Mechanical arm compliance control method for die casting mold manufacturing line

Abstract

This application relates to the field of robotic arm control technology, specifically to a compliant control method for a robotic arm used in a die-casting mold manufacturing production line. The method includes: acquiring sensor data for each joint of the robotic arm during its movement on the die-casting mold manufacturing production line; analyzing the compensation coefficient at the same moment using the low-rank essential structure between different sensor data, and using this as the proportional term parameter for the PID controller controlling the robotic arm; and combining the compensation coefficient at the same moment with the distance between the desired and actual positions of the robotic arm joints to achieve compliant control of the robotic arm. This application aims to adjust the PID controller using the specific states of the robotic arm joints to achieve compliant control of the robotic arm.

Description

Technical Field

[0001] This application relates to the field of robotic arm control technology, specifically to a method for compliant control of robotic arms used in die-casting mold manufacturing production lines. Background Technology

[0002] On the die-casting mold manufacturing production line, robotic arms can perform a variety of functions, including automated material handling, precise operation and auxiliary processing, as well as quality inspection and post-processing. The robotic arms are adapted and adjusted according to the mold material and specific process to meet the production line's requirements for processing efficiency and production accuracy.

[0003] During the operation of a robotic arm, due to the complexity of the mold material and structure, as well as the manufacturing requirements such as batch changes, the robotic arm may not only perform simple point-to-point movements, but also engage in complex multi-axis collaborative movements such as high-speed reversal movements. When the robotic arm performs complex movements, compliant control is usually implemented to solve the problems of safety and accuracy during robotic arm interaction. Compliant control enables the robotic arm to adjust the force and trajectory when handling die-casting molds to avoid generating excessive impact force on the mold.

[0004] When controlling a robotic arm, the sensor signals of the robotic arm are usually analyzed first before control is performed. When the robotic arm performs complex movements on the manufacturing production line, there is a large delay error between the real-time processing of the sensor signals and the system response of the robotic arm compliance control. This causes the robotic arm to stagnate or jam during the control process, making it difficult to meet the requirements of compliance control. Summary of the Invention

[0005] To address the aforementioned technical problems, this application provides a method for compliant control of robotic arms in die-casting mold manufacturing lines, thereby resolving the existing issues.

[0006] The compliant control method for robotic arms used in die-casting mold manufacturing production lines in this application adopts the following technical solution:

[0007] One embodiment of this application provides a method for compliant control of a robotic arm in a die-casting mold manufacturing production line, the method comprising the following steps:

[0008] Acquire sensor data for each joint of the robotic arm during its movement on the die-casting mold manufacturing production line;

[0009] A higher-order tensor is constructed using sensor data, its acquisition time, and joint numbers; the higher-order tensor is decomposed into a kernel tensor and the product of multiple factor matrices using Tucker decomposition.

[0010] By using the similarity parameter of the projection coefficients of the same sensor data collected between any two joints at the same time in the corresponding factor matrix, the coupling strength parameter between any two joints at each time is determined; and the coupling strength matrix is ​​constructed using the coupling strength parameters between all joints at the same time.

[0011] The gain factor at each time step is determined iteratively by measuring the difference between the coupling strength matrices at adjacent time steps; the compensation coefficient at each time step is determined based on the size of the off-diagonal elements in the coupling strength matrix at each time step.

[0012] By combining the compensation coefficient at the same moment with the distance between the desired position and the actual position of the robotic arm joint, compliant control of the robotic arm can be achieved.

[0013] Preferably, the sensor data includes torque time-series data, velocity time-series data, and acceleration time-series data.

[0014] Preferably, there are 5 factor matrices, with the first three factor matrices corresponding to the three types of sensor data, respectively.

