A robot collaborative sewing method and system based on human-machine skill transfer

Abstract

The present application belongs to the technical field of industrial robots, and particularly relates to a robot collaborative sewing method and system based on human-machine skill transfer, which comprises the following steps: acquiring the hand movement trajectory and fabric tension of a demonstrator during sewing; obtaining a sewing initial trajectory according to a dynamic motion primitive and the acquired hand movement trajectory; obtaining a sewing desired tension according to a force primitive and the acquired fabric tension; acquiring real-time sewing tension of the fabric, combining the obtained sewing desired tension and an impedance control model to obtain a tension compensation trajectory for robot sewing; judging in real time whether the fabric is wrinkled to obtain a fabric wrinkle compensation trajectory; and obtaining an actual sewing trajectory of the robot according to the obtained sewing initial trajectory, sewing tension compensation trajectory and wrinkle compensation trajectory, so as to complete robot collaborative sewing.

Description

Technical Field

[0001] This invention belongs to the field of industrial robot technology, specifically relating to a robot collaborative sewing method and system based on human-machine skill transfer. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] As a labor-intensive industry, the garment manufacturing sector faces the challenge of improving efficiency, increasing speed, reducing costs, and upgrading its development amidst labor shortages and high labor costs. Currently, automation in the garment manufacturing industry is mainly reflected in sewing machinery, which has become an advanced equipment manufacturing industry integrating optics, mechanics, and electronics, providing complete sewing technology solutions for textiles, garments, automobiles, building materials, aerospace, and other fields. However, a large number of professional personnel are still needed to operate the equipment, and there is an urgent need to de-specialize operations. As the garment manufacturing market gradually enters the era of personalized consumption, consumers have increasingly higher requirements for clothing styles and quality. Products are becoming more customized and the processes are becoming more complex, while delivery times are getting shorter. Orders are shifting from single-variety, large-volume to multi-variety, small-volume, making the advantages of single automated equipment less obvious.

[0004] To adapt to the practice of intelligent manufacturing in the sewing industry, the dynamic processing module composed of "robots + automated intelligent sewing equipment" is an inevitable transformation in future sewing production. However, sewing differs from operations such as sanding and assembly. Due to the flexible fabrics it handles, which exhibit material anisotropy and difficulty in controlling deformation, robots struggle to complete sewing tasks in a precise and efficient manner in collaboration with sewing machines. Traditional robot-assisted sewing methods primarily rely on visual detection for sewing trajectory planning, using simple linear functions to correspond to the collaborative speed between the robot and the sewing machine, and depending on open-loop position control to achieve sewing. This approach cannot achieve precise collaborative sewing, and the robot cannot adjust its posture according to the fabric's condition during the sewing process, easily leading to fabric wrinkles and sewing failures. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a robot collaborative sewing method and system based on human-machine skill transfer. Within the framework of human-machine skill transfer, it combines multimodal information such as tactile, force, and visual senses to transfer the sewing skills of human experts to the robot. The robot can dynamically adjust itself according to the sewing process, making it widely applicable and achieving compliant collaborative sewing. This effectively improves the robot's ability to manipulate soft materials and other easily deformable objects.

[0006] According to some embodiments, the first aspect of the present invention provides a robot collaborative sewing method based on human-machine skill transfer, employing the following technical solution:

[0007] A robotic collaborative sewing method based on human-machine skill transfer includes:

[0008] Acquire the demonstrator's hand movement trajectory and fabric tension during the sewing process;

[0009] Based on the dynamic motion primitives and the acquired hand movement trajectory, the initial sewing trajectory is obtained;

[0010] Based on the force element and the obtained fabric tension, the desired sewing tension is obtained;

[0011] The real-time sewing tension of the fabric is obtained, and combined with the obtained sewing expected tension and impedance control model, the tension compensation trajectory of robot sewing is obtained.

[0012] Real-time determination of whether the fabric is wrinkled, and obtaining the fabric wrinkle compensation trajectory;

[0013] Based on the obtained initial sewing trajectory, sewing tension compensation trajectory, and wrinkle compensation trajectory, the actual sewing trajectory of the robot is obtained, and the robot collaborative sewing is completed.

[0014] As a further technical limitation, the obtained sewing tension compensation trajectory and wrinkle compensation trajectory are superimposed on the initial sewing trajectory of the robot, and the superimposed trajectory is the actual sewing trajectory of the robot.

