Robot gluing three-dimensional conformal corner path generation method based on linear laser online guidance

Abstract

This invention relates to a method for generating a 3D conformal corner path for robot adhesive application based on online line laser guidance. The method includes: acquiring point cloud data of the area to be processed on a curved workpiece using a line laser contour sensor; preprocessing the point cloud data and extracting at least one key feature point, such as a boundary point or inflection point; converting the coordinate values ​​of each key feature point in the visual coordinate system to coordinate values ​​in the base coordinate system using a transformation matrix; fitting the point cloud of each key feature point in the base coordinate system to obtain a spatial curve function; re-selecting points at equal intervals based on the spatial curve function; calculating the tangent vector and normal vector of the current point position by combining previous and subsequent points; and generating a path based on the processing position and posture requirements. This method can obtain the processing pose at each data acquisition point of the curved workpiece, including the position of the path points in space and the posture corresponding to the path points that meets the process requirements, providing accurate processing path guidance for the system.

Description

Technical Field

[0001] This invention belongs to the technical field of machining path generation methods for curved workpieces, and particularly relates to a method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance. Background Technology

[0002] The field of aircraft manufacturing involves complex geometries. An aircraft's geometry (aerodynamic shape) is jointly formed by the geometries of its main components, such as the wings, fuselage, and tail (divided into horizontal or vertical tail). The geometry of modern aircraft must meet requirements for aerodynamic characteristics and stealth performance.

[0003] Hyperbolic aluminum sheet, also known as hyperbolic aluminum sheet skin, is a special aluminum material made of aluminum alloy and processed using a special technique to achieve a hyperbolic shape. Unlike traditional flat shapes, the curved hyperbolic shape better adapts to curves or complex surface structures, providing better fit and aesthetics. Furthermore, hyperbolic aluminum sheet boasts advantages such as lightweight, high strength, high rigidity, pressure resistance, impact resistance, vibration resistance, oxidation resistance, corrosion resistance, durability, and good plasticity. It is widely used in aerospace, construction, automotive, and shipbuilding industries, playing a crucial role in constructing aircraft fuselages, ship hulls, building facades, and automobile bodies.

[0004] Complex curved surface profiles are very common in the aerospace manufacturing industry, making technologies for machining complex curved surface workpieces particularly important. However, hyperboloids are structurally complex and difficult to machine, and current machining methods are mostly based on manual processing or theoretical mathematical models. While manual processing is more flexible, its subjective nature results in low accuracy and efficiency in machining complex curved surfaces, and the pollution in most machining environments can cause irreversible harm to human health. Machining based on theoretical mathematical models requires preparing a mathematical model corresponding to the workpiece in advance, and planning the machining method and path based on the mathematical model through offline programming. However, this method has limitations; it is poorly adaptable to workpiece deformation caused by gravity or tooling forces, and to deviations between the actual pose and the mathematical model pose caused by workpiece placement errors. It cannot simultaneously meet the requirements of high-precision, high-flexibility online, conformal machining of complex curved surface workpieces. Summary of the Invention

[0005] To overcome the aforementioned shortcomings of existing complex curved surface workpiece processing technologies, this invention proposes a novel method for generating three-dimensional conformal corner paths for robot adhesive application based on online line laser guidance. This method can perform online conformal processing based on the actual contour of complex curved surface workpieces.

[0006] Since complex surfaces often exhibit a three-dimensional hyperboloid shape, this method plans a spatial rotation angle encompassing three dimensions when guiding the robot to process complex surfaces. This is also the key design point in the method of this invention.

[0007] Specifically, this invention provides a method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance. This method includes:

[0008] S1. Use a line laser contour sensor to acquire point cloud data of the area to be processed on the curved workpiece;

[0009] S2. Preprocess the obtained point cloud data and extract at least one key feature point. The extracted key feature point is required to describe the spatial structural features of the area to be processed on the curved workpiece.