[0015] Preferably, the method for determining the coupling strength parameter between any two joints at each moment is as follows:

[0016] The coupling strength parameter between the i-th and (i+1)-th joints at time k is expressed as: :

[0017]

[0018] In the formula, M is the number of types of sensor data. , These represent the data collected by the i-th joint and the (i+1)-th joint at time k, respectively, from the m-th sensor in the corresponding factor matrix. The coefficient sequence within the matrix, wherein the coefficient sequence is formed by the data of the m-th sensor in the corresponding factor matrix. It consists of the projection coefficients on all column vectors within the vector. It is a similarity parameter between two coefficient sequences. It is the coupling weight of the m-th sensor data dimension, and its calculation formula is as follows: , It is the number of columns in the factor matrix corresponding to the data dimension of the m-th sensor. It is the number of columns in the factor matrix corresponding to the p-th sensor data dimension.

[0019] Preferably, the elements on the main diagonal of the coupling strength matrix are 0, and the remaining element values ​​are the coupling strength parameters between the two joints in the row and column corresponding to them at the same time.

[0020] Preferably, the method for iteratively determining the gain factor at each time step is as follows:

[0021]

[0022] In the formula, , These are the gain factors at time k and time k-1, respectively. , , These are the coupling strength matrices at time k, k-1, and k-2, respectively. It is a matrix and The results of the difference measurement between them It is a matrix and The results of the difference measurement between them.

[0023] Preferably, the difference measure includes, but is not limited to, Euclidean distance, relative error, and Manhattan distance.

[0024] Preferably, the method for determining the compensation coefficient at each time point is as follows:

[0025]

[0026] In the formula, It is the compensation coefficient at time k. This is the initial preset value. It is the mean of the off-diagonal elements in the coupling strength matrix at time k.

[0027] Preferably, the method for achieving compliant control of the robotic arm is as follows:

[0028] S1, obtain the target path of the robotic arm, including the expected position of each joint at each time step;

[0029] S2, the sum of the distances between the actual positions and the desired positions of all joints at each time step is used as the control error at each time step;

[0030] S3 generates the control input of the PID controller for each time step based on the compensation coefficient and control error at each time step;

[0031] S4 controls the robotic arm based on the control input of the PID controller at each time step.

[0032] Preferably, in step S3, the method for calculating the control quantity of the PID controller at each time point is as follows:

[0033] The control quantity at time k is expressed as... :

[0034]

[0035] In the formula, It is the compensation coefficient at time k, which serves as the proportional term parameter when the PID controller controls the robotic arm. , Let k and t be the control errors, respectively. for The derivative at time k, , , These are the preset integral and derivative terms when the PID controller controls the robotic arm.

[0036] This application has at least the following beneficial effects:

[0037] First, this application effectively captures the coupling relationship between multi-axis joints of the robotic arm when it is in a complex motion state while handling a die-casting mold by decomposing higher-order tensors, thus achieving accurate extraction of the mechanical transmission characteristics between joints. Second, this application adaptively adjusts the gain factor and compensation coefficient at each moment based on the coupling strength between joints and the change amplitude of the coupling strength at each moment when the robotic arm is in multi-axis motion on the manufacturing line. The output of the PID controller is adjusted according to the specific state of the robotic arm joints at each moment to ensure the response speed and control accuracy when controlling the robotic arm, thereby achieving compliant control of the robotic arm. Attached Figure Description

[0038] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 A flowchart of the robotic arm compliance control method for a die-casting mold manufacturing production line provided in this application. Detailed Implementation

[0040] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the robotic arm compliance control method for a die-casting mold manufacturing production line proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0042] The following description, in conjunction with the accompanying drawings, details the specific scheme of the robotic arm compliance control method for die-casting mold manufacturing production lines provided in this application.

[0043] One embodiment of this application provides a method for compliant control of a robotic arm in a die-casting mold manufacturing production line.

[0044] Specifically, the following methods for compliant control of robotic arms in die-casting mold manufacturing lines are provided. Please refer to [link / reference]. Figure 1 The method includes the following steps:

[0045] Step 1: Obtain sensor data for each joint of the robotic arm during its movement on the die-casting mold manufacturing production line.

