Robot welding circular path method and apparatus

Abstract

The application discloses a robot welding circular ring track method and device, and relates to the field of computer vision detection. A point cloud image of a circular ring mark is acquired through camera module shooting, and a fitting point cloud set is obtained through noise point filtering. Three target point clouds are determined from the fitting point cloud set, a target spherical surface equation and a target section are fitted and constructed according to a space coordinate system, a target circular ring equation is determined based on the fitted target spherical surface and the target section, and an inserted point cloud and corresponding space coordinates are calculated and generated according to the selected target point cloud on the circular ring. The posture data of the robot mechanical arm end is determined based on the inserted point cloud and the target point cloud, and the motion track of the welding point is generated. The scheme filters the point cloud image containing noise points, fits the closest circular ring equation, and determines the posture data of the robot mechanical arm end using the least number of point clouds, thereby improving the circular ring welding efficiency while meeting the welding precision requirement.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of computer vision detection, in particular to a robot welding circular ring track method and device. BACKGROUND

[0002] In the field of finishing technology, when a robot is welding, a camera module needs to shoot a workpiece, identify or mark according to the image content obtained, and realize spatial positioning to ensure the precision of the robot welding.

[0003] For conventional image recognition, a robot arm can accurately weld according to the marks on the workpiece, but for some finishing fields, the point cloud captured by the camera is incomplete, resulting in many noise points. If the welding track of the robot is directly processed based on the point cloud, the welding track will be incomplete. Especially for circular ring welding points, under the condition of incomplete or offset point cloud track, a complete circular ring track cannot be fitted by computer vision, so the welding track must be incomplete. SUMMARY

[0004] The robot welding circular ring track method and device provided by the embodiments of the present application solve the problems of incomplete and position offset of the robot welding track.

[0005] In one aspect, a robot welding circular ring track method is provided, including the following steps:

[0006] A point cloud image of a circular ring mark is captured by a camera module, and a fitting point cloud set is obtained through noise point filtering;

[0007] Three target point clouds are determined from the fitting point cloud set, and a target spherical surface equation and a target tangent surface are fitted and constructed according to a space coordinate system established; the target spherical surface constructed by the target spherical surface equation is covered with the most fitting point clouds except the target point clouds, and the target tangent surface is a robot welding surface;

[0008] A target circular ring equation is determined based on the fitted target spherical surface and the target tangent surface, and an inserted point cloud and its corresponding space coordinates are calculated and generated according to a target point cloud selected on the circular ring; the inserted point cloud is located between adjacent target point clouds and within the target circular ring;

[0009] The pose data of the end of a robot arm is determined based on the inserted point cloud and the target point cloud, and a motion track of a welding point is generated.

[0010] Specifically, the fitting point cloud set is obtained through noise point filtering, including:

[0011] All point clouds in the point cloud image are extracted to obtain a point cloud set; the point cloud set contains noise point data;

[0012] Traverse all point clouds, respectively, to count the average neighbor distance between each point cloud and K neighbor point clouds ;

[0013] Calculate the average neighbor distance of all point clouds The mean distance And the standard deviation Determine the point cloud greater than the distance threshold Noise data, filter to obtain the fitting point cloud set;

[0014] The distance threshold Is expressed as follows:

[0015]

[0016] Where, The proportion coefficient of the point cloud.

[0017] Specifically, the three target point clouds are determined from the fitting point cloud set, and the target spherical surface equation and the target tangent plane are fitted and constructed according to the established spatial coordinate system, comprising:

[0018] Establish a spatial coordinate system, and combine the fitting point clouds in turn to construct a plurality of construction sets and corresponding verification sets, the construction set containing three fitting point clouds for constructing a spherical surface equation, and the verification set containing other fitting point clouds except the construction spherical surface equation;

[0019] Respectively based on the three fitting point cloud coordinates and the tangent plane equation composed of each group to construct a candidate spherical surface equation;

[0020] Substitute the other point clouds in the verification set into the candidate spherical surface equation to verify, and determine the number of point clouds covered by the spherical surface;

[0021] The candidate spherical surface equation with the largest number of point clouds covered by the spherical surface is determined as the target spherical surface equation, and all point clouds contained in the target construction set and the target verification set are determined as the target point clouds on the target annulus. The plane where the target point cloud is located is determined as the target tangent plane.