[0010] S3. By using a transformation matrix, the coordinate values ​​of each key feature point in the visual coordinate system are transformed into coordinate values ​​in the base coordinate system, thus obtaining the set of key feature points in the base coordinate system;

[0011] S4. Use the least squares method or B-spline curve fitting method to fit the point cloud of each key feature point in the base coordinate system to obtain the spatial curve function;

[0012] S5. Based on the space curve function and process parameters, re-select points at equal intervals;

[0013] S6. Calculate the tangent vector and normal vector of the current point by combining the preceding and following points;

[0014] S7. If a single point cloud data contains only one key feature point, then the path is generated directly based on the tangent vector and normal vector of the current point position, combined with the processing position and attitude requirements; if a single point cloud data contains two or more key feature points, then the remaining feature points are fitted with spatial curves, and the intersection points with the normal vector or plane of the current point position are calculated, and the path is generated in combination with the processing position and attitude requirements.

[0015] Furthermore, when using the robot adhesive coating three-dimensional conformal corner path generation method based on line laser online guidance of the present invention for online processing, the line laser contour sensor is installed in front of the processing end head in the direction of travel so that the path points corresponding to the processing position are calculated before the processing end head travels to a certain processing position.

[0016] Furthermore, in the method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance of the present invention, the processing end is an adhesive application head or a milling cutter head.

[0017] Furthermore, in step S3 of the method for generating a three-dimensional conformal rotation path for robot adhesive application based on online line laser guidance, the key feature points in the key feature point set under the base coordinate system exist in the order of data acquisition.

[0018] Furthermore, in step S4 of the method for generating a three-dimensional conformal rotation path for robot adhesive application based on online line laser guidance, the least squares method or B-spline curve fitting method is used to fit the point cloud of each key feature point in the base coordinate system to obtain a spatial curve function; including:

[0019] (1) p1, p2, and p3 are three adjacent key feature points in the base coordinate system. Connect the current key feature point p2 with the preceding and subsequent key feature points to construct three vectors. and Based on these three vectors and the current key feature point p2, the tangent vector is calculated using the following formula. and normal vector

[0020]

[0021]

[0022] A(x-x0)+B(y-y0)+C(z-z0)=0;

[0023] (2) Construct the angle bisector based on the tangent vector and the current key feature point p2. Or through the current point and vector Establish planes S1 and S2, and the angle bisector, respectively. The solution is obtained by using the point normal equation, and the angle bisector is obtained by using the key feature point p2 shared by all faces, as well as the mean vector of faces S1 and S2. The normal vector.

[0024] Due to the surface Determined by both surfaces S1 and S2, and taking into account the relative positions of preceding and subsequent points p1 and p3 with the current point p2, the surface... The spatial relationship can integrate the spatial structure of the preceding and following curved surfaces, smoothly expressing the cross-section of the area to be processed at point p2. This cross-section, compared to the two-dimensional field of view formed by the line laser during actual operation, more closely matches the actual structural characteristics of the complex curved surface to be processed. Because the line laser sensor and the processing end are rigidly fixed, their relative positional relationship is fixed and cannot flexibly adapt to the changing spatial structure of the surface to be processed. However, the ideal cross-section calculated by mathematical formula based on the spatial positional relationship of the actual feature points on the workpiece in the base coordinate system, using the workpiece as a reference, has higher reliability, stability, and accuracy.

[0025] Furthermore, in the method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance of the present invention, if surfaces S1 and S2 coincide, the angle bisector surface... That is, the coincident plane.

[0026] Furthermore, in the method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance, due to processing requirements, the processing end is often positioned parallel to the surface to be processed, perpendicular to the surface to be processed, or offset by a specified angle in a certain direction at a parallel or perpendicular angle for path planning. However, regardless of the angle between the processing end and the surface to be processed, the TCP always falls on the surface. Inside, since the normal vector is also located on the surface Therefore, after offsetting the normal vector by a specified distance according to the processing requirements, the attitude vector is also located on the surface. Inside, and according to processing requirements, it forms a specified angle with the normal vector.