[0046] First, to facilitate subsequent analysis of the coupling relationship between the joints of the robotic arm during complex movements on the die-casting mold manufacturing production line, each joint on the robotic arm is numbered sequentially. The numbering order can be based on the distance of the joint from the robotic arm / manipulator, including both ascending and descending order of distance. In other embodiments, the joints can also be numbered sequentially from the joint closest to the base towards the end effector. This application does not impose any special restrictions on the method of joint numbering.

[0047] Furthermore, torque sensors and speed sensors are used to synchronously collect torque timing data, speed timing data, and acceleration timing data for each joint during the movement of the robotic arm. The data acquisition period is the complete time period from start to stop of the robotic arm on the die-casting mold manufacturing line. In this embodiment, the acquisition frequency is set to 100Hz. In other embodiments, the frequency can be set according to the complexity of the robotic arm's prescribed actions. The setting standard is that the higher the complexity, the higher the frequency of sensor data changes within the same time period. A higher sampling frequency can be set to ensure that the sampled data can contain as complete data information as possible.

[0048] Step two: Utilize the low-rank essential structure between different sensor data to analyze the compensation coefficients at the same moment, which will serve as the proportional term parameters when the PID controller controls the robotic arm.

[0049] In the die-casting mold manufacturing line, due to factors such as mold shape, multi-station collaboration, and high speed and high load, the robotic arm is in complex motion. Not only is there a certain coupling effect between the joints of the robotic arm—that is, the movement of one joint is transmitted through the linkage mechanism, causing a linkage effect in other joints—but also, different mechanical data at each joint influence each other. For example, during joint movement, greater acceleration usually requires greater torque to maintain the motion state. Therefore, when performing compliant control of the robotic arm, it is necessary to analyze the degree of coupling influence between joints and dynamically adjust the compensation coefficient based on the coupling strength between joints.

[0050] First, a 5-dimensional high-order tensor is constructed by combining sensor data, sensor data acquisition time, and joint number, represented as follows: (a,b,c,d,t), where a represents a one-dimensional matrix composed of torque time series data; b represents a one-dimensional matrix composed of velocity time series data; c represents a one-dimensional matrix composed of joint numbers on the robotic arm; d represents a one-dimensional matrix composed of acceleration time series data; and t represents a one-dimensional matrix composed of acquisition times.

[0051] Secondly, Tucker decomposition is used to decompose the higher-order tensor into the product of a kernel tensor and five factor matrices. Each element in the kernel tensor represents the degree of interaction between different dimensions, and the five factor matrices represent the feature matrices in the torque, velocity, acceleration, joint, and time dimensions, respectively. , , , , Tucker decomposes tensor processing into a well-known technique in the field of tensor processing, and the specific process will not be elaborated here.

[0052] In this model, each element in the kernel tensor represents the degree of coupling between column vectors in the factor matrix across the five dimensions; the larger the element value, the stronger the coupling between the column vectors. Each column in the factor matrix represents a principal component after dimensionality reduction for the corresponding data dimension. Each element in the factor matrix represents the projection coefficient of the collected data onto the principal component corresponding to that element's column for the corresponding data dimension; the larger the projection coefficient, the stronger the correlation between the collected data and the principal component. Taking the torque dimension as an example, the element in the j-th row and u-th column of the factor matrix corresponding to the torque dimension represents the projection coefficient of the j-th collected torque data onto the u-th principal component. Decomposing higher-order tensors can effectively capture the coupling relationships between multi-axis joints when the robotic arm is in motion while processing a die-casting mold, achieving accurate extraction of the mechanical transmission characteristics between joints.

[0053] Then, for any joint, taking the i-th joint as an example, the coupling effect between the i-th joint and the other joints is analyzed by using the kernel tensor and the elements in the five factor matrices.

[0054] Specifically, torque data, velocity data, and acceleration data collected at different joints at the same time are obtained in the factor matrix. , , The projection coefficients on the same column vectors are used to form the sequence of projection coefficients of the original data collected at each joint at each time step in each dimension across all column vectors in each factor matrix. This sequence is taken as the coefficient sequence of the original data under that factor matrix. At a given time step, the stronger the motion coupling between joints, the more similar the motion trajectory information of the robotic arm contained in the data collected at the joints, the closer the magnitudes of the projection coefficients on the same principal component of the factor matrix, and the more similar the element distributions among the coefficient sequences under the same factor matrix.

