An automatic textile piece feeding closed-loop judgment method, system, device and medium

Abstract

The application discloses a kind of textile piece automatic feeding closed loop judging method, system, equipment and medium, method includes: obtaining the basic parameter of piece to be grabbed, according to the basic parameter and preset grabbing planning model, determine grabbing scheme, and set the initial judgment threshold of perception unit;According to the grabbing scheme, pre-grabbing action is executed, and multiple-source perception data are obtained;The multiple-source perception data are judged with the initial judgment threshold, and the judgment result of current grabbing state is obtained;According to the judgment result, corresponding transfer blanking or closed loop error correction operation is executed, and grabbing and abnormal error correction process data are recorded, the judgment threshold of same type piece and grabbing parameter are updated, feeding cycle is carried out, realize textile piece automatic feeding closed loop judgment, adapt to batch, multi-style garment flexible production.

Description

Technical Field

[0001] This invention relates to the fields of intelligent garment manufacturing, robotic flexible grasping, and automated loading and unloading technology, and in particular to a closed-loop judgment method, system, equipment, and medium for automatic loading of textile pieces. Background Technology

[0002] In the process of intelligent upgrading of garment manufacturing, the layer-by-layer loading and unloading of textile pieces has become a bottleneck in automated sewing production lines. The stacking of multiple layers of fabric, resulting from cutting, requires individual layer-by-layer grasping. Misalignment can lead to color differences and defective products, while missed grasps and over-grabbing are the two main types of failures in automated loading. Currently, some existing technical solutions in the industry use thickness sensors to detect the thickness of the fabric pieces and set thresholds to distinguish between single and multiple layers. However, single thickness detection cannot identify missed grasping situations and only performs simple material arrangement, lacking closed-loop control logic and secondary grasping capabilities. Furthermore, it does not incorporate mechanical models such as fabric bending force and contact force, making it difficult to adapt to complex conditions such as knitted, woven, and lightweight fabrics, resulting in a high misjudgment rate. Some solutions use photoelectric or proximity sensors to detect the position of the fabric pieces to assist in picking, but these sensors only provide positional information and cannot quantitatively determine the number of layers or the grasping status. Without the support of mechanical perception and visual fusion algorithms, they do not solve the fundamental problem of separating stacked fabric pieces. Another approach utilizes changes in electrostatic capacitance to distinguish between single and multiple pieces of fabric. However, electrostatic detection methods cannot identify complete missed grasps and fail to detect antistatic and thick, stiff fabrics, limiting their application scenarios. While a few systems involving piece separation integrate visual inspection units, they focus on the backend sorting and packaging stages, neglecting real-time closed-loop judgment of missed / multiple grasps during the feeding process. They lack iterative correction mechanisms after detecting anomalies and have not yet established a fully closed-loop logic. In summary, existing technologies generally rely on single-type sensors, unable to simultaneously identify both missed and multiple grasp anomalies; they are mostly open-loop detection, directly stopping or discharging materials after anomaly handling, lacking adaptive parameter adjustment and closed-loop error correction capabilities; they lack deep integration of fabric mechanics models, resulting in poor adaptability to all types of fabrics, and the feeding accuracy cannot meet the high-precision requirements of automated garment production.

[0003] To address the two types of abnormalities—missing or over-grabbing—in the automatic feeding process of layered textile pieces, this invention proposes a closed-loop judgment method for automatic feeding of textile pieces, coupled with a hierarchical adaptive error correction strategy, to achieve high-accuracy layer-by-layer feeding of textile pieces of all categories, and is compatible with automated garment sewing robot production lines. Summary of the Invention

[0004] In view of the above-mentioned existing problems, the present invention provides a closed-loop judgment method, system, equipment and medium for automatic feeding of textile pieces, which solves two types of abnormal problems in the automatic feeding process of textile pieces: empty grab (i.e., no piece is grabbed) and too many grab (i.e., grabbing ≥2 layers of pieces).

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a closed-loop judgment method for automatic feeding of textile cut pieces, comprising: Obtain the basic parameters of the cut piece to be grasped, determine the grasping scheme based on the basic parameters and the preset grasping planning model, and set the initial judgment threshold of the sensing unit. Perform pre-grabbing actions according to the grasping scheme and acquire multi-source sensing data; The multi-source sensing data is compared with the initial judgment threshold to obtain the judgment result of the current grasping state; Based on the judgment result, the corresponding transfer and unloading or closed-loop error correction operation is executed, and the data of the grabbing and abnormal error correction process is recorded. The judgment threshold and grabbing parameters of the same type of cut pieces are updated, and the feeding cycle is carried out to realize the automatic feeding closed-loop judgment of textile cut pieces.