[0015] As a further technical limitation, the specific process of determining whether the fabric is wrinkled in real time and obtaining the fabric wrinkle compensation trajectory is as follows:

[0016] Using the direction θ n Gabor filter extracts wrinkles f n , where f n It represents the number of pixels in the wrinkles;

[0017] Using Gaussian Mixture Model (GMM) clustering, the wrinkled clusters {w1,...,w} are obtained. n}, each fold w n Contains the corresponding pixel N n ;

[0018] Select the w with the most pixels from the fold cluster. nmax To obtain the maximum fold, get w. nmax The maximum value of the middle pixel on the x-axis and y-axis. max y max With minimum value x min y min Calculate the endpoint (x) of the largest fold. p1 ,y p1 ), (x p2 ,y p2 ) Length l and midpoint (x) c ,yc ); where, (x p1 ,y p1 )=(x max ,y),(x p2 ,y p2 )=(x min ,y),l=max(|(x max ,y)-(x min ,y)|,|(x,y max )-(x,y min )|)

[0019] During each robot flattening cycle, along the direction of the perpendicular bisector of the maximum wrinkle. Drag The distance, i.e., the fabric wrinkle compensation trajectory X I .

[0020] As a further technical limitation, the specific process of obtaining the real-time sewing tension of the fabric and combining it with the obtained desired sewing tension and impedance control model to obtain the tension compensation trajectory of the robot sewing is as follows:

[0021] The physical control model of "mass-damping-spring" is adopted as the impedance control model, that is: Where M, B, and K represent the inertia matrix, damping matrix, and stiffness matrix, respectively; These represent the acceleration, velocity, and position of the robot's end effector, respectively. Let F represent the desired acceleration, velocity, and position of the robot control system, respectively; force error E = F r -F e As the driving force of the model, F r For the desired sewing tension, F e For real-time tension in the sewing environment;

[0022] By combining position-based impedance control, the contact force between the robot's end tool and the environment is obtained through a force sensor. The force error enables the impedance controller to generate a position correction amount e for the robot sewing, thus obtaining the tension compensation trajectory of the robot sewing.

[0023] As a further technical limitation, force data f is obtained from each contact point of the distributed tactile sensor. ix ,f iy Taking the sewing direction as the positive direction, and assuming the rotation angle of the tactile sensor is β, the force is concentrated at the operation center point for analysis. The overall force data F along the sewing direction is... y Force data F perpendicular to the sewing direction x for: Let the angle between the sewing tension F along the line connecting the operation center point and the sewing point and the sewing direction be α. Then the desired sewing tension is F = F x / sinα.

[0024] As a further technical limitation, a dynamic motion primitive model is used to learn the acquired hand movement trajectory during the process of obtaining the initial sewing trajectory; the dynamic motion primitive model used is: Where α and β are constant coefficients, and α = 4β; p is the target position, and x is the position at any time. Let the velocity be at any given moment. τ is the acceleration at any given time; s is the system phase; τ is the time scaling factor; α s is a constant coefficient, and f(s) is a nonlinear forcing term.

[0025] As a further technical limitation, in the process of obtaining fabric tension, force data of the fabric contact point is obtained. Taking the sewing direction as the positive direction, the angle between the line connecting the fabric contact point and the sewing point and the opposite direction of the sewing direction is obtained. The fabric tension is obtained by combining the obtained angle with the force data of the fabric contact point.

[0026] According to some embodiments, a second aspect of the present invention provides a robot collaborative sewing system based on human-machine skill transfer, employing the following technical solution:

[0027] A robotic collaborative sewing system based on human-machine skill transfer includes:

[0028] The acquisition module is configured to acquire the movement trajectory of the demonstrator's hands and the fabric tension during the sewing process;

[0029] The calculation module is configured to obtain the initial sewing trajectory based on the dynamic motion primitives and the acquired hand movement trajectory; obtain the desired sewing tension based on the force primitives and the acquired fabric tension; acquire the real-time sewing tension of the fabric, and combine the obtained desired sewing tension and impedance control model to obtain the tension compensation trajectory for robot sewing; and determine in real time whether the fabric is wrinkled to obtain the fabric wrinkle compensation trajectory.

[0030] The sewing module is configured to obtain the robot's actual sewing trajectory based on the obtained initial sewing trajectory, sewing tension compensation trajectory, and fold compensation trajectory, thereby completing the robot's collaborative sewing.

[0031] According to some embodiments, a third aspect of the present invention provides a computer-readable storage medium, employing the following technical solution:

[0032] A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in the robot collaborative sewing method based on human-machine skill transfer as described in the first aspect of the present invention.