[0022] Specifically, the target annulus equation is determined based on the fitted target spherical surface and the target tangent plane, and the inserted point cloud and its corresponding fitted inserted point cloud are calculated and generated according to the selected target point cloud on the annulus, comprising:

[0023] Randomly generate M coordinate points on the circle according to the target annulus equation, and sort according to the position relationship;

[0024] According to Euler's formula, the distance between two adjacent coordinate points is calculated And the number of inserted points is calculated according to the following formula :

[0025]

[0026] in, This represents the Euclidean distance between the i-th group of adjacent coordinates. This represents the number of points that need to be inserted between the i-th group of adjacent coordinate points. This indicates the maximum gap in the welding trajectory of the robotic arm;

[0027] Based on adjacent coordinate points and Given a point cloud vector, calculate the nth insertion point on the vector between two points. The calculation formula is as follows:

[0028]

[0029]

[0030] in, This represents the vector between two coordinate points.

[0031] Specifically, determining the posture data of the robot arm's end effector based on the inserted point cloud and the target point cloud, and generating the motion trajectory of the welding point, includes:

[0032] Based on the center of the target annular circle equation and the insertion point cloud The spatial coordinates are used to determine the placement point cloud. The corresponding fitted target point cloud on the annulus , means as follows:

[0033]

[0034] Among them This represents the radius of the target annulus. This indicates the placement point cloud. To the center The distance is such that the fitted insertion point cloud is located above the target ring;

[0035] The target point cloud and the fitted insertion point cloud are identified as circular welding points and added to the welding point set;

[0036] Determine the vertical downward tilt angle of the robot arm's end effector. The attitude data of the circular welding point relative to the end of the robot arm are calculated one by one and converted into motion trajectory.

[0037] Specifically, the determination of the vertical downward tilt angle of the robot arm's end effector... , sequentially calculate the pose data of the circular ring welding point relative to the end of the robot manipulator, including:

[0038] sequentially select a circular ring welding point from the set of welding points and a neighbor welding point , and determine a unit normal vector relative to the target sphere according to the tilt angle and ; the unit normal vector is perpendicular to the robot manipulator;

[0039] with the negative vector direction as the z-axis, the path scanning direction as the y-axis, and the direction of the line connecting the current circular ring welding point and the neighbor welding point as the positive direction of the y-axis, a path pose coordinate system is established; the x-axis is determined based on the right-hand rule;

[0040] determine the robot arm end pose coordinates in the path pose coordinate system, obtain the Euler angles through the Euler transformation equation and send them to the robot to form the motion trajectory of the target welding point.

[0041]

[0042] When :

[0043]

[0044]

[0045]

[0046] Otherwise:

[0047]

[0048]

[0049]

[0050] wherein, represents a spatial matrix constructed in the path pose coordinate system with as the target welding point, x-axis, y-axis, z-axis, and Euler angle parameters.

[0051] On the other hand, the present application provides a robot welding circular ring trajectory device, comprising:

[0052] The point cloud acquisition module is used to capture point cloud images of the circular markers through the camera module, and obtain a fitted point cloud set through noise filtering;

[0053] The sphere determination module is used to determine three target point clouds from the fitted point cloud set, and to fit and construct the target sphere equation and target cross surface according to the established spatial coordinate system; the target sphere constructed by the target sphere equation is covered with the most fitted point clouds other than the target point clouds, and the target cross surface is the robot welding surface;

[0054] The point cloud generation module is used to determine the target ring equation based on the fitted target sphere and the target cross-section, and to generate an insertion point cloud and its corresponding spatial coordinates based on the target point cloud selected on the ring; the insertion point cloud is located between adjacent target point clouds and within the target ring;

[0055] The trajectory generation module is used to determine the posture data of the robot arm end based on the inserted point cloud and the target point cloud, and to generate the motion trajectory of the welding point.

