Adaptation method for frame type assembly component

Abstract

The invention relates to the field of automatic assembly of production lines in the automobile manufacturing industry, and discloses a frame type assembly component-oriented adaptation method, which comprises the following steps of: detecting and recording a space pose of a component to be assembled and a space pose of a frame type component, inputting data into an adaptation mathematical model, and calculating a space pose transformation relation between the data and the space pose transformation relation; and an adaptive pose offset vector is calculated, and the robot completes installation according to the adaptive pose offset vector. Compared with the prior art, the laser vision measurement method based on the laser technology and the traditional laser vision detection technology can obtain high-resolution contour point feature information, and the shape and the features of the surface of the frame type component can be accurately measured; the measured point cloud data can provide accurate point feature information, the formed detection algorithm has a good adaptation effect on the frame type component, the algorithm calculation amount is small, the method is suitable for online real-time assembly, and the method has good adaptability to the contour machining error of the frame type component.

Description

Technical Field

[0001] The present invention relates to the technical field of automated assembly in the automotive manufacturing production line, and specifically relates to an adaptation method for frame-type assembly components. Background Art

[0002] Intelligent vision measurement and adaptation technology can improve industrial production efficiency and product quality, and is an important part of future intelligent manufacturing.

[0003] The positioning and assembly of frame-type components is a key link in the automated assembly in the field of automotive manufacturing. For the automated assembly of car hoods, doors, and windshields, most of them use cameras to position the vehicle body, such as the method based on calibration plates, coordinate measurement methods, methods based on SNK cameras, etc.; among them, the calibration plates have high accuracy requirements, cannot be damaged during use and occupy storage space, and high repeatability is required to position the calibration plates in the camera's field of view, with high costs; the coordinate measurement method requires manual operation of the measuring arm to adjust the position of the camera, and at the same time, it is also necessary to accurately aim at the object to be measured through a rotation and tilt device. This process is relatively cumbersome and time-consuming, and will seriously reduce production efficiency; when measuring based on multiple SNK cameras, it is not easy to achieve precise calibration, resulting in measurement errors and affecting the positioning accuracy of workpieces. Summary of the Invention

[0004] Therefore, the present invention provides an adaptation method for frame-type components to solve problems such as low production efficiency and difficulty in achieving precise calibration.

[0005] To achieve the above object, the present invention provides the following technical solution: An adaptation method for frame-type components, and the steps of the method are as follows:

[0006] S1: Use a robot to grasp the component to be assembled, detect the local point feature information of the component to be assembled through a laser vision sensor to determine the spatial pose of the component to be assembled, and record the spatial pose of the component to be assembled;

[0007] S2: Detect the local point feature information of the frame-type component through the laser vision sensor to determine the spatial pose of the frame-type component, and record the spatial pose of the frame-type component; wherein, the laser vision sensor can form a point cloud image;

[0008] S3: Construct an adaptation mathematical model between the component to be assembled and the frame-type component, input the local point feature information of the component to be assembled and the spatial pose of the component to be assembled into the adaptation mathematical model, and input the local point feature information of the frame-type component and the spatial pose of the frame-type component into the adaptation mathematical model;

[0009] S4: Calculate the adaptation pose offset vector between the component to be assembled and the frame-type component through the adaptation mathematical model;

[0010] S5: The robot installs the to-be-assembled component on the frame-type component according to the adapted pose offset vector.

[0011] As a preferred technical solution of the present invention, the relationship between the local point feature information of the to-be-assembled component and the spatial pose of the to-be-assembled component is as follows:

[0012] T C = Ψ(P1, P2,..., P n )

[0013] where P i (i = 1, 2,..., n) is the local point feature information of the to-be-assembled component extracted by the laser vision sensor in the robot coordinate system, Ψ is the spatial pose function of the to-be-assembled component, and T C is the homogeneous transformation matrix of the spatial pose of the to-be-assembled component relative to the robot coordinate system.

[0014] As a preferred technical solution of the present invention, the relationship between the local point feature information of the frame-type component and the spatial pose of the frame-type component is as follows:

[0015]

[0016] where, is the local point feature information of the frame-type component in the robot coordinate system, Ω is the spatial pose function of the frame-type component, and T F is the homogeneous transformation matrix of the spatial pose of the frame-type component relative to the robot coordinate system.