[0055] Based on the above analysis, a coupling strength parameter is constructed here to characterize the strength of the motion coupling between two joints during the movement of the robotic arm. The coupling strength parameter between the i-th and (i+1)-th joints at time k is expressed as: :

[0056]

[0057] In the formula, M is the number of types of sensor data, which is taken as 3 in this embodiment. , These represent the data collected by the i-th joint and the (i+1)-th joint at time k, respectively, from the m-th sensor in the corresponding factor matrix. The coefficient sequence within the matrix, wherein the coefficient sequence is formed by the data of the m-th sensor in the corresponding factor matrix. It consists of the projection coefficients on all column vectors within the vector. This is a similarity parameter between two coefficient sequences. The similarity parameter is used to evaluate the similarity between the coefficient sequences; the larger the similarity parameter value, the higher the similarity. The specific method for obtaining the similarity parameter is as follows:

[0058] In Example 1, the similarity measure between two coefficient sequences is calculated as the similarity parameter between the coefficient sequences. The similarity measure includes, but is not limited to, Pearson correlation coefficient and Jaccard coefficient. In Example 2, the difference measure between two coefficient sequences is calculated first, and then the difference measure is negatively mapped. The negative mapping method includes, but is not limited to, taking the reciprocal and taking the opposite number. The difference measure method includes, but is not limited to, DTW distance, value variance (VSD), and Euclidean distance.

[0059] in, It is the coupling weight of the m-th sensor data dimension, and its calculation formula is as follows:

[0060]

[0061] In the formula, It is the number of columns in the factor matrix corresponding to the data dimension of the m-th sensor. It is the number of columns in the factor matrix corresponding to the p-th sensor data dimension.

[0062] Among them, the more principal components (i.e., the more columns) in the factor matrix of the m-th sensor data dimension, the more drastic the changes in the data collected under the m-th sensor data dimension during the multi-axis motion of the robotic arm, and more principal components are needed to capture the main changes in the data as the robotic arm moves; the fewer columns in the factor matrix, the smoother the data changes in this dimension, the more concentrated the data features, and a small number of principal components can reflect the main data information as the robotic arm moves.

[0063] Furthermore, based on the above steps, the coupling strength parameter between any two joints at each time step is calculated to construct the coupling strength matrix at each time step. Taking the coupling strength matrix at time k as an example, the coupling strength matrix in the i-th row and i+1-th column is... The elements on the diagonal are set to 0.

[0064] Subsequently, the changes in coupling strength between joints during the movement of the robotic arm are evaluated in real time based on the differences between coupling strength matrices at different times, and the compensation coefficients are dynamically adjusted accordingly.

[0065] Specifically, when the coupling strength between joints is higher, the compensation coefficient when controlling the robotic arm should be smaller to avoid overcompensation due to high coupling. When the coupling strength is lower, the compensation coefficient when controlling the robotic arm should be higher to suppress the coupling effect of joints during the movement of the robotic arm.

[0066] On the other hand, gain factors are usually used to control the response speed of compensation coefficients to system errors. When the change in coupling strength between joints is greater at adjacent moments, that is, the difference between coupling strength matrices at adjacent moments is greater, a larger gain factor should be used to control the response speed to respond quickly in order to avoid the accumulation of coupling errors when controlling the robotic arm. Conversely, when the change in point coupling strength between joints is small at adjacent moments, a smaller gain factor should be used to achieve real-time control of the robotic arm, so as to avoid pauses and stutters when the robotic arm performs actions such as mold closing, opening, and ejection on the manufacturing line, and to ensure the smoothness of control.