[0033] According to some embodiments, the fourth aspect of the present invention provides an electronic device, which adopts the following technical solution:

[0034] An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps in the robot collaborative sewing method based on human-machine skill transfer as described in the first aspect of the present invention.

[0035] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0036] This invention, based on a human-machine skill transfer framework, improves the robot's ability to manipulate soft materials and promotes the application of robots in the garment manufacturing industry. It combines trajectory information and tactile force information during the sewing process, and realizes robot-assisted compliant sewing based on multi-modal representation, which is applicable to different fabric materials. It incorporates a wrinkle observer into the traditional impedance control model, and dynamically adjusts the sewing operation based on the wrinkle images during the fabric sewing process to avoid sewing wrinkles and improve sewing quality. Attached Figure Description

[0037] The accompanying drawings, which form part of this embodiment, are used to provide a further understanding of this embodiment. The illustrative embodiments and their descriptions are used to explain this embodiment and do not constitute an improper limitation of this embodiment.

[0038] Figure 1 This is a flowchart of the robot collaborative sewing method based on human-machine skill transfer in Embodiment 1 of the present invention;

[0039] Figure 2 This is a schematic diagram of the robot collaborative sewing skill learning process in Embodiment 1 of the present invention;

[0040] Figure 3 This is a schematic diagram of tension estimation during the sewing process in Embodiment 1 of the present invention;

[0041] Figure 4 This is a flowchart of the fold observer in Embodiment 1 of the present invention;

[0042] Figure 5 This is a flowchart of the robot collaborative sewing of fabric based on impedance control in Embodiment 1 of the present invention;

[0043] Figure 6 This is a structural block diagram of the robot collaborative sewing system based on human-machine skill transfer in Embodiment 2 of the present invention. Detailed Implementation

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0045] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, 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.

[0046] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0047] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0048] In this invention, terms such as "fixed connection," "connected," and "linked" should be interpreted broadly, indicating a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can determine the specific meaning of these terms in this invention based on the specific circumstances, and they should not be construed as limitations on the invention.

[0049] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0050] Example 1

[0051] Embodiment 1 of this invention introduces a robot collaborative sewing method based on human-machine skill transfer.

[0052] like Figure 1 The illustrated robotic collaborative sewing method based on human-machine skill transfer includes:

[0053] Acquire the demonstrator's hand movement trajectory and fabric tension during the sewing process;

[0054] Based on the dynamic motion primitives and the acquired hand movement trajectory, the initial sewing trajectory is obtained;

[0055] Based on the force element and the obtained fabric tension, the desired sewing tension is obtained;

[0056] The real-time sewing tension of the fabric is obtained, and combined with the obtained sewing expected tension and impedance control model, the tension compensation trajectory of robot sewing is obtained.

[0057] Real-time determination of whether the fabric is wrinkled, and obtaining the fabric wrinkle compensation trajectory;

[0058] Based on the obtained initial sewing trajectory, sewing tension compensation trajectory, and wrinkle compensation trajectory, the actual sewing trajectory of the robot is obtained, and the robot collaborative sewing is completed.

[0059] like Figure 2 As shown, the robot collaborative sewing method based on human-machine skill transfer in this embodiment mainly includes three stages: expert sewing demonstration, sewing skill modeling, and sewing skill reproduction. First, in the expert sewing demonstration stage, the motion trajectory of the demonstrator's hand during the sewing process is extracted using an existing gesture recognition network and converted into a robot reference trajectory through a hand-eye calibration matrix. At the same time, a tactile sensor installed on the demonstrator's hand provides a reference tension during the fabric sewing process. Second, in the sewing skill modeling stage, the robot reference motion trajectory is modeled using the dynamic motion primitive method. Then, based on the contact force, the desired fabric tension during the robot sewing process is estimated. A set of linearly combined Gaussian functions is used to form a force primitive to model the sewing tension characteristic curve. A regularized system simultaneously drives the dynamic motion primitive (DMP) model and the force primitive (FP) model to ensure phase synchronization between the motion trajectory and the force curve. The motion trajectory output by the DMP model is extracted and combined with the force primitive and the wrinkle observer. Impedance control is used to reproduce the robot's sewing operation. When the robot moves, it will be subjected to the driving force from the trajectory generated by the DMP and the correction force from the force control.

[0060] As one or more implementation methods, a camera is used to capture a video of a demonstrator sewing, and a media pipe model is used to extract the movement trajectory X of the index finger in the video. At the same time, information from a tactile sensor installed in the index finger during this process is collected, and the tension F of the fabric held during sewing is estimated.