[0056] The beneficial effects of the technical solution provided in this application include at least the following: by filtering a noisy point cloud image to eliminate noise deviating from the ring, a set of fitted point clouds is generated. Then, based on the fitted point cloud, the optimal target point cloud is found, and a target sphere and its corresponding target ring equation that can cover the most fitted point cloud data are constructed. To improve welding accuracy, an insertion point cloud can be generated from the target point cloud selected from the target ring equation. Based on the insertion point cloud and the confirmed fitted point cloud, and while meeting accuracy requirements, the posture data of the robot arm's end effector is determined using the minimum number of point clouds, and the motion trajectory of the welding point is generated, thereby improving the robot's welding efficiency and accuracy. Attached Figure Description

[0057] Figure 1 This is a flowchart of the robot circular welding ring trajectory method provided in the embodiments of this application;

[0058] Figure 2 It involves capturing point cloud images of the circular markers using a camera module;

[0059] Figure 3 This is a flowchart of a robot circular welding ring trajectory method provided in another embodiment of this application;

[0060] Figure 4 It is the acquired posture data of the robot's robotic arm end effector;

[0061] Figure 5 This is a structural block diagram of the robot circular welding loop trajectory method provided in the embodiments of this application. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0063] In this article, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0064] Figure 1 This is a flowchart of the robot circular welding ring trajectory method provided in the embodiments of this application, which includes the following steps:

[0065] Step 101: Capture point cloud images of the circular markers using the camera module, and obtain a fitted point cloud set through noise filtering.

[0066] Point cloud images captured by the camera module, such as Figure 2 As shown in the figure, all the point cloud data form a ring-like structure, and there are many noisy data points around the ring-like structure, meaning that the generated point cloud is unevenly distributed.

[0067] Noise filtering involves retaining the point cloud that approximates a ring (approximately forming a ring) while filtering out point clouds that deviate from the ring shape, either inside or outside the ring. This step improves the accuracy of subsequent ring fitting. The point cloud obtained after noise filtering is the fitted point cloud, and the resulting set is the fitted point cloud set.

[0068] Step 102: Determine three target point clouds from the fitted point cloud set, and construct the target spherical equation and target tangent based on the established spatial coordinate system.

[0069] The circular ring (i.e., the welding trajectory ring) can be viewed in space as the intersection of a spherical surface and a tangent plane. Since at least three coordinate points are needed to form a sphere, three fitted point clouds must be selected from the fitted point cloud set to construct the spherical equation and the tangent plane equation. The intersection of the sphere and the tangent plane is the ring equation. This scheme requires identifying at least three target point clouds. The target spherical equation formed by these three target point clouds will cover the largest number of fitted point clouds in space, meaning that the number of other fitted point clouds (excluding these three target point clouds) falling on the target sphere will be the largest. The corresponding target tangent plane is the robot welding surface, i.e., the surface where the target ring is located.

[0070] Step 103: Determine the target ring equation based on the fitted target sphere and target tangent, and generate the insertion point cloud and its corresponding spatial coordinates based on M coordinate points randomly selected on the ring.

[0071] Once the target sphere and target cross-section are determined, the equation of the target annulus and the target sphere are obtained. The selection of point clouds is to facilitate robot positioning of welding points; higher point cloud density results in higher welding accuracy. When the number of point clouds cannot achieve the required welding accuracy, it is necessary to generate insertion point clouds based on the annulus. Insertion point clouds require selecting at least two coordinate point clouds from the target annulus, generating several insertion point clouds and their corresponding spatial coordinates between these two coordinate point clouds, and ensuring that these insertion point clouds are located within the target annulus. It should be noted that the selection of spatial coordinates can be determined based on actual conditions, and this application does not impose any limitations on this.

[0072] Step 104: Determine the posture data of the robot arm end effector based on the spatial coordinates of the insertion point cloud and the coordinate point cloud, and generate the motion trajectory of the welding point.

[0073] The purpose of calculating the insertion point cloud is to cut a circle, since the welding points must be connected by straight lines. This step constructs a set of welding points from the insertion point cloud and the previously determined coordinate point cloud. Moreover, from the robot's perspective, there is a certain angle between the point cloud data of each welding point and the welding torch, so it is necessary to solve for the posture data of the robot arm's end effector, including relative spatial coordinates and Euler angle coordinates, in order to generate the motion trajectory of the welding point.

[0074] In summary, this application filters noisy point cloud images to eliminate noise points deviating from the circular ring, generating a fitted point cloud set. Then, based on the fitted point cloud, the optimal target point cloud is identified, constructing a target sphere and its corresponding target circular ring equation that covers the most fitted point cloud data. To improve welding accuracy, an insertion point cloud can be generated from the coordinate point cloud selected from the target circular ring equation. Based on the insertion point cloud and the confirmed fitted point cloud, the posture data of the robot arm's end effector is determined, and the motion trajectory of the welding point is generated, thereby improving the robot's welding efficiency and accuracy.

