement vector, use K-nearest-neighbor inverse distance weighted interpolation to calculate the deformation of the trajectory points, and update the trajectory pose.

[0059] In a possible implementation, S3 further includes:

[0060] This step is used to detect the deformation of the workpiece surface and adjust the position of the trajectory attached to the surface according to the surface deformation. no-burr Every point P in i , its deformation displacement vector to the target point cloud is v i , the specific steps are as follows:

[0061] Construct an energy function with three terms:

[0062] E=E fit +E reg +E cons

[0063] The data items are Used to measure point cloud alignment error;

[0064] The regularization term is Used to constrain the continuity of the deformation field;

[0065] The tolerance constraint is E cons=γ·H(|v i |-τ), used for tolerance constraints, where τ is the deformation threshold set based on process tolerance, H(x) is a step function used to impose a penalty when the deformation exceeds the threshold. i To constrain: |v i |≤τ, the deformation of the trajectory points is calculated using K-nearest neighbor inverse distance weighted interpolation to form a deformation field. The trajectory posture is updated through deformation field interpolation, and the surface deformation is applied to the trajectory, thereby achieving the purpose of adjusting the tool trajectory.

[0066] S4. Trajectory speed optimization based on grid analysis: The flash point cloud is projected onto the growth plane and a grid system is established. The contour points are sorted and iteratively expanded to generate an equidistant outer contour. The flash height is calculated based on the number of layers and layer height. The optimal feed speed is obtained through mapping and interpolation, and the speed is smoothed.

[0067] In a possible implementation, S4 further includes:

[0068] This step is used to adjust the feed speed of the track according to the height of the non-edited text to form an efficient track. The specific steps are as follows:

[0069] S401, projecting the flash point cloud onto the growth plane and establishing a grid system with a grid size of δ;

[0070] S402, sorting the contour points using a kd-tree nearest neighbor search algorithm to obtain the original contour C0;

[0071] S403, perform iterative expansion operation on C0 to generate n equidistant outer contours {C1, C2, ..., C n}, calculate the height of the fin corresponding to each trajectory point according to the number of layers and layer height;

[0072] S404, establishing a mapping relationship between burr height and feed speed using pre-measured optimal grinding parameters of the material, and obtaining an optimal trajectory feed speed by performing height interpolation calculation;

[0073] S405: Smoothing the trajectory speed to ensure that the speed variation between adjacent points does not exceed 20%.

[0074] Specifically, refer to the appendix of the manual. Figure 2First, scanned data of a real workpiece and a CAD model with parting surfaces are used as initial input, and both undergo data preprocessing. Subsequently, the initial pose is obtained through rigid registration, and feature analysis is used to segment the point cloud to produce a flash point cloud. Next, corrections are made from both a speed and pose perspective. First, the flash height distribution is analyzed, and speed correction is achieved through velocity mapping. Second, the TC-nicp algorithm is used to correct the trajectory and achieve pose correction. Finally, the results of speed and pose correction are integrated to generate a deflashing trajectory for machining.

[0075] The present invention also provides an adaptive deflashing system for molded parts, which is applied to an adaptive deflashing method for molded parts, comprising:

[0076] The data acquisition system, execution system, control system and software system are characterized by:

[0077] The data acquisition system includes a point cloud camera and a hand-eye calibration system. The point cloud camera is installed in an eye-on-hand manner, acquiring multi-viewpoint point cloud data as the robot moves. The point cloud data is transmitted via Ethernet connection to an industrial control computer. The hand-eye calibration system includes a calibration board and calibration algorithm software to establish the conversion relationship between the camera coordinate system and the robot base coordinate system.

[0078] The execution system includes a flexible grinding unit, a rotary file, and an industrial robot body. The flexible grinding unit has a floating compensation function and is connected to the standard interface of the robot end through a flange. The rotary file uses a cylindrical carbide tool. The tool diameter is selected according to the characteristics of the workpiece and maintains a 90° processing angle with the workpiece surface. The industrial robot body is a 6-degree-of-freedom articulated industrial robot with a payload of no less than 10kg.