[0061] like Figure 3 As shown, force data f is obtained for each contact point of the distributed tactile sensor. ix ,f iy Taking the sewing direction as the positive direction, let the rotation angle of the tactile sensor be β, which can be obtained from the rotation angle of the robotic arm's end effector. For ease of calculation, the force is concentrated at the operation center point for analysis, and the overall force data F along the sewing direction is considered. y Force data F perpendicular to the sewing direction x for: Let the angle between the sewing tension F along the line connecting the operation center point and the sewing point and the sewing direction be α. Then, the expected tension during the sewing process is estimated as F = F x / sinα.

[0062] As one or more implementation methods, the task trajectory is learned based on the DMP algorithm, and the mathematical model of the DMP used is as follows: Where α and β are constant coefficients, α = 4β; p is the target position, and x is the position at any given time. Let the velocity be at any given moment. Let be the acceleration at any given time; s be the system phase; τ be the time scaling factor, where the system phase s satisfies Also known as a regular system, α s The coefficients are constant and satisfy the following conditions: Where s is a function of time t; f(s) is a nonlinear forcing term.

[0063] To fit the trajectories of complex motions, the forcing term needs to be able to represent arbitrarily complex curves. In the DMP model, the forcing term is defined using a weighted linear combination of a set of Gaussian functions:

[0064]

[0065]

[0066] Where x0 is the initial position. Let ω be the Gaussian function. i Here, c represents the weighting coefficients, N is the number of basis functions, and c represents the weighting coefficients. i h represents the center coordinates of the Gaussian function. i Let V be the variance of the Gaussian kernel function. By determining the number of Gaussian kernel functions and their weight parameters, the characteristics of the demonstration trajectory can be learned.

[0067] Through visual demonstration by the demonstrator, combined with hand-eye calibration, the position, velocity, and acceleration information of the robot's end effector at any given moment in the XY direction can be obtained. Then, by setting constant coefficients in the model, the f(s) value at each moment can be calculated based on the mathematical model of the DMP (Distributed Dynamic Model).

[0068] Based on the multiple f obtained demo (s t The value learns the weights of each Gaussian kernel function, such that f t (s) values ​​should be as close as possible to a series of f values ​​obtained from the demonstration trajectory information. t The (s) value represents the function approximation problem.

[0069] By minimizing the lower error function ξ(s(t))=s(t)(p-x0); The Locally Weighted Regression (LWR) algorithm is used to learn the weight parameters, and finally the weights of the i-th Gaussian kernel function are obtained. in,

[0070] Complete the parameter ω i The estimation is completed, and the parameters in the model are determined. Then, the starting point, target point and running time in the model are set according to the actual task of the robot. Then, the motion trajectory can be generated under the drive of the system phase s.

[0071] Similarly to the DMP model, M Gaussian kernel functions are used to fit the tension curve during the teaching process, while a regularized system is used for driving, i.e. Based on the tension data collected at each moment during the demonstration, a sequence can be obtained. Similarly, by minimizing the error The weight parameters are learned using a locally weighted regression algorithm, and the solution is obtained. in, By learning the weighting parameters, a model can be obtained that fits the tension during the demonstration process.

[0072] As one or more implementation methods, such as Figure 4 As shown, fabric tension changes can serve as a priori criterion for determining non-coordinated sewing; while fabric wrinkles are a result of non-coordinated sewing, therefore a wrinkle observer can serve as a posterior criterion for determining non-coordinated sewing. A constant force tracking impedance controller is designed by combining these two pieces of information, and stable coordinated sewing is ensured by adjusting the impedance control coefficient in real time.

[0073] A color image I of the fabric captured by a local camera is acquired, and the fabric foreground is segmented using HSV. Then, an application direction θ is applied to the segmented fabric image. n Gabor filter extracts wrinkles f n The fold cluster with the most pixels is selected as the largest fold. The length l of the largest fold and the vertical bisector r of the largest fold are calculated. The robot adjusts its pose along the direction of the vertical bisector.

[0074] As one or more implementation methods, the robot impedance control model is as follows: Where M, B, and K are the inertia, damping, and stiffness matrix parameters in the robot impedance control model; These are the actual position, velocity, and acceleration of the robot's end effector, respectively. The desired position, velocity, and acceleration output by the DMP model.