[0075] Figure 3 This is a flowchart of a robot circular welding ring trajectory method according to another embodiment of this application, which includes the following steps:

[0076] Step 301: Extract all point clouds from the point cloud image to obtain a point cloud set; the point cloud set contains noise data.

[0077] Step 302: Traverse all point clouds and calculate the average neighbor distance between each point cloud and its K neighbor point clouds. .

[0078] like Figure 1 As shown in the diagram, for each selected point cloud, the distances between the selected point and its K neighboring point clouds are calculated, and then the average distance is taken. The value of K is determined by the system settings.

[0079] Step 303: Calculate the mean distance and standard deviation of the average neighbor distance of all point clouds, identify point clouds with distances greater than the distance threshold as noisy data, filter them to obtain a set of fitted point clouds.

[0080] For all point clouds, the average neighbor distance is calculated. Then calculate these average neighbor distances. Standard deviation Distance from the mean The distance threshold formula is used to determine whether the data is noisy.

[0081] Distance threshold It is expressed as follows:

[0082]

[0083] in, This represents the scaling factor of the point cloud, which is set according to the actual parameters.

[0084] For average neighbor distance Greater than the distance threshold The points are identified as noise data and filtered out. The remaining point cloud is used as the fitted point cloud and added to the fitted point cloud set.

[0085] Step 304: Establish a spatial coordinate system and combine the fitted point cloud in a polling manner to construct several sets of construction sets and corresponding validation sets.

[0086] A spatial coordinate system is established based on specific circumstances to determine the three-dimensional spatial coordinates of each fitted point cloud. The process of determining the target point cloud, the target sphere equation, and the target cross-section requires a round-robin approach to construct a construction set and a corresponding validation set. Specifically, three fitted point clouds are selected as the construction set using a round-robin method, and the rest serve as the validation set.

[0087] Step 305: Construct candidate spherical equations based on the coordinates of the three fitted point clouds and the tangent equations of each group.

[0088] Three fitted point clouds were selected to construct candidate spherical equations and candidate tangent equations, while the fitted point cloud data in the validation set were used to validate the candidate spherical equation.

[0089] The parametric equations that make up a torus can be decomposed into a sphere and a plane. That is, the point cloud formed by a plane passing through the sphere is a circle in three-dimensional space; solve the system of equations (parametric equations of the circle).

[0090]

[0091] Step 306: Substitute other point clouds from the validation set into the candidate spherical equation for validation to determine the number of point clouds that the sphere can cover.

[0092] For all candidate spherical equations, substitute the fitted point cloud from the corresponding validation set into the candidate spherical equations to verify the number of points in the fitted point cloud that the candidate sphere can cover.

[0093] Step 307: The candidate spherical equation with the largest number of point clouds covered by the spherical surface is determined as the target spherical equation. All point clouds contained in the target construction set and target verification set corresponding to the target spherical equation are determined as the target point clouds on the target ring, and the plane where the target point clouds are located is determined as the target sectional plane.

[0094] All possible scenarios are constructed through a polling method. The target spherical equation and the target tangent are found from all candidate spherical equations. The intersection of the target spherical equation and the target tangent is the target annulus.

[0095] Step 308: Randomly generate M coordinate points on the circle according to the target annular equation, and sort them according to their positional relationships.

[0096] In one possible implementation, 10 coordinate points on the circle can be randomly generated according to the target annular equation, and these 10 points can be sorted in order according to their positional relationship.

[0097] Step 309: Calculate the distance between any two adjacent coordinate points using Euler's formula. The number of insertion points is calculated according to the following formula. .

[0098] The points generated on the circle are discontinuous, while the robot welding is carried out along the straight-line distance between the points, which is to cut the circle. In order to restore the circular trajectory to the greatest extent and minimize the number of point clouds, it is necessary to "fill in the gaps" between two adjacent points, that is, to place and insert point clouds.

[0099] The formula for calculating the number of point clouds is as follows:

[0100]

[0101] in, This represents the Euclidean distance between the i-th group of adjacent coordinates. This represents the number of points that need to be inserted between the i-th group of adjacent coordinate points. This indicates the maximum gap (accuracy requirement) in the welding trajectory of the robotic arm.