[0079] The control system includes an industrial computer, an electrical control unit, a robot controller, and a communication interface module. The industrial computer is used for data processing and method implementation, and calculates the processing trajectory. The electrical control unit integrates a PLC controller to achieve coordinated control of various execution components. It contains a safety protection circuit and has an emergency stop function and a safety protection mechanism. The robot controller provides real-time trajectory planning and interpolation functions, and has an Ethernet communication interface for data exchange with the industrial computer. The communication interface module uses industrial Ethernet as the main communication method and supports TCP / IP and Modbus communication protocols.

[0080] The software system includes a point cloud processing module developed based on Open3D and a host computer communication module based on open_ABB.

[0081] In one possible implementation, the point cloud camera is a Revopoint Surface HD 50 infrared structured light camera.

[0082] In a possible implementation, the industrial robot body is an ABB IRC2600-20 robot body, and the robot controller is an ABB IRC5 controller.

[0083] In a possible implementation, the flexible polishing unit is Hongyao FD08, and the industrial control computer is configured with an Intel i5-9300H processor and 20GB memory.

[0084] In the embodiment provided by the present invention, the adaptive deflashing system for molded parts is composed of a data acquisition system, an execution system, a control system and a software system, and the subsystems work together to realize automated deflashing.

[0085] The data acquisition system includes a point cloud camera and a hand-eye calibration system. The point cloud camera, mounted in an eye-on-hand configuration, captures multi-viewpoint point cloud data as the robot moves. It supports 3D vision sensors such as structured light, binocular, and time-of-flight, and communicates with an industrial computer via Ethernet. The hand-eye calibration system, which includes a calibration board and algorithm software, establishes the transformation relationship between the camera coordinate system and the robot's base coordinate system.

[0086] The execution system consists of a flexible grinding unit, a rotary file, and the industrial robot itself. The flexible grinding unit connects to the standard end-effector interface of the robot via a flange and features floating compensation to accommodate variations in the workpiece surface contour. The rotary file is a cylindrical carbide tool with a diameter tailored to the workpiece's characteristics, maintaining a 90° perpendicular angle to the workpiece surface during machining. The industrial robot itself is an articulated unit with six degrees of freedom and a payload of at least 10 kg, meeting the end-effector weight requirements.

[0087] The control system consists of an industrial computer, an electrical control unit, a robot controller, and a communication interface module. The industrial computer is responsible for data processing, algorithm implementation, and machining trajectory calculation. The electrical control unit integrates a PLC controller to coordinate and control the actuators. It also includes safety protection circuits, an emergency stop function, and multiple protection mechanisms. The robot controller provides real-time trajectory planning and interpolation, interacting with the industrial computer via Ethernet. The communication interface module uses Industrial Ethernet and supports protocols such as TCP / IP and Modbus.

[0088] The software system is based on the Windows 10 platform and is programmed in Python. It includes a point cloud processing module based on Open3D and a host computer communication module based on open_ABB.

[0089] The system connection relationship is as follows: the point cloud camera is installed at the end of the robot, the flexible grinding unit is connected to the end of the robot through a flange, the industrial control computer is connected to the robot controller, electrical control unit and point cloud camera through Ethernet, and the electrical control unit is connected to the actuator through a hard wire.

[0090] The operation process is as follows: first, the coordinate system conversion relationship is established through the hand-eye calibration system; the robot drives the camera to scan the workpiece to obtain point cloud data; the industrial control computer processes the data and generates an adaptive trajectory; the robot executes the trajectory, combining force control and position compensation to remove burrs, and each system works together to complete the processing task.

[0091] Reference Manual Figure 3 -Attached Figure 4 , showing a schematic structural diagram of an adaptive deflashing device for a molded part provided by an embodiment of the present invention.