[0027] Furthermore, the present invention also includes the following method for generating a three-dimensional conformal rotation path for robot adhesive application based on online line laser guidance: converting the calculated vector into a rotation matrix, and then converting the rotation matrix into Euler angles;

[0028] For the same vector Let on the unit orthogonal basis The coordinates below are [x, y, z]. T After one rotation, the unit orthogonal basis It became a vector The coordinates under the new orthogonal basis are [x′, y′, z′]. T This combination immediately yields the rotation matrix R; let the Euler angles of the three axes x, y, z be θ. x θ y θ z The sine and cosine values ​​are c and c, respectively. x s x c y s y c z s z The algorithm is as follows:

[0029]

[0030] θ x =atan2(r 32 r 33 )

[0031]

[0032] θ z =atan2(r 21 ,r11 )

[0033] Furthermore, in step S4 of the method for generating a three-dimensional conformal rotation path for robot adhesive application based on online line laser guidance, the least squares method or B-spline curve fitting method is used to fit the point cloud of each key feature point in the base coordinate system to obtain a spatial curve function; including:

[0034] The least squares method is used to fit the space curve, and the final space curve parameters are obtained by solving the overdetermined equations. For the Nth order curve polynomial, the following equation is given:

[0035]

[0036] Where y is the y-axis coordinate of the higher-order curve, x is the x-axis coordinate of the independent variable, i is the specific degree of different degree terms, and k is the parameter of each degree term;

[0037] Further, the above equation can be constructed into a system of equations:

[0038] Y = XK;

[0039] Where Y is an [M×1] dimensional matrix composed of parameter points, X is an [M×6] dimensional matrix composed of parameter points, M represents the number of coordinate points involved in the fitting, and K is the [6×1] dimensional coefficient matrix to be determined;

[0040] By applying the least-squares solution theorem for overdetermined equations, we can obtain:

[0041] K = (X T X) -1 X T Y;

[0042] Among them, X T X is the transpose of the [M×6]-dimensional matrix X composed of parameter points;

[0043] Then, the coefficient matrix K is obtained through matrix operations; the spatial fitting curve corresponding to the set of key feature points can be obtained through the above method.

[0044] Furthermore, the present invention's method for generating a three-dimensional conformal turning path for robot adhesive application based on online line laser guidance also includes: using the least squares method or B-spline curve fitting method to perform overall fitting of the calculated path points and their corresponding postures in three-dimensional space, and performing equal-interval sampling and processing path optimization according to processing requirements to meet the robot's command requirements.

[0045] In summary, the machining poses at each data acquisition point of the complex curved surface workpiece can be obtained, including the positions of path points in space and the poses (Euler angles) corresponding to the path points that meet the process requirements. The higher-order fitting curves corresponding to this set of feature points are obtained using the above method, and the obtained path is proportionally divided using a quantitative truncation method, thereby achieving further segmentation of the path points and providing precise machining path guidance for the system. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the following drawings are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 A schematic diagram of the mathematical model of a partial outline of the window frame of the AC-311 helicopter.

[0048] Figure 2 This is a schematic diagram of the installation position of the line laser contour sensor and the processing end according to an embodiment of the present invention (the direction of travel points upwards in this image).

[0049] Figure 3 This is a schematic diagram of an online conformal machining process according to an embodiment of the present invention.

[0050] Figure 4 This is a schematic diagram of the preceding segment path according to an embodiment of the present invention.

[0051] Figure 5 This is a schematic diagram of the point cloud fitting curve function composed of multiple key feature points in the base coordinate system according to an embodiment of the present invention.

[0052] Figure 6 This is a schematic diagram of the spatial structure of the angle bisector plane according to an embodiment of the present invention.

[0053] Figure 7 This is a schematic diagram of a high-order fitting curve corresponding to a set of feature points according to an embodiment of the present invention.

[0054] Figure 8 This is a schematic diagram illustrating the solution of the normal vector of the surface to be processed according to an embodiment of the present invention.

[0055] Figure 9 This is a schematic diagram illustrating the implementation process of the method of the present invention. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. This invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this invention.