[0102] Step 310, based on adjacent coordinate points and Given a point cloud vector, calculate the nth insertion point on the vector between two points. .

[0103] The calculation formula is as follows:

[0104]

[0105]

[0106] in, express and The vector between two coordinate points, , .

[0107] Step 311, Based on the center and insertion point cloud of the target annular equation Spatial coordinates to determine the placement point cloud The corresponding fitted target point cloud on the annulus .

[0108] The inserted point cloud is located within the annulus. When the number of selected point clouds (10) is relatively small, directly welding based on the inserted point clouds would affect the accuracy of the annulus. This solution, after determining the inserted point cloud, uses the center of the annulus and the extension line of the inserted point cloud to determine the fitting target point cloud located on the annulus. The fitting target point cloud is then used as the welding point. This way, the welding accuracy can be met without increasing the number of point clouds.

[0109] Sure Following the spatial coordinates, the coordinates of each fitted target point cloud are then calculated. This is represented as follows:

[0110]

[0111] The parsed representation is as follows:

[0112]

[0113] Among them Indicates the radius of the target annulus. Indicates the placement of point clouds To the center The distance is such that the fitted point cloud is located above the target ring. .

[0114] Step 312: The fitted target point cloud and the inserted point cloud are identified as circular welding points and added to the welding point set.

[0115] It should be noted that the verification set for determining the target spherical equation is not included in this scheme. This is because the density distribution of these point clouds is uneven and their overall number is large, which affects the welding efficiency.

[0116] Step 313: Determine the vertical downward tilt angle of the robot arm's end effector. The attitude data of the circular welding point relative to the end of the robot arm are calculated one by one and converted into motion trajectory.

[0117] The welding arm of a robot is often not vertically downward, and this is related to the placement of the workpiece and the position of the ring. When there is an angle between the end effector of the robot arm and the vertical direction, it is necessary to determine the downward tilt angle of the end effector. The orientation data of the circular welding point relative to the end effector of the robot arm are calculated successively. Specifically, the following steps are included:

[0118] A. Select the annular welding points from the set of welding points one by one. and neighboring welding points And determine the unit normal vector relative to the target sphere based on the tilt angle. and Unit normal vector Perpendicular to the robotic arm.

[0119] The welding process requires the robotic arm's end to tilt downwards. Therefore, the robot's posture needs to be adjusted, which can be translated into tilting each point upwards. The degree, and each point is along the tangent direction perpendicular to that point outward from the center of the circle, and the unit normal vector of each point. It can be represented as follows:

[0120]

[0121]

[0122]

[0123] For two adjacent welding points, let the current point be... The next point is Let the combination of its point and normal vector be expressed as:

[0124]

[0125]

[0126] B, with The negative vector direction is the z-axis, the path scanning direction is the y-axis, and the direction of the line connecting the current ring welding point and the neighboring welding point is the positive y-axis direction, thus establishing a path attitude coordinate system; the x-axis is determined based on the right-hand rule.

[0127] Each welding point needs to have its own path attitude coordinate system established. The current welding point and normal vector can be represented as follows: and ;

[0128] To establish the path point attitude coordinate system, starting from position M1, the normal vector... The negative vector is the Z-axis; the scanning direction along the path points is the Y-axis, that is, the direction of the line connecting the current path point and the next path point is the positive direction of the y-axis; the direction of the X-axis is determined according to the right-hand rule, as follows:

[0129]

[0130]

[0131]

[0132] C. Determine the robot arm end-effector attitude coordinates in the path attitude coordinate system, obtain the Euler angles through the Euler transformation equation and send them to the robot to form the motion trajectory of the target welding point.

[0133] The Euler angle transformation equation is expressed as follows:

[0134]

[0135] when hour:

[0136]

[0137]

[0138]

[0139] otherwise:

[0140]

[0141]

[0142]

[0143] The obtained Euler angles OAT are At this point, we have obtained the robot's coordinate and pose information, which can be transmitted to the robot to complete its trajectory movement. The specific generated robot coordinate and pose information is as follows: Figure 4 As shown, a portion of the coordinates is extracted in the form of three-dimensional coordinates plus Euler angle coordinates. `atan2` is a function that, in C, returns the azimuth angle. The C function prototype for `atan2` is `double atan2(double y, double x)`, which returns the arctangent of y / x in radians.