[0092] The present invention also provides an adaptive deflashing device for molded parts, which uses a system base 22 as a basic support, on which a workbench 11 is arranged. The workbench 11 is equipped with a clamp 12 for fixing the workpiece to be processed. An air pump 13 provides power for the system operation. An ABB robotic arm 17 is installed on the system base 22, and its end is connected to a flexible grinding unit 14. The front end of the flexible grinding unit 14 is equipped with a rotary file 16 to perform the actual deflashing operation. The point cloud camera 15 is installed in an eye-in-hand style and moves with the ABB robotic arm 17 to obtain multi-view three-dimensional point cloud data of the workpiece. The electrical control cabinet 18 integrates a PLC controller and safety protection circuit to coordinate and control various execution components and ensure system safety; the ABB control cabinet 19 cooperates with the ABB robotic arm 17 to achieve real-time trajectory planning, interpolation and motion control; and the grinding unit controller 21 accurately controls the flexible grinding unit 14. During operation, the fixture 12 fixes the workpiece, the point cloud camera 15 scans and collects data, which is processed by the electrical control cabinet 18 to generate a processing trajectory. The ABB robotic arm 17 drives the flexible grinding unit 14 and the rotary file 16 to execute the trajectory. Combined with the power of the air pump 13, the various components work together through the system base 22 to complete the automated adaptive deflashing processing of molded parts.

[0093] Table 1 shows a comparison of the key performance indicators of the method, system, and device provided by the present invention when implemented with those based on traditional non-rigid registration tuning methods.

[0094] Table 1 Comparison of trajectory error and deflashing time between this method and the traditional method

[0095]

[0096] like Figure 5 The figure shows the trajectory deviation distribution of the proposed method and the traditional non-rigid method. The blue bars represent the proposed method, and the red bars represent the traditional non-rigid method. The mean deviation of the proposed method is 0.70±0.37mm, while the mean of the traditional method is 1.00±0.32mm. This shows that the deviation distribution of the proposed method is more concentrated in the low-error range, effectively reducing the deviation between the trajectory and the true value, verifying the superiority of the proposed method in trajectory accuracy control.

[0097] like Figure 6 As shown in the figure, (a), (b), and (c) show the trajectory error distribution based on rigid registration, traditional non-rigid registration, and this method respectively. The trajectories based on rigid registration (a) and traditional non-rigid registration (b) deviate greatly from the true value, while the trajectory of this method (c) (red) is closer to the true value (green), showing smaller errors in both two-dimensional contours and three-dimensional spatial distribution, proving that this method is superior to traditional registration methods in trajectory generation accuracy.

[0098] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0099] (1) In the present invention, the tolerance-constrained TC-NICP registration algorithm is used to achieve accurate recognition of workpiece surface deformation and dynamic adjustment of trajectory posture. The algorithm introduces process tolerance constraints in the point cloud registration process, effectively suppressing the excessive deformation problem of traditional non-rigid registration methods, and reducing the average error of trajectory points from 1.00 mm in traditional methods to within 0.71 mm. By compensating for the deformation of the workpiece caused by cooling shrinkage in real time, the fit between the trajectory and the actual processing position is significantly improved, ensuring that the tool path accurately fits the complex curved surface, thereby improving the accuracy and consistency of the deburring operation, and is particularly suitable for the post-processing of high-precision molded parts;

[0100] (2) In the present invention, a trajectory speed tuning method based on grid analysis realizes dynamic optimization of processing parameters by establishing an adaptive mapping relationship between burr height and feed speed. This technology projects the burr point cloud onto the growth plane and constructs a grid system. It combines the kd tree algorithm and iterative expansion profile generation technology to accurately measure the burr height distribution in different areas. According to the pre-calibrated material grinding parameters, it automatically matches the optimal feed speed and ensures that the speed change rate of adjacent trajectory points does not exceed 20% through speed smoothing. Compared with traditional uniform parameter processing, this method can reduce the processing time by 70%, significantly improving production efficiency while ensuring the quality of deburring.