[0017] As a preferred technical solution of the present invention, the application function of the adapted mathematical model is:

[0018]

[0019] where n is the number of local point features of the component contour, is the adapted local point feature information of the to-be-assembled component in the robot coordinate system, U(x) is the pose adaptation transformation matrix between the to-be-assembled component and the frame-type component; z is the pose adaptation offset vector, including spatial position and attitude information; ||·|| E represents the Euclidean norm; Γ is the adaptation function, including distance optimization adaptation of local point features and attitude optimization adaptation of local point features; is the distance adaptation matrix of the i-th local point, and d i is the adaptation distance of the i-th local point.

[0020] As a preferred technical solution of the present invention,

[0021] S401: Set the distance adaptation matrix for the i-th local point Set the adaptation distance d for the i-th local point i , Set the initial offset vector z for pose adaptation 0；

[0022] S402: According to the described adaptation mathematical model, set the adaptable target value ε of the adaptation function Γ(z), and use the non-linear least squares algorithm to iteratively solve the pose adaptation offset vector z between the component to be assembled and the frame component. The solution formula is:

[0023]

[0024] Where J is the Jacobian matrix of the mathematical model, I is the identity matrix, μ is the damping parameter, h is the iteration step vector, r(z) is the distance optimization adaptation and pose optimization adaptation vector of the features of each local point, and K is the number of iteration steps;

[0025] S403: Calculate the pose adaptation offset vector z that meets the adaptable target value ε in S402 w .

[0026] As a preferred technical solution of the present invention, the pose adaptation offset vector z includes an adaptation position and an adaptation pose; its description form can be based on any one of the homogeneous transformation matrix, Euler angles, RPY angles, and quaternions.

[0027] As a preferred technical solution of the present invention, the selected value range of the number of local point feature information detected for the component to be assembled is 6 - 10.

[0028] As a preferred technical solution of the present invention, the selected value range of the number of local point feature information detected for the frame component is 6 - 10.

[0029] As a preferred technical solution of the present invention, the robot adopts one or several of industrial standard joint robots, collaborative robots, and humanoid robots.

[0030] Has the following beneficial effects:

[0031] Compared with the prior art, the laser vision measurement method based on laser technology and traditional laser vision detection technology can obtain high-resolution contour point feature information, and can accurately measure the shape and features of the surface of the frame component; the measured point cloud data can provide accurate point feature information, and the formed detection algorithm has a good adaptation effect on the frame component, with a small amount of algorithm calculation, is suitable for online real-time assembly, and has good self-adaptability to the contour processing error of the frame component. Description of the Drawings

[0032] Figure 1 It is a schematic flowchart of an embodiment of the present invention;

[0033] Figure 2 It is a schematic diagram of the sequence of each local point feature of the robot acquisition component in the embodiment of the present invention;

[0034] Figure 3 It is a schematic diagram of a mathematical model of the adaptation method between the component to be assembled and the frame-type component based on local point feature information in the embodiment of the present invention. Detailed implementation manners

[0035] Next, the technical solutions of the present invention will be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0036] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be understood as a limitation of the present invention. In addition, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0037] In the description of the present invention, it should be noted that unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0038] An adaptation method for frame-type assembly components, the method steps include:

[0039] S1: Use a robot to grasp the component to be assembled, detect the local point feature information of the component to be assembled through a laser vision sensor to determine the spatial pose of the component to be assembled, and record the spatial pose of the component to be assembled;

[0040] S2: Detect the local point feature information of the frame-type component through a laser vision sensor to determine the spatial pose of the frame-type component, and record the spatial pose of the frame-type component; wherein, the laser vision sensor can form a point cloud image;

[0041] S3: Construct an adaptation mathematical model between the component to be assembled and the frame - type component. Input the local point feature information and the spatial pose of the component to be assembled into the adaptation mathematical model, and input the local point feature information and the spatial pose of the frame - type component into the adaptation mathematical model;

[0042] S4: Calculate the adaptation pose offset vector between the component to be assembled and the frame - type component through the adaptation mathematical model;

[0043] S5: The robot installs the component to be assembled on the frame - type component according to the adaptation pose offset vector.

[0044] Specifically, the robot in step S1 can be an industrial standard joint robot, a collaborative robot, or a humanoid robot. An industrial standard joint robot usually consists of multiple joints, and each joint can achieve motion at a specific angle or direction. Through the coordinated cooperation of the joints, it can complete complex tasks. Its advantages are high position accuracy and repeatability, and it can operate stably under long - term and high - intensity working conditions after strict quality control and testing. It has mature programming languages and control systems for realizing automated production; A collaborative robot can safely complete tasks with construction workers in the working environment. Its advantages are lightweight, equipped with various sensors to achieve safe and efficient human - machine collaboration, and simple programming; A humanoid robot has a body structure, joint configuration, and motion ability similar to humans, and can walk, operate objects, and interact with people like humans in various environments. Its advantages are high - performance motion control ability, high - perception interaction ability, and adaptability.