[0057] At the same time, it should be understood that the scope of protection of the present invention is not limited to the specific embodiments described below; it should also be understood that the terminology used in the embodiments of the present invention is for describing specific embodiments and not for limiting the scope of protection of the present invention.

[0058] Example: A method for generating a 3D conformal corner path for robot adhesive application based on online line laser guidance (e.g., Figure 9 (As shown)

[0059] To address the online, conformal machining needs of complex curved surface workpieces in aerospace and automotive manufacturing, this study selects typical spatial curved surface workpieces and utilizes a flexible and applicable line laser contour sensor to acquire visual data. The research focuses on developing an online, conformal machining technology for complex curved surfaces based on line laser sensor point cloud data, with the main components being the calculation of three-dimensional spatial rotation angles and path planning.

[0060] Taking the AC-311 civilian helicopter as an example, its fuselage is teardrop-shaped and its window frame has a complex curved structure, which makes spatial path planning difficult when processing such as applying sealant to its window frame.

[0061] Based on the window frame outline of the AC-311 helicopter, a partial outline diagram of the hyperboloid complex structure window frame is drawn to explain the spatial angle calculation method. For example... Figure 1 As shown, the mathematical model simulates a complex curved window frame that bends to the right. The inner side of the window frame has an inner step structure that is thinner than the surrounding surface. This structure is used to place gaskets, apply sealant, and install the windshield.

[0062] Because online machining is performed, the path points corresponding to a machining position need to be calculated before the end-effector reaches that position. Therefore, the line laser profile sensor needs to be installed in front of the machining end (applying head or milling cutter head, etc.) in the direction of travel. Figure 2 As shown, the mounting position of the line laser contour sensor and the processing end can be designed in this form, with the direction of travel pointing upwards in this image.

[0063] like Figure 3As shown, the gray object on the left is the workpiece to be processed, and the orange object on the right is the robot's robotic arm. The robotic arm moves the end in the center of the image to perform online, conformal processing. The top of the image represents the main direction of movement, and blue represents the line laser. During processing, the line laser scans the workpiece, acquires point cloud data, and performs a series of calculations. It integrates the calculation results from multiple locations to complete segment path planning and combines the overall contour to update the overall path in real time. The gray cylinder at the lower end of the laser line is the processing end, completing the online, conformal processing.

[0064] When the workpiece to be processed has a complex hyperboloid structure, point cloud data of the workpiece can be acquired at the starting position (a position where one end of the workpiece can be scanned or a planned starting position). As the end point moves a certain distance in the direction of travel, multiple point cloud data will be acquired. Because there is a certain distance between the line laser and the processing end point, the line laser can continuously scan a distance from the starting position while ensuring that the processing end point does not enter the processing area and does not come into contact with the workpiece. It is required that the distance the line laser continuously scans from the starting position is less than the installation distance between the line laser and the processing end point, and a distance for the end point radius needs to be reserved based on the end point model and features. We define the path of the preceding sequence as a straight line moving from the starting position, collecting point cloud data, and ensuring that the end point does not enter the processing area.

[0065] Taking the previous section as an example, the calculation method for the overall structure of the workpiece to be processed is the same as that of the previous section. For example... Figure 4 As shown, multiple line laser point cloud data can be acquired in the preceding stage. By preprocessing the point cloud data and extracting key feature points, the yellow feature points shown in the figure can be obtained. Since the window frame structure has inner and outer steps, and the field of view of the selected line laser sensor is larger than the width of the inner step, regardless of whether the laser line strictly follows the surface normal vector direction of the hyperboloid edge, feature points at the intersection of the inner and outer steps (points falling on the inner step surface) and feature points at the boundary of the inner step can be acquired within the field of view. These two types of feature points, with the intersection of the inner and outer steps as the first key feature point and the boundary point of the inner step as the second key feature point, together describe the structural features of the inner step, i.e., the area to be processed, in space. Using a hand-eye transformation matrix, the coordinate values ​​of each key feature point in the visual coordinate system can be converted to coordinate values ​​in the base coordinate system. Thus, a set of key feature points in the base coordinate system can be obtained, and the points exist in a specific order according to the data acquisition sequence.