[0075] like Figure 5 As shown, the actual sewing contact tension F e The desired contact tension F output by the FP model r The deviation input impedance control model is used to obtain the end position correction amount e of the robot sewing process, and then the correction amount e is compared with the reference position information X. rAnd the compensation X detected by the fold observer r By combining the trajectories, the final robot sewing trajectory is obtained.

[0076] This embodiment, based on a human-machine skill transfer framework, improves the robot's ability to manipulate soft materials and promotes the application of robots in the garment manufacturing industry. By combining trajectory information and tactile force information during the sewing process, and based on multi-modal representation, it achieves compliant sewing through robot collaboration, applicable to different fabric materials. A wrinkle observer is added to the traditional impedance control model, and the sewing operation is dynamically adjusted according to the wrinkle images during the fabric sewing process to avoid sewing wrinkles and improve sewing quality.

[0077] Example 2

[0078] Embodiment 2 of the present invention introduces a robot collaborative sewing system based on human-machine skill transfer.

[0079] like Figure 6 The illustrated robotic collaborative sewing system based on human-machine skill transfer includes:

[0080] The acquisition module is configured to acquire the movement trajectory of the demonstrator's hands and the fabric tension during the sewing process;

[0081] The calculation module is configured to obtain the initial sewing trajectory based on the dynamic motion primitives and the acquired hand movement trajectory; obtain the desired sewing tension based on the force primitives and the acquired fabric tension; acquire the real-time sewing tension of the fabric, and combine the obtained desired sewing tension and impedance control model to obtain the tension compensation trajectory for robot sewing; and determine in real time whether the fabric is wrinkled to obtain the fabric wrinkle compensation trajectory.

[0082] The sewing module is configured to obtain the robot's actual sewing trajectory based on the obtained initial sewing trajectory, sewing tension compensation trajectory, and fold compensation trajectory, thereby completing the robot's collaborative sewing.

[0083] The detailed steps are the same as those of the robot collaborative sewing method based on human-machine skill transfer provided in Example 1, and will not be repeated here.

[0084] Example 3

[0085] Embodiment 3 of the present invention provides a computer-readable storage medium.

[0086] A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of the robot collaborative sewing method based on human-machine skill transfer as described in Embodiment 1 of the present invention.

[0087] The detailed steps are the same as those of the robot collaborative sewing method based on human-machine skill transfer provided in Example 1, and will not be repeated here.

[0088] Example 4

[0089] Embodiment 4 of the present invention provides an electronic device.

[0090] An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor. When the processor executes the program, it implements the steps in the robot collaborative sewing method based on human-machine skill transfer as described in Embodiment 1 of the present invention.

[0091] The detailed steps are the same as those of the robot collaborative sewing method based on human-machine skill transfer provided in Example 1, and will not be repeated here.

[0092] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A robot collaborative sewing method based on human-machine skill transfer, characterized in that, include: Acquire the demonstrator's hand movement trajectory and fabric tension during the sewing process; Based on the dynamic motion primitives and the acquired hand movement trajectory, the initial sewing trajectory is obtained; In the process of acquiring the initial sewing trajectory, a dynamic motion primitive model is used to learn the acquired hand movement trajectory; the dynamic motion primitive model used is: ;in, and All are constant coefficients, and ; For the target location, For any given time, Let the velocity be at any given moment. Let be the acceleration at any given moment; s For system phase; This is a time scaling factor. It is a nonlinear forcing term; Based on the force element and the obtained fabric tension, the desired sewing tension is obtained; The real-time sewing tension of the fabric is obtained, and combined with the obtained desired sewing tension and impedance control model, the tension compensation trajectory of the robot sewing is obtained. The specific process is as follows: The physical control model of "mass-damping-spring" is adopted as the impedance control model, that is: ;in, These represent the inertia matrix, damping matrix, and stiffness matrix, respectively. These represent the acceleration, velocity, and position of the robot's end effector, respectively. These represent the desired acceleration, velocity, and position of the robot control system, respectively; force error. As the driving force of the model, For sewing desired tension, For real-time tension in the sewing environment; Combining position-based impedance control, the contact force between the robot's end-effector and the environment is obtained through a force sensor. The force error is used to enable the impedance controller to generate a position correction for the robot's sewing process. e That is, to obtain the tension compensation trajectory of the robot sewing; Real-time determination of fabric wrinkles and obtaining fabric wrinkle compensation trajectory, the specific process is as follows: The direction adopted is Gabor filter extracts wrinkles ,in It represents the number of pixels in the wrinkles; Clustering using Gaussian mixture model yields wrinkled clusters. Each fold Contains corresponding pixels ; Select the cluster with the most pixels. To obtain the maximum fold, Medium pixel axis, Maximum value of the axis , and minimum value , Calculate the endpoints of the largest fold. , length and midpoint ;in, , , , ; During each robot flattening cycle, along the direction of the perpendicular bisector of the maximum wrinkle. Drag The distance, i.e., the fabric wrinkle compensation trajectory. X I ; Based on the obtained initial sewing trajectory, sewing tension compensation trajectory, and wrinkle compensation trajectory, the actual sewing trajectory of the robot is obtained, and the robot collaborative sewing is completed.