[0144] In summary, this scheme filters noisy point cloud images to eliminate noise points deviating from the ring, generating a fitted point cloud set. Then, based on the fitted point cloud, the optimal target point cloud is identified, constructing a target sphere and its corresponding target ring equation that covers the most fitted point cloud data. To improve welding accuracy without increasing the number of point clouds, an insertion point cloud is generated from the coordinate point cloud selected from the target ring equation. The fitted target point cloud located on the ring is determined on the extension line connecting the insertion point cloud and the center of the ring. The fitted target point cloud and the insertion point cloud are then identified as welding points and added to the weld point set. For cases where the robot arm's end effector has a certain tilt angle, this needs to be converted into the tilt angle of the welding point. Based on the coordinates of the welding arm, the welding point is converted into coordinates plus Euler angle parameters to improve positioning accuracy. This scheme, by fitting the ring and generating the welding trajectory, can restore the original point cloud's ring trajectory to the greatest extent possible, and completes ring welding with the minimum number of point clouds while meeting the robot's welding accuracy requirements, significantly improving robot welding efficiency.

[0145] Figure 5 This is a structural block diagram of the robotic circular welding track device provided in an embodiment of this application. It includes:

[0146] The point cloud acquisition module 501 is used to capture point cloud images of the circular markers through the camera module and obtain a fitted point cloud set through noise filtering.

[0147] The sphere determination module 502 is used to determine three target point clouds from the fitted point cloud set, and to fit and construct a target sphere equation and a target cross surface according to the established spatial coordinate system; the target sphere constructed by the target sphere equation is covered with the most fitted point clouds other than the target point clouds, and the target cross surface is the robot welding surface;

[0148] The point cloud generation module 503 is used to determine the target ring equation based on the fitted target sphere and the target cross-section, and to generate an insertion point cloud and its corresponding spatial coordinates based on M coordinate points randomly selected on the ring; the insertion point cloud is located between adjacent target point clouds and within the target ring;

[0149] The trajectory generation module 504 is used to determine the posture data of the end effector of the robot arm based on the spatial coordinates of the insertion point cloud and the coordinate point cloud, and to generate the motion trajectory of the welding point.

[0150] The preferred embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above. The devices and structures not described in detail should be understood as being implemented in a conventional manner in the art. Any person skilled in the art can make many possible changes and modifications, or equivalent changes to equivalent embodiments without departing from the technical solution of the present invention. This does not affect the substantive content of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for robotic welding of circular tracks, characterized in that, include: Point cloud images of the circular markers are captured by the camera module, and a fitted point cloud set is obtained by noise filtering. Three target point clouds are determined from the set of fitted point clouds, and the target spherical equation and target tangent are constructed by fitting according to the established spatial coordinate system. Specifically, the spatial coordinate system is established, and the fitted point clouds are combined in a polling manner to construct several sets of construction sets and corresponding verification sets. The construction set contains at least three fitted point clouds used to construct the spherical equation, and the verification set contains other fitted point clouds except those used to construct the spherical equation. Candidate spherical equations are constructed based on the coordinates of the three fitted point clouds in each group and the tangent equations they form. The other fitted point clouds in the validation set are substituted into the candidate spherical equation for validation to determine the number of point clouds that the sphere can cover. The candidate spherical equation with the largest number of point clouds covered by the spherical surface is determined as the target spherical equation. All point clouds contained in the corresponding target construction set and target verification set are determined as the target point cloud on the target ring. The plane where the target point cloud is located is the target tangent and the robot welding surface. The target ring equation is determined based on the fitted target sphere and the target tangent, and an insertion point cloud and its corresponding spatial coordinates are generated by computing M coordinate points randomly selected on the ring; the insertion point cloud is located between adjacent coordinate point clouds and is located within the target ring. Based on the insertion point cloud and the coordinate point cloud, the posture data of the robot arm end effector is determined, and the motion trajectory of the welding point is generated.

2. The method according to claim 1, characterized in that, The process of obtaining the fitted point cloud set through noise filtering includes: Extract all points from the point cloud image to obtain a point cloud set; the point cloud set contains noise data; Iterate through all point clouds and calculate the average neighbor distance between each point cloud and its K neighbor point clouds. ; Calculate the average neighbor distance for all point clouds. mean distance and standard deviation It will be greater than the distance threshold. The point cloud is used to identify noisy data, and after filtering, the fitted point cloud set is obtained. Distance threshold It is expressed as follows: in, This represents the scaling factor of the point cloud.