[0101] (3) In the present invention, a complete vision-guided deburring system architecture is constructed, realizing full-process automation from data acquisition, trajectory planning to execution control. The system integrates three-dimensional vision sensors, industrial robots, flexible grinding units and industrial control systems. Through eye-in-hand point cloud acquisition, hand-eye calibration and multi-subsystem collaborative control, workpiece scanning, burr segmentation, trajectory generation and deburring operations can be completed without human intervention. The automated operation mode not only reduces the intensity of manual labor, but also eliminates human operation errors through standardized processes, so that the processing accuracy and consistency are significantly improved.

[0102] The above content is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

[0103] There are a few points to note:

[0104] (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention. Other structures may refer to conventional designs.

[0105] (2) For the sake of clarity, the thickness of layers or regions in the drawings used to describe the embodiments of the present invention are exaggerated or reduced, that is, these drawings are not drawn to scale. It is understood that when an element such as a layer, film, region, or substrate is referred to as being "on" or "under" another element, the element may be "directly" on or "under" the other element or intervening elements may be present.

[0106] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to form new embodiments.

[0107] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for adaptively deflashing a molded part, characterized in that: include: S1. Data acquisition and preprocessing: An eye-on-hand point cloud camera is used to collect point cloud data of the workpiece surface. The ICP algorithm is used for multi-view point cloud stitching. The point cloud is processed by an octree-based voxel downsampling algorithm. The feature points of the parting surface are manually selected, and the RANSAC plane fitting algorithm is used to obtain the flash growth plane parameters. S2. Flash point cloud segmentation: Use the FPFH algorithm to roughly align the scanned point cloud with the CAD model point cloud, extract the potential flash area based on the registration results, calculate the Euclidean distance feature and the normal vector difference feature, and segment the flash point cloud using the decision function and threshold; S3, tolerance-constrained trajectory pose optimization: Construct an energy function containing data terms, regularization terms, and tolerance constraint terms, constrain the displacement vector, use K-nearest-neighbor inverse distance weighted interpolation to calculate the trajectory point deformation, and update the trajectory pose; S4. Trajectory speed optimization based on grid analysis: The flash point cloud is projected onto the growth plane and a grid system is established. The contour points are sorted and iteratively expanded to generate an equidistant outer contour. The flash height is calculated based on the number of layers and layer height. The optimal feed speed is obtained through mapping and interpolation, and the speed is smoothed.

2. A method for adaptively deflashing a molded part according to claim 1, characterized in that: Said S1 further comprises: The fringe growth plane parameters include the position p of the fringe growth plane plane and normal vector n plane .

3. The method for adaptively deflashing a molded part according to claim 2, wherein: Said S2 further comprises: The extraction of potential flash area includes extracting the distance from the flash growth plane {p plane ,n plane Potential flash area p within a certain range on both sides potential , and calculate the feature quantity.

4. A method for adaptively deflashing a molded part according to claim 3, characterized in that: The calculation of the characteristic quantity specifically includes: Suppose point p in the scanned point cloud i The corresponding point in the CAD model point cloud is q i , and calculate the following feature quantities: Euclidean distance feature: f dist (p i )=|p i -q i | Normal vector difference characteristics: in, and Point p i and q i The unit normal vector at ; Based on the above features, the judgment function for extracting the flash point cloud is defined as follows: B(p i )=ω1f dist (p i )+ω2f normal (p i ) Among them, ω1 and ω2 are weight coefficients, which are determined through experimental optimization. The extraction rule of the flash point cloud is: P burr =p i ∣B(p i )>B threshold ,p i ∈P potential Among them, B threshold is the judgment threshold; Finally, P burr Segmented from the scan point cloud, the pure scan data P with the flash removed is obtained no-burr .