[0045] Specifically, the laser vision sensor can form a point - cloud image. By calculating the time when the reflected laser returns, the distance between each point on the surface of the target object and the laser vision sensor can be calculated, and then the three - dimensional coordinate information of each point on the surface of the target object can be obtained. The point - cloud image can be used to accurately measure parameters such as the size and shape of the object, improving production quality and efficiency; By extracting the local point feature information of the component to be assembled, the spatial pose of the component to be assembled is determined; By extracting the local point feature information of the frame - type component, the spatial pose of the frame - type component is determined.

[0046] Specifically, the following relationship exists between the local point feature information of the component to be assembled and the spatial pose of the component to be assembled:

[0047] T C = Ψ(P1, P2,..., P n )

[0048] where, P i(i = 1, 2, ..., n) is the local point feature information of the component to be assembled extracted by the laser vision sensor in the robot coordinate system, Ψ is the spatial pose function of the component to be assembled, and T C is the homogeneous transformation matrix of the spatial pose of the component to be assembled relative to the robot coordinate system.

[0049] Specifically, the local point feature information of the frame-type component and the spatial pose of the frame-type component have the following relationship:

[0050]

[0051] Among them, is the local point feature information of the frame-type component in the robot coordinate system, Ω is the spatial pose function of the frame-type component, and T F is the homogeneous transformation matrix of the spatial pose of the frame-type component relative to the robot coordinate system.

[0052] Specifically, the application function of the adaptation mathematical model is:

[0053]

[0054] Among them, n is the number of local point features of the component contour, is the adapted local point feature information of the component to be assembled in the robot coordinate system, U(x) is the pose adaptation transformation matrix between the component to be assembled and the frame-type component; z is the pose adaptation offset vector, including spatial position and attitude information; ||·|| E represents the Euclidean norm; Γ is the adaptation function, including the distance optimization adaptation of local point features and the attitude optimization adaptation of local point features; is the distance adaptation matrix of the i-th local point, and d i is the adaptation distance of the i-th local point.

[0055] The adaptation mathematical model is described by the non-linear least squares function Γ, and has and includes two parts. Among them,

[0056] The first part is the distance optimal adaptation evaluation function Γ1 of local point features. The function Γ1 calculates the distance between the local point feature of the i-th frame-type component and the adapted local point feature of the component to be assembled, and then calculates the absolute value of the difference between the distance between the local point feature of the i-th frame-type component and the adapted local point feature of the component to be assembled and the adaptation distance of the i-th local point;

[0057] The second part is the attitude optimal adaptation evaluation function Γ2 of local point features. The function Γ2 calculates the deviation between two poses. The first pose uses the homogeneous transformation matrix T of the spatial pose of the frame-type component relative to the robot coordinate system F; The second pose is represented by the ideal assembly pose of the component to be assembled, and the second pose is equal to the product of the pose adaptation transformation matrix U(x) between the component to be assembled and the frame component and T C between them.

[0058] Through the application formula of the adaptation mathematical model, the adaptation relationship between the component to be assembled and the frame component is clarified, and technical features such as local point feature information, pose adaptation transformation matrix, pose adaptation offset vector, Euclidean norm, adaptation function, distance adaptation matrix, and adaptation distance are described. These technical features cooperate with each other to accurately calculate the pose offset between the component to be assembled and the frame component, so as to achieve high-precision assembly. Each parameter and variable in the formula work together to ensure the accuracy and efficiency of the adaptation process, which helps to improve the assembly accuracy and efficiency.

[0059] Specifically, the pose adaptation offset vector z includes an adaptation position and an adaptation attitude, and its description form can be based on any one of homogeneous transformation matrix, Euler angle, RPY angle, and quaternion. Among them, the homogeneous transformation matrix realizes the translation operation by adding the fourth dimension, so that rotation and translation can be represented uniformly, which is convenient for calculating coordinate transformation and composite transformation. In robot applications, it is used to describe the position and attitude of the robot end effector in three-dimensional space and the transformation relationship between different coordinate systems.