[0006] As a preferred embodiment of the closed-loop judgment method for automatic feeding of textile pieces according to the present invention, the method for determining the feeding scheme based on the basic parameters and the preset feeding planning model includes: Based on the fabric weight and cut area in the basic parameters, and combined with the single-point output force of the soft robotic arm, determine the minimum number of grippers involved in grasping. According to the preset shape-preserving gripping rules, the gripping points of each gripper are determined at the outer contour boundary and corner intersection of the cut piece; Based on the minimum number of grippers and the gripping points, a gripping plan is obtained.

[0007] The beneficial effect of this preferred technical solution is that by combining the fabric cutting parameters to determine the number and position of grippers, the gripping force is balanced, and the fabric slippage is prevented.

[0008] As a preferred embodiment of the closed-loop judgment method for automatic feeding of textile cut pieces according to the present invention, the method includes: performing a pre-grabbing action according to the grasping scheme and acquiring multi-source sensing data, including: The control software robotic arm moves to the top of the stack of layered cut pieces according to the gripping point and performs a pre-pressing action; During the pre-pressing and grasping process, the contact force data of the soft robotic arm is collected by force sensors; Visual morphological data of the captured area of ​​the cut piece is collected through visual units; Based on the stiffness coefficient, friction factor, and thickness change rate in the basic parameters, and combined with the downward pressure of the pre-pressing action, the grasping state characterization value of the cut piece is obtained.

[0009] The beneficial effects of this preferred technical solution are that the present invention performs pre-pressing and collects multi-source sensing data to comprehensively obtain the cutting piece grasping condition information, providing a basis for state judgment.

[0010] In a preferred embodiment of the closed-loop judgment method for automatic feeding of textile pieces according to the present invention, the judgment result of the current grasping state by judging the multi-source sensing data with the initial judgment threshold includes: The contact force data is compared with the contact force threshold range in the initial judgment threshold to obtain the first comparison result; The visual morphology data is compared with the visual morphology baseline value in the initial judgment threshold to obtain the second comparison result; The capture status representation value is compared with the qualified range of the representation value in the initial judgment threshold to obtain the third comparison result; Based on the combined results of the first, second, and third comparisons, the current capture status is output as one of the following: empty capture, normal single-layer capture, or multiple captures.

[0011] The beneficial effect of this preferred technical solution is that by comparing and judging the grasping status from multiple dimensions, it reduces the misjudgment of a single sensor and improves the reliability of piece recognition.

[0012] As a preferred embodiment of the closed-loop judgment method for automatic feeding of textile cut pieces according to the present invention, the method outputs one of the following based on the first comparison result, the second comparison result, and the third comparison result: the current grasping state is determined by combining the results of the first comparison, the second comparison, and the third comparison, including: […]. In response to the first comparison result that the contact force data is less than the empty grasp threshold in the contact force threshold range, the second comparison result that the visual morphology data does not match the cut piece outline reference, and the third comparison result that the grasping state characterization value is close to zero, the current grasping state is determined to be empty grasping. In response to the first comparison result that the contact force data is within the single-layer force range of the contact force threshold range, the second comparison result that the visual morphology data matches the single-layer cut piece morphology benchmark, and the third comparison result that the grasping state characterization value is within the qualified range of characterization value, the current grasping state is determined to be a normal single layer. In response to the first comparison result that the contact force data exceeds the multi-layer force threshold in the contact force threshold range, the second comparison result that the visual morphology data presents a multi-layer stacked outline, and the third comparison result that the grasping state characterization value exceeds the qualified characterization value range, the current grasping state is determined to be grasping multiple objects.

[0013] As a preferred embodiment of the closed-loop judgment method for automatic feeding of textile cut pieces according to the present invention, the method includes: performing corresponding transfer and unloading or closed-loop error correction operations based on the judgment result. When the judgment result is a normal single layer, the transfer and unloading operation is executed; In response to the judgment result of missing the grab, a missing grab error correction operation is performed. The missing grab error correction operation includes adjusting the grab point and the downward speed and re-executing the grab and judgment. If it is judged to be missing again, the remaining amount of the cut piece stack is checked. If the stack is empty, a material replenishment alarm is triggered. If the stack has remaining amount, the downward pressure and the gripper opening are adjusted and the grab and judgment are performed again. In response to the judgment result of "grabbing too many", a "grabbing too many" error correction operation is performed. The "grabbing too many" error correction operation includes maintaining the clamping posture and lifting, using the difference in interlayer friction to perform layer peeling, adjusting the gripper clamping force, opening distance and transfer speed, and re-performing the gripping and judgment. If the "grabbing too many" result is still obtained after the number of layer peelings reaches the preset value, the cut piece is released and reset.