5. The method for adaptively deflashing a molded part according to claim 3, wherein: Said S3 further comprises: This step is used to detect the deformation of the workpiece surface and adjust the position of the trajectory attached to the surface according to the surface deformation. no-burr Every point P in i , its deformation displacement vector to the target point cloud is v i , the specific steps are as follows: Construct an energy function with three terms: E=E fit +E reg +E cons The data items are Used to measure point cloud alignment error; The regularization term is Used to constrain the continuity of the deformation field; The tolerance constraint is E cons =γ·H(|v i |-τ), used for tolerance constraint, where τ is the deformation threshold set based on process tolerance, H( x ) is a step function used to impose a penalty when the deformation exceeds the threshold. i To constrain: |v i |≤τ, the deformation of the trajectory points is calculated using K-nearest neighbor inverse distance weighted interpolation to form a deformation field. The trajectory posture is updated through deformation field interpolation, and the surface deformation is applied to the trajectory, thereby achieving the purpose of adjusting the tool trajectory.

6. The method for adaptively deflashing a molded part according to claim 1, wherein: Said S4 further comprises: This step is used to adjust the feed speed of the track according to the height of the non-edited text to form an efficient track. The specific steps are as follows: S401, projecting the flash point cloud onto the growth plane and establishing a grid system with a grid size of δ; S402, sorting the contour points using a kd-tree nearest neighbor search algorithm to obtain the original contour C0; S403, perform iterative expansion operation on C0 to generate n equidistant outer contours {C1, C2, ..., C n }, calculate the height of the fin corresponding to each trajectory point according to the number of layers and layer height; S404, establishing a mapping relationship between burr height and feed speed using pre-measured optimal grinding parameters of the material, and obtaining an optimal trajectory feed speed by performing height interpolation calculation; S405: Smoothing the trajectory speed to ensure that the speed variation between adjacent points does not exceed 20%.

7. An adaptive deflashing system for molded parts, comprising a data acquisition system, an execution system, a control system, and a software system, characterized in that: The data acquisition system includes a point cloud camera and a hand-eye calibration system. The point cloud camera adopts the eye-on-hand installation method to acquire multi-view point cloud data as the robot moves. The point cloud data is transmitted by connecting to the industrial control computer via Ethernet. The hand-eye calibration system includes a calibration board and calibration algorithm software for establishing the conversion relationship between the camera coordinate system and the robot base coordinate system. The execution system includes a flexible grinding unit, a rotary file, and an industrial robot body. The flexible grinding unit has a floating compensation function and is connected to the standard interface of the robot end through a flange. The rotary file uses a cylindrical carbide tool with a diameter selected according to the characteristics of the workpiece and maintains a 90° processing angle with the workpiece surface. The industrial robot body is a 6-degree-of-freedom articulated industrial robot with a payload of no less than 10 kg. The control system includes the industrial control computer, an electrical control unit, a robot controller and a communication interface module. The industrial control computer is used for data processing and method implementation, and calculation of the processing trajectory. The electrical control unit is integrated with a PLC controller to achieve coordinated control of various execution components, and contains a safety protection circuit, an emergency stop function and a safety protection mechanism. The robot controller provides real-time trajectory planning and interpolation functions, and has an Ethernet communication interface for data exchange with the industrial control computer. The communication interface module uses industrial Ethernet as the main communication method and supports TCP / IP and Modbus communication protocols. The software system includes a point cloud processing module developed based on Open3D and a host computer communication module based on open_ABB.

8. A method for adaptively deflashing a molded part according to claim 7, characterized in that: include: The point cloud camera is a revopoint Surface HD 50 infrared structured light camera.

9. The method for adaptively deflashing a molded part according to claim 7, wherein: include: The industrial robot body is an ABB IRC2600-20 robot body, and the robot controller is an ABB IRC5 controller.

10. The method for adaptively deflashing a molded part according to claim 7, wherein: include: The flexible polishing unit is Hongyao FD08, and the industrial control computer is configured with an Intel i5-9300H processor and 20GB memory.

Images