[0060] Specifically, the specific implementation steps of calculating the pose adaptation offset vector between the component to be assembled and the frame component through the adaptation mathematical model in step S4 are as follows:

[0061] S401: Set the distance adaptation matrix of the i-th local point Set the adaptation distance d of the i-th local point i , and set the initial pose adaptation offset vector z0;

[0062] S402: According to the adaptation mathematical model, set the adaptable target value ε of the adaptation function Γ(z), and use the nonlinear least squares algorithm to iteratively solve the pose adaptation offset vector z between the component to be assembled and the frame component. The solution formula is:

[0063]

[0064] Among them, J is the Jacobian matrix of the mathematical model, I is the identity matrix, μ is the damping parameter, h is the iteration step vector, r(z) is the distance optimization adaptation and attitude optimization adaptation vector of each local point feature, and K is the number of iteration steps;

[0065] S403: Calculate the pose adaptation offset vector z that satisfies the adaptable target value ε in S402 w .

[0066] Specifically, the number of local point feature information of the component to be assembled is set to n, and the value range of n is 6 - 10; the number of local point feature information of the frame - type component is also set to n, and the value range of n is 6 - 10.

[0067] Example 1:

[0068] In this example, a laser vision sensor is fixed on the tooling fixture of the robot for measurement.

[0069] When n = 7, 7 position measurement points are used to extract local point feature information. The measurement sequence of the robot is as Figure 2 shown. According to the planned trajectory of the robot, the robot is operated to move to the specified position to collect local point feature information, and the pre - processing algorithm is used to filter the image point feature information, and the current pose is recorded.

[0070] S1. Teach the robot to grasp the component to be assembled, and use the laser vision sensor to detect the local point feature information of the component to be assembled. When n = 7, then P i (i = 1, 2,..., n) is the local point feature information of the component to be assembled extracted by the laser vision sensor in the robot coordinate system, and record the spatial pose relationship of the component to be assembled:

[0071] T C = Ψ(P1, P2,..., P n ) Equation (1)

[0072] S2. Use the laser vision sensor to detect the local point feature information of the frame - type component. When n = 7, then is the local point feature information of the frame - type component extracted by the laser vision sensor in the robot coordinate system, and record the spatial pose relationship of the frame - type component:

[0073]

[0074] As Figure 3 shown is a schematic diagram of the mathematical model of the adaptation method between the component to be assembled and the frame - type component based on local point feature information (only the relationship between two measurement points and the component spatial pose is drawn in the figure, and other measurement points have the same spatial pose relationship with the component);

[0075] S3. Build the adaptation mathematical model between the component to be assembled and the frame - type component (i = 1, 2,..., 7) as:

[0076]

[0077] Substitute the local point feature information P i of the component to be assembled and the spatial pose T C, local point feature information of the frame component and the spatial pose T F ;

[0078] S4. Solve the adaptation pose offset vector between the component to be assembled and the frame component;

[0079] The specific implementation process of this step S4 is as follows:

[0080] S401. Set the distance adaptation matrix of the i-th local point Set the adaptation distance d of the i-th local point i , and set the initial pose adaptation offset vector z0;

[0081] S402. According to the adaptation mathematical model in step S3 described by formula (3), set the adaptable target value ε of the adaptation function Γ(z), and use the nonlinear least squares algorithm to iteratively solve the pose adaptation offset vector z between the component to be assembled and the frame component. The solution formula is:

[0082]

[0083] In formula (4), J is the Jacobian matrix of the mathematical model, μ is the damping parameter, h is the iteration step vector, r(z) is the distance optimization adaptation and pose optimization adaptation vector of each local point feature, and K is the number of iteration steps.

[0084] S403. Obtain the pose adaptation offset vector z that satisfies the adaptable target value ε in step S402 w .

[0085] S5. The robot automatically installs the component to be assembled on the frame component according to the pose offset vector.

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptation method for frame-type assembly components, characterized in that The steps of the method include: S1: Use a robot to grasp the component to be assembled, detect the local point feature information of the component to be assembled through a laser vision sensor to determine the spatial pose of the component to be assembled, and record the spatial pose of the component to be assembled; S2: Detect the local point feature information of the frame-type component through the laser vision sensor to determine the spatial pose of the frame-type component, and record the spatial pose of the frame-type component; wherein, the laser vision sensor can form a point cloud image; S3: Construct an adaptation mathematical model between the component to be assembled and the frame-type component, input the local point feature information of the component to be assembled and the spatial pose of the component to be assembled into the adaptation mathematical model, and input the local point feature information of the frame-type component and the spatial pose of the frame-type component into the adaptation mathematical model; S4: Calculate the adaptation pose offset vector between the component to be assembled and the frame-type component through the adaptation mathematical model; S5: The robot installs the component to be assembled on the frame-type component according to the adaptation pose offset vector.