[0066] Due to mechanical vibrations or noise in the laser point cloud data during data acquisition by the robot carrying the end effector, the key feature points are not smooth in the base coordinate system. The least squares method or B-spline curve fitting method is used to fit the point cloud formed by the first and second key feature points in the base coordinate system, respectively, to obtain curve functions.

[0067] like Figure 5 As shown, p1, p2, and p3 are three adjacent key feature points in the base coordinate system. Connecting the current key feature point p2 with the preceding and subsequent key feature points respectively constructs three vectors. and Based on these three vectors and the current key feature point p2, the tangent vector is calculated using the following formula. and normal vector

[0068]

[0069]

[0070] A(x-x0)+B(y-y0)+C(z-z0)=0;

[0071] Construct the angle bisector based on the tangent vector and the current key feature point p2. Or through the current point and vector Establish planes S1 and S2, and the angle bisector, respectively. The solution is obtained by using the point normal equation, and the angle bisector is obtained by using the key feature point p2 shared by all faces, as well as the mean vector of faces S1 and S2. The normal vector.

[0072] Since both surfaces S1 and S2 contain p2, their relative positions may be intersecting or coincident. If surfaces S1 and S2 intersect, the line of intersection is l, and point p2 lies on line l. The line of intersection l can be solved by simultaneously solving surfaces S1 and S2.

[0073] like Figure 6 As shown, based on planes S1 and S2, construct the angle bisector of the two planes. Point p2 and the line of intersection l are both on the plane. The angle bisector can also be solved using the point normal equation, by using the key point p2 shared by all faces, and the mean vector of faces S1 and S2 as the angle bisector. The normal vector.

[0074] If surfaces S1 and S2 coincide, the angle bisector is the plane that coincides with S1.

[0075] Due to the surface Determined by both surfaces S1 and S2, and taking into account the relative positions of preceding and subsequent points p1 and p3 with the current point p2, the surface... The spatial relationship can integrate the spatial structure of the preceding and following curved surfaces, smoothly expressing the cross-section of the area to be coated at point p2. This cross-section, compared to the two-dimensional field of view formed by the line laser during actual operation, better matches the actual structural characteristics of the complex curved surface to be processed. Because the line laser sensor and the processing end are rigidly fixed, their relative positional relationship is fixed and cannot flexibly adapt to the changing spatial structure of the surface to be processed. However, the ideal cross-section calculated by mathematical formula based on the spatial positional relationship of the actual feature points on the workpiece in the base coordinate system, using the workpiece as a reference, has higher reliability, stability, and accuracy.

[0076] Due to processing requirements, the machining end is often positioned parallel to the surface to be machined, perpendicular to the surface to be machined, or offset by a specified angle in a certain direction, either parallel or perpendicular, for path planning. However, regardless of the angle between the machining end and the surface to be machined, the TCP (Current Travel Point) always falls on the surface. Inside, since the normal vector is also located on the surface Therefore, after offsetting the normal vector by a specified distance according to the processing requirements, the attitude vector is also located on the surface. Inside, and according to processing requirements, it forms a specified angle with the normal vector.

[0077] Space curve fitting employs the least squares method, obtaining the final curve parameters by solving overdetermined equations. As shown in the following equation, for an Nth-order curve polynomial:

[0078]

[0079] Where y is the y-axis coordinate of the higher-order curve, x is the x-axis coordinate of the independent variable, i is the specific degree of different degree terms, and k is the parameter of each degree term. Furthermore, the above equation can be constructed into a system of equations:

[0080] Y = XK;

[0081] Where Y is an [M×1] dimensional matrix composed of parameter points, and X is an [M×6] dimensional matrix composed of parameter points. T Let X be the transpose of the [M×6] dimensional matrix composed of parameter points, where M represents the number of coordinate points involved in the fitting, and K is the [6×1] dimensional coefficient matrix to be solved. Using the least squares solution theorem for overdetermined equations, we can obtain:

[0082] K = (X T X) -1 X T Y;

[0083] The coefficient matrix K is then obtained through matrix operations. Using the above method, the system will obtain the higher-order fitting curves corresponding to this set of feature points.