[0067] For each time point, taking time point k as an example, the off-diagonal elements in the coupling strength matrix at time point k are larger, indicating that there is a strong coupling between multiple joints of the robotic arm at time point k. At this time, more precise control is required to avoid control imbalance. At the same time, the greater the difference between the coupling strength matrices at time point k and time point k-1, the more abrupt the joint coupling occurs in a short period of time. The movement of some joints will force the other joints to perform unexpected actions. For example, when the robotic arm grasps and transports a die-casting mold, the abrupt coupling will force other joints to adjust to maintain the overall balance of the robotic arm, which may easily lead to grasping or transport failure.

[0068] Based on the above analysis, we first determine the real-time update method of the gain factor based on the degree of difference between the coupling strength matrices at adjacent time points. The formula for calculating the gain factor at time k is:

[0069]

[0070] In the formula, , These are the gain factors at time k and time k-1, respectively. , , These are the coupling strength matrices at time k, k-1, and k-2, respectively. It is a matrix and The results of the difference measurement between them It is a matrix and The difference between the matrices is measured, and the larger the value of the difference measure, the greater the degree of difference between the matrices. The difference measure includes, but is not limited to, Euclidean distance, relative error, and Manhattan distance. Specifically, at the initial first and second time points, the robotic arms are in the early stages of motion initiation, and the joint coupling between the robotic arms is considered relatively weak. In this embodiment, the value is 0.5.

[0071] Then, the compensation coefficient for each time step is determined by combining the magnitudes of the off-diagonal elements in the coupling strength matrix at each time step. The compensation coefficient at time k is expressed as... :

[0072]

[0073] In the formula, This is the initial preset value; in this embodiment, it is set to 0.5. It is the mean of the off-diagonal elements in the coupling strength matrix at time k, used to reflect the overall strength of the coupling between all joints on the robotic arm.

[0074] Step 3: By combining the compensation coefficient at the same moment with the distance between the desired position and the actual position of the robotic arm joint, the compliant control of the robotic arm is achieved.

[0075] S1, with the connection point between the mounting base of the robotic arm and the first joint as the origin of the coordinate system, according to the specified task of the robotic arm handling the die-casting mold on the manufacturing production line, the desired posture of the robotic arm at each moment, or the desired position of the joint, is manually set. The target path is generated using techniques such as polynomial interpolation or spline curve fitting. This is a commonly used technique in the field of robotic arm control, and the specific process will not be elaborated here.

[0076] S2, the actual positions of all joints are measured in real time by sensors. The sum of the distances between the actual and desired positions of all joints at each time step is used as the control error U at each time step. The control error at time k is denoted as... .

[0077] S3 generates the control quantity of the PID controller at each time step based on the compensation coefficient and control error at each time step, and expresses the control quantity at time step k as follows: :

[0078]

[0079] In the formula, It is the compensation coefficient at time k, which serves as the proportional term parameter when the PID controller controls the robotic arm, thereby achieving compliant control of the robotic arm. , Let k and t be the control errors, respectively. for The derivative at time k, , , These are the integral and derivative terms preset by the PID controller when controlling the robotic arm, respectively. In this embodiment, they are set to 0.01 and 0.05, respectively.

[0080] S4. The robotic arm is controlled based on the control quantity of the PID controller at each moment. In embodiment 1, the control quantity at each moment can be used as the current reference value of the robotic arm drive motor. The mechanical arm driver outputs torque, and the control of the robotic arm is realized based on the torque. In another embodiment, the control quantity output at each moment is converted into a PWM duty cycle through linear transformation. The PWM controller is used to convert the PWM duty cycle into a PWM signal. The drive circuit of the driver controls the robotic arm in real time according to the PWM signal.

[0081] The above technical features constitute the preferred embodiment of this application, which has strong adaptability and the best implementation effect. Unnecessary technical features can be added or removed according to actual needs to meet the needs of different situations.