2. An adaptation method for frame-type assembly components according to claim 1, characterized in that: The functional relationship between the local point feature information of the component to be assembled and the spatial pose of the component to be assembled is as follows: T C = Ψ(P1, P2,..., P n ) Among them, P i (i = 1, 2, ..., n) is the local point feature information of the component to be assembled extracted by the laser vision sensor in the robot coordinate system, Ψ is the spatial pose function of the component to be assembled, T C is the homogeneous transformation matrix of the spatial pose of the component to be assembled relative to the robot coordinate system.

3. An adaptation method for frame-type assembly components according to claim 1, characterized in that: The relationship between the local point feature information of the frame-type component and the spatial pose of the frame-type component is as follows: Among them, is the local point feature information of the frame-type component in the robot coordinate system, Ω is the spatial pose function of the frame-type component, and T F is the homogeneous transformation matrix of the spatial pose of the frame-type component relative to the robot coordinate system.

4. The adaptation method for a frame - type assembly component according to claim 1, wherein: The application function of the adaptation mathematical model is: where n is the number of local point features of the component contour, is the local point feature information after adaptation of the component to be assembled in the robot coordinate system, U(x) is the pose adaptation transformation matrix between the component to be assembled and the frame component; z is the pose adaptation offset vector, including spatial position and attitude information; ||·|| E represents the Euclidean norm; Γ is the adaptation function, including distance optimization adaptation of local point features and attitude optimization adaptation of local point features; is the distance adaptation matrix of the i-th local point, d i is the adaptation distance of the i-th local point.

5. An adaptation method for frame-type assembly components according to claim 4, characterized in that: The adaptation function Γ has and includes two parts, wherein, The first part is the distance optimal adaptation evaluation function Γ1 of the local point features. The function Γ1 calculates the distance between the local point features of the i-th frame-type component and the local point features after adaptation with the component to be assembled, and then calculates the absolute value of the difference between the distance between the local point features of the i-th frame-type component and the local point features after adaptation with the component to be assembled and the adaptation distance of the i-th local point; The second part is the pose optimal adaptation evaluation function Γ2 of local point features. The function Γ2 calculates the deviation between two poses. The first pose is represented by the homogeneous transformation matrix T of the spatial pose of the frame-type component relative to the robot coordinate system F ; The second pose is represented by the ideal assembly pose of the component to be assembled. The second pose is equal to the product between the pose adaptation transformation matrix U(x) between the component to be assembled and the frame-type component and T C .

6. A fitting method for a frame-type assembly member according to claim 4, characterized in that, The step of calculating the adaptation pose offset vector between the component to be assembled and the frame-type component through the adaptation mathematical model includes: S401: Set the distance adaptation matrix of the i-th local point Set the adaptation distance d of the i-th local point i , and set the initial offset vector z0 of pose adaptation; S402: According to the adaptation mathematical model, set the adaptable target value ε of the adaptation function Γ(z), and use the nonlinear least squares algorithm to iteratively solve the pose adaptation offset vector z between the component to be assembled and the frame-type component. The solution formula is: where J is the Jacobian matrix of the mathematical model, I is the identity matrix, μ is the damping parameter, h is the iteration step vector, r(z) is the distance optimization adaptation and pose optimization adaptation vector of each local point feature, and K is the number of iteration steps; S403: Calculate the pose adaptation offset vector z that meets the adaptable target value ε in S402 w 。 7. An adaptation method for frame-type assembly components according to claim 4, characterized in that: The pose adaptation offset vector z includes an adaptation position and an adaptation pose; its description form can be based on any one of homogeneous transformation matrices, Euler angles, RPY angles, and quaternions.

8. A fitting method for frame-type assembly components according to claim 1, characterized in that: The selected value range of the number of local point feature information detected for the component to be assembled is 6-10.

9. A fitting method for frame-type assembly components according to claim 1, characterized in that: The selected value range of the number of local point feature information detected for the frame-type component is 6-10.

10. A fitting method for frame-type assembly components according to claim 1, characterized in that: The robot adopts one or several of industrial standard joint robots, collaborative robots, and humanoid robots.

Images