2. The robot collaborative sewing method based on human-machine skill transfer as described in claim 1, characterized in that, Based on the initial sewing trajectory of the robot, the obtained sewing tension compensation trajectory and fold compensation trajectory are superimposed respectively. The superimposed trajectory is the actual sewing trajectory of the robot.

3. The robot collaborative sewing method based on human-machine skill transfer as described in claim 1, characterized in that, Acquire force data at each contact point of the distributed haptic sensor Taking the sewing direction as the positive direction, let the rotation angle of the tactile sensor be . By concentrating the force at the operation center point for analysis, the overall force data along the sewing direction is analyzed. Force data perpendicular to the sewing direction for: , ; Set the sewing tension along the line connecting the operation center point and the sewing point. F The angle between the sewing direction and the sewing direction is Then the expected sewing tension is .

4. The robot collaborative sewing method based on human-machine skill transfer as described in claim 1, characterized in that, In the process of obtaining fabric tension, the force data of the fabric contact point is obtained. Taking the sewing direction as the positive direction, the angle between the line connecting the fabric contact point and the sewing point and the opposite direction of the sewing direction is obtained. The fabric tension is obtained by combining the obtained angle with the force data of the fabric contact point.

5. A robotic collaborative sewing system based on human-machine skill transfer, characterized in that, include: The acquisition module is configured to acquire the movement trajectory of the demonstrator's hands and the fabric tension during the sewing process; The calculation module is configured to obtain the initial sewing trajectory based on dynamic motion primitives and the acquired hand movement trajectory. During the acquisition of the initial sewing trajectory, a dynamic motion primitive model is used to learn the acquired hand movement trajectory. The dynamic motion primitive model used is as follows: ;in, and All are constant coefficients, and ; For the target location, For any given time, Let the velocity be at any given moment. Let be the acceleration at any given moment; s For system phase; This is a time scaling factor. This is a nonlinear forcing term; based on the force element and the obtained fabric tension, the desired sewing tension is obtained. Specifically, a physical control model of "mass-damping-spring" is used as the impedance control model, i.e.: ;in, These represent the inertia matrix, damping matrix, and stiffness matrix, respectively. These represent the acceleration, velocity, and position of the robot's end effector, respectively. These represent the desired acceleration, velocity, and position of the robot control system, respectively; force error. As the driving force of the model, For sewing desired tension, For real-time tension in the sewing environment; combined with position-based impedance control, the contact force between the robot's end effector and the environment is obtained through a force sensor. The force error is used to generate a position correction amount for the robot's sewing. e This involves obtaining the tension compensation trajectory for robot sewing; acquiring the real-time sewing tension of the fabric; combining the obtained desired sewing tension and impedance control model to obtain the tension compensation trajectory for robot sewing; and determining in real-time whether the fabric is wrinkled to obtain the fabric wrinkle compensation trajectory. The specific process is as follows: using a direction... Gabor filter extracts wrinkles ,in This represents the number of pixels in the folds; clustering is performed using a Gaussian mixture model to obtain fold clusters. Each fold Contains corresponding pixels Select the cluster with the most pixels. To obtain the maximum fold, Medium pixel axis, Maximum value of the axis , and minimum value , Calculate the endpoints of the largest fold. , length and midpoint ;in, , , , During each robot flattening cycle, along the direction of the perpendicular bisector of the maximum wrinkle. Drag The distance, i.e., the fabric wrinkle compensation trajectory. X I ; The sewing module is configured to obtain the robot's actual sewing trajectory based on the obtained initial sewing trajectory, sewing tension compensation trajectory, and fold compensation trajectory, thereby completing the robot's collaborative sewing.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the robot collaborative sewing method based on human-machine skill transfer as described in any one of claims 1-4.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the steps of the robot collaborative sewing method based on human-machine skill transfer as described in any one of claims 1-4.

Images