Claims

1. A method for compliant control of robotic arms used in die-casting mold manufacturing production lines, characterized in that, The method includes the following steps: Acquire sensor data for each joint of the robotic arm during its movement on the die-casting mold manufacturing production line; A higher-order tensor is constructed using sensor data, its acquisition time, and joint numbers; the higher-order tensor is decomposed into a kernel tensor and the product of multiple factor matrices using Tucker decomposition. By using the similarity parameter of the projection coefficients of the same sensor data collected between any two joints at the same time in the corresponding factor matrix, the coupling strength parameter between any two joints at each time is determined; and the coupling strength matrix is ​​constructed using the coupling strength parameters between all joints at the same time. The gain factor at each time step is determined iteratively by measuring the difference between the coupling strength matrices at adjacent time steps; the compensation coefficient at each time step is determined based on the size of the off-diagonal elements in the coupling strength matrix at each time step. By combining the compensation coefficient at the same moment with the distance between the desired position and the actual position of the robotic arm joint, compliant control of the robotic arm can be achieved. The method for determining the coupling strength parameter between any two joints at each time point is as follows: The coupling strength parameter between the i-th and (i+1)-th joints at time k is expressed as: : In the formula, M is the number of types of sensor data. , These represent the data collected by the i-th joint and the (i+1)-th joint at time k, respectively, from the m-th sensor in the corresponding factor matrix. The coefficient sequence within the matrix, wherein the coefficient sequence is formed by the data of the m-th sensor in the corresponding factor matrix. It consists of the projection coefficients on all column vectors within the vector. It is a similarity parameter between two coefficient sequences. It is the coupling weight of the m-th sensor data dimension, and its calculation formula is as follows: , It is the number of columns in the factor matrix corresponding to the data dimension of the m-th sensor. It is the number of columns in the factor matrix corresponding to the data dimension of the p-th sensor; The method for iteratively determining the gain factor at each time step is as follows: In the formula, , These are the gain factors at time k and time k-1, respectively. , , These are the coupling strength matrices at time k, k-1, and k-2, respectively. It is a matrix and The results of the difference measurement between them It is a matrix and The results of the difference measurement between them.

2. The robotic arm compliance control method for a die-casting mold manufacturing line as described in claim 1, characterized in that, The sensor data includes torque time-series data, velocity time-series data, and acceleration time-series data.

3. The robotic arm compliance control method for a die-casting mold manufacturing production line as described in claim 2, characterized in that, There are five factor matrices, with the first three factor matrices corresponding to three types of sensor data, respectively.

4. The robotic arm compliance control method for a die-casting mold manufacturing line as described in claim 1, characterized in that, The elements on the main diagonal of the coupling strength matrix are 0, and the values ​​of the remaining elements are the coupling strength parameters between the two joints in the row and column at the same time.

5. The robotic arm compliance control method for a die-casting mold manufacturing line as described in claim 1, characterized in that, The dissimilarity measures include, but are not limited to, Euclidean distance, relative error, and Manhattan distance.

6. The robotic arm compliance control method for a die-casting mold manufacturing line as described in claim 1, characterized in that, The method for determining the compensation coefficient at each time point is as follows: In the formula, It is the compensation coefficient at time k. This is the initial preset value. It is the mean of the off-diagonal elements in the coupling strength matrix at time k.

7. The robotic arm compliance control method for a die-casting mold manufacturing line as described in claim 1, characterized in that, The method for achieving compliant control of the robotic arm is as follows: S1, obtain the target path of the robotic arm, including the expected position of each joint at each time step; S2, the sum of the distances between the actual positions and the desired positions of all joints at each time step is used as the control error at each time step; S3 generates the control input of the PID controller for each time step based on the compensation coefficient and control error at each time step; S4 controls the robotic arm based on the control input of the PID controller at each time step.

8. The robotic arm compliance control method for a die-casting mold manufacturing line as described in claim 7, characterized in that, In step S3, the calculation method for the control quantity of the PID controller at each time point is as follows: The control quantity at time k is expressed as... : In the formula, It is the compensation coefficient at time k, which serves as the proportional term parameter when the PID controller controls the robotic arm. , Let k and t be the control errors, respectively. for The derivative at time k, , , These are the preset integral and derivative terms when the PID controller controls the robotic arm.

Images