[0084] like Figure 7As shown, the red dashed line represents the location of the point cloud data acquired by the line laser sensor. The red dots are the first key feature points extracted and transformed into the base coordinate system based on the visual algorithm, and the red curve is the spatial curve C1 obtained after fitting. The yellow dots are the second key feature points extracted and transformed into the base coordinate system based on the visual algorithm, and the yellow arc is the fitted curve C2 obtained by the spatial curve fitting algorithm based on the first key feature points in the base coordinate system. The blue straight line represents the angle bisector calculated using the first key feature points before and after. The space curve obtained through fitting is perpendicular to each angle bisector. The intersections yield intersection points p′1, p′2, p′3..., p′ n Furthermore, the tangent vector and normal vector method is suitable for cases with only one feature point, and can also be used for multiple feature points. This can be achieved by offsetting the reference keypoint along the normal vector by a specified distance. The attitude vector is calculated jointly from the normal vector, tangent vector, and process requirements.

[0085] like Figure 8 As shown, connecting the same surface Two points p1 and p′1 give a vector This refers to the direction vector within the current cross-section that is parallel to the surface to be machined. If the process requires the machined end to be perpendicular to the surface to be machined, this can be achieved by using this surface... normal vector and Solving vectors by cross product If the process requires the machined end to be at a different angle to the surface being machined, it can also be based on... Or the normal vector at the current key feature point When performing calculations, it is important to determine the vectors. The direction.

[0086] The final processing end vector calculated through the above steps represents the target posture corresponding to that path point during actual processing, and can be used to guide the robot in posture adjustment. Since this embodiment uses a KUKA robot as the experimental basis, and the KUKA robot's communication commands require defining the end-effector posture in Euler angles, the vector is converted to Euler angles. The vector-to-Euler angle conversion involves converting the direction vector into a rotation matrix, and then converting the rotation matrix into Euler angles.

[0087] For the same vector Let on the unit orthogonal basis The coordinates below are [x, y, z]. T After one rotation, the unit orthogonal basis It became a vector The coordinates under the new orthogonal basis are [x′, y′, z′]. TThis combination immediately yields the rotation matrix R; let the Euler angles of the three axes x, y, z be θ. x θ y θ z The sine and cosine values ​​are c and c, respectively. x s x c y s y c z s z The algorithm is as follows:

[0088]

[0089] θ x =atan2(r 32 r 33 )

[0090]

[0091] θ z =atan2(r 21 ,r 11 )

[0092] In summary, the machining poses at each data acquisition point of the complex curved workpiece can be obtained, including the positions of path points in space and the corresponding poses (Euler angles) that meet the process requirements. To meet the robot's command requirements, it is necessary to perform overall fitting of the path points and their corresponding poses in three-dimensional space, using the least squares method or B-spline curve fitting method. Sampling is then performed at equal intervals according to machining requirements to optimize the machining path. The above methods will obtain the high-order fitting curves corresponding to this set of feature points. By quantitatively truncating the path, it can be proportionally divided to achieve further segmentation of the path points, providing precise machining path guidance for the system.