3. The method according to claim 1, characterized in that, The process of determining the target annular equation based on the fitted target sphere and the target tangent, and generating an insertion point cloud and its corresponding fitted insertion point cloud based on M randomly selected coordinate points on the annular surface, includes: M coordinate points on the circle are randomly generated according to the target annular equation, and then sorted according to their positional relationships; Calculate the distance between two adjacent coordinate points using Euler's formula. The number of insertion points is calculated according to the following formula. : in, This represents the Euclidean distance between the i-th group of adjacent coordinates. This represents the number of points that need to be inserted between the i-th group of adjacent coordinate points. This indicates the maximum gap in the welding trajectory of the robotic arm; Based on adjacent coordinate points and Given a point cloud vector, calculate the nth insertion point on the vector between two points. The calculation formula is as follows: in, This represents the vector between two coordinate points.

4. The method according to claim 3, characterized in that, The step of determining the posture data of the robot arm's end effector based on the insertion point cloud and the coordinate point cloud, and generating the motion trajectory of the welding point, includes: Based on the center of the target annular circle equation and the insertion point cloud The spatial coordinates are used to determine the placement point cloud. The corresponding fitted target point cloud on the annulus , means as follows: Among them This represents the radius of the target annulus. This indicates the placement point cloud. To the center The distance is such that the fitted insertion point cloud is located above the target ring; The fitted target point cloud and the inserted point cloud are identified as annular welding points and added to the welding point set; Determine the vertical downward tilt angle of the robot arm's end effector. The attitude data of the circular welding point relative to the end of the robot arm are calculated one by one and converted into motion trajectory.

5. The method according to claim 4, characterized in that, The vertical downward tilt angle of the robot arm's end effector is determined. The attitude data of the circular welding point relative to the end effector of the robot arm are calculated successively, including: Select the annular welding points from the set of welding points one by one. and neighboring welding points And determine the unit normal vector relative to the target sphere based on the tilt angle. and Unit normal vector Perpendicular to the robotic arm; by The negative vector direction is the z-axis, the path scanning direction is the y-axis, and the direction of the line connecting the current ring welding point and the neighboring welding point is the positive y-axis direction, thus establishing a path attitude coordinate system; the x-axis is determined based on the right-hand rule. The robot arm end-effector attitude coordinates in the path attitude coordinate system are determined, and the Euler angles are obtained through the Euler transformation equation and sent to the robot to form the motion trajectory of the target welding point.

6. The method according to claim 5, characterized in that, The Euler angle transformation equation is expressed as follows: when hour: otherwise: in, Indicated by The spatial matrix constructed for the target weld point in the path attitude coordinate system. x-axis The y-axis is The z-axis is and These are Euler angle parameters.

7. A robotic welding ring trajectory device, characterized in that, include: The point cloud acquisition module is used to capture point cloud images of the circular markers through the camera module, and obtain a fitted point cloud set through noise filtering; The sphere determination module is used to determine three target point clouds from the fitted point cloud set, and to fit and construct the target spherical equation and target tangent based on the established spatial coordinate system. Specifically, the spatial coordinate system is established, and the fitted point clouds are combined in a polling manner to construct several sets of construction sets and corresponding verification sets. The construction set contains at least three fitted point clouds used to construct the spherical equation, and the verification set contains other fitted point clouds except those used to construct the spherical equation. Candidate spherical equations are constructed based on the coordinates of the three fitted point clouds in each group and the tangent equations they form. The other fitted point clouds in the validation set are substituted into the candidate spherical equation for validation to determine the number of point clouds that the sphere can cover. The candidate spherical equation with the largest number of point clouds covered by the spherical surface is determined as the target spherical equation. All point clouds contained in the corresponding target construction set and target verification set are determined as the target point cloud on the target ring. The plane where the target point cloud is located is the target tangent and the robot welding surface. The point cloud generation module is used to determine the target ring equation based on the fitted target sphere and the target cross-section, and to generate an insertion point cloud and its corresponding spatial coordinates based on M coordinate points randomly selected on the ring; the insertion point cloud is located between adjacent target point clouds and within the target ring; The trajectory generation module is used to determine the posture data of the robot arm end based on the insertion point cloud and the coordinate point cloud, and to generate the motion trajectory of the welding point.

Images