[0093] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to obtain equivalent embodiments without departing from the scope of the technical solution of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the technical solution of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance, characterized in that, The method includes: S1. Use a line laser profile sensor to acquire multiple point cloud data of the area to be processed on the curved workpiece along the travel direction; S2. Preprocess each point cloud data obtained and extract two or more key feature points. The extracted key feature points are required to describe the spatial structural features of the area to be processed on the curved workpiece. S3. By using a transformation matrix, the coordinate values ​​of each key feature point in the visual coordinate system are transformed into coordinate values ​​in the base coordinate system, thus obtaining the set of key feature points in the base coordinate system; S4. Using the least squares method or B-spline curve fitting method, fit the point clouds of key feature points with the same relative positions in multiple point cloud data in the base coordinate system to obtain multiple corresponding spatial curve functions, including: (1) These are three adjacent key feature points in the base coordinate system, connected to the current key feature point. Construct three vectors with the preceding and subsequent key feature points. , and And based on these three vectors and the current key feature points The tangent vector is calculated using the following formula. and normal vector : ； ； ； (2) Based on the tangent vector and the current key feature points Construct the angle bisector ; or through the current point and vector , Establish planes respectively , the angle bisector Solving for the problem using the point normal equation, and by using key feature points shared by all faces. and plane The mean vector is used as the angle bisector. The normal vector; Due to the requirements of the machining process, the machining end is used for path planning in an orientation that is parallel to the surface to be machined, perpendicular to the surface to be machined, or offset by a specified angle in a certain direction at a parallel or perpendicular angle. The TCP always falls on the angle bisector plane. Inside, since the normal vector also lies on the angle bisector. Therefore, after offsetting the normal vector by a specified distance according to the processing requirements, the attitude vector also lies on the angle bisector. Inside, and according to processing requirements, it forms a specified angle with the normal vector; The method for machining curved workpieces also includes: converting the calculated vector into a rotation matrix, and then converting the rotation matrix into Euler angles; For the same vector Let the unit orthogonal basis be The coordinates below are After one rotation, the unit orthogonal basis It became a vector The coordinates under the new orthogonal basis are This combination immediately yields the rotation matrix. Let the Euler angles of the three axes x, y, z be respectively... The sine and cosine values ​​are respectively The algorithm is as follows: ； ； The machining method for curved workpieces also includes: using the least squares method or B-spline curve fitting method to fit the calculated path points and their corresponding postures in three-dimensional space as a whole, and performing equal-interval sampling and machining path optimization according to machining requirements to meet the robot's instruction requirements; S5. Based on the space curve function and process parameters, re-select points at equal intervals; S6. Calculate the tangent vector and normal vector of the current point by combining the preceding and following points; S7. If a single point cloud data contains two or more key feature points, determine the normal vector or angle bisector of one of the key feature points, calculate the intersection of the spatial curve that does not contain the key feature point with the normal vector or angle bisector, and generate a path based on the processing position and attitude requirements.

2. The method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance according to claim 1, characterized in that, When performing online machining using a curved workpiece machining method, the line laser contour sensor is installed in front of the machining end head in the direction of travel so that the path points corresponding to the machining position are calculated before the machining end head travels to a certain machining position.

3. The method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance according to claim 2, characterized in that, The processing end is a glue applicator.

4. The method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance according to claim 1, characterized in that, The key feature points in the key feature point set under the base coordinate system mentioned in step S3 exist in the order of data acquisition.

5. The method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance according to claim 1, characterized in that, If plane Coincident, the angle bisector That is, the coincident plane.

6. The method for generating a three-dimensional conformal corner path for robot adhesive application based on online line laser guidance according to claim 1, characterized in that, Step S4 describes using the least squares method or B-spline curve fitting method to fit key feature points with the same relative arrangement in multiple point cloud data in the base coordinate system, obtaining multiple corresponding spatial curve functions, including: The least squares method is used to fit the space curve, and the final space curve parameters are obtained by solving the overdetermined equations. The curve polynomial of order 1 is shown in the following equation: ； Where y is the y-axis coordinate of the higher-order curve, x is the x-axis coordinate of the independent variable, i is the specific degree of different degree terms, and k is the parameter of each degree term; Further, the above equation can be constructed into a system of equations: ； in, Composed of parameter points 3D matrix Composed of parameter points 3D matrix This indicates the number of coordinate points involved in the fitting process. For those in demand dimensional coefficient matrix; By applying the least-squares solution theorem for overdetermined equations, we can obtain: ； Among them, X T X is the transpose of the [M × 6] dimensional matrix X composed of parameter points; The coefficient matrix is ​​then obtained through this matrix operation. The above method can be used to obtain a set of spatial fitting curves corresponding to key feature points.

Images