[0014] As a preferred embodiment of the closed-loop judgment method for automatic feeding of textile cut pieces according to the present invention, the following steps are included: recording data of the grasping and error correction process, updating the judgment threshold and grasping parameters of the same type of cut pieces, and performing the feeding cycle: Record and store data on the capture and error correction process, and update the judgment threshold and capture parameters for the same type of cut pieces based on the stored process data; The system counts the frequency of missed or excessive captures within a predetermined batch. When the frequency of the fault exceeds the preset alarm threshold, a maintenance prompt signal is issued. After completing the transfer and unloading of the current cut piece, control the robotic arm to reset to the starting position of the cut piece stack and enter the next round of feeding cycle.

[0015] Secondly, the present invention provides an automatic closed-loop judgment system for feeding textile cut pieces, comprising: The grasping planning module is used to obtain the basic parameters of the cut piece to be grasped, determine the grasping scheme based on the basic parameters and the preset grasping planning model, and set the initial judgment threshold of the sensing unit. The pre-grabbing and data acquisition module is used to perform pre-grabbing actions according to the grasping scheme and acquire multi-source sensing data; The grasping state judgment module is used to judge the multi-source sensing data with the initial judgment threshold to obtain the judgment result of the current grasping state; The graded response and closed-loop judgment module is used to perform corresponding transfer and unloading or closed-loop error correction operations based on the judgment result, record the data of the capture and abnormal error correction process, update the judgment threshold and capture parameters of the same type of cut pieces, and perform the feeding cycle to realize the automatic feeding closed-loop judgment of textile cut pieces.

[0016] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the automatic feeding closed-loop judgment method for textile cut pieces.

[0017] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the automatic feeding closed-loop judgment method for textile cut pieces.

[0018] Compared with existing technologies, the advantages of this invention are as follows: This invention achieves near-high accuracy in layer-by-layer feeding of textile pieces by combining multi-model and multi-sensor fusion judgment with closed-loop error correction, solving the problems of missed or excessive feeding of garment pieces. This invention embeds a self-developed fabric mechanics model, dynamically adapting to all types of fabrics, including knitted, woven, chemical fiber, and lightweight / thick / stiff fabrics, and is compatible with garment pieces of different shapes and weights. This invention uses closed-loop error correction instead of direct machine stoppage / material arrangement, reducing manual intervention and improving the overall utilization rate of automated sewing production lines. This invention has parameter self-learning capabilities, continuously optimizing the feeding and judgment logic over long-term operation, adapting to batch production and flexible production of multiple garment styles. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the overall process logic of a closed-loop judgment method for automatic feeding of textile cut pieces according to an embodiment of the present invention; Figure 2 This is an overall closed-loop control flowchart of an automatic feeding closed-loop judgment method for textile cut pieces provided in one embodiment of the present invention; Figure 3 A diagram showing the shape and gripping point of a textile fabric cutting piece in an embodiment of the present invention. Figure 4 A diagram showing the dropping posture and corner control of an automatic feeding closed-loop judgment method for textile pieces provided in one embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the principle of pre-pressing different gripping points to generate air pressure difference in an automatic feeding closed-loop judgment method for textile pieces according to an embodiment of the present invention, wherein (a) is the gripping center position and (b) is the gripping edge position. Detailed Implementation

[0021] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0022] Example 1, referring to Figures 1-5 As an embodiment of the present invention, a closed-loop judgment method for automatic feeding of textile cut pieces is provided, such as... Figure 2 As shown, it includes: S100: Obtain the basic parameters of the cut piece to be grasped, determine the grasping scheme based on the basic parameters and the preset grasping planning model, and set the initial judgment threshold of the sensing unit. S200: Performs pre-grabbing actions according to the grasping plan and acquires multi-source sensing data; S300: Compare the multi-source sensing data with the initial judgment threshold to obtain the judgment result of the current grasping state; In one optional embodiment, judging the multi-source sensing data with the initial judgment threshold can be done by normalizing the contact force data; extracting edge and contour features from the visual morphology data to form a visual feature vector; normalizing the grasping state representation value; and inputting the normalized contact force, visual feature vector, and grasping state representation value into a pre-trained multi-classification model; the model directly outputs the classification result: one of the following three: grasping empty, normal single layer, or grasping multiple objects. In another optional embodiment, judging the multi-source sensing data with the initial judgment threshold can also be done by mapping the contact force data to a contact force score in the 0-1 interval; mapping the visual morphology data to a visual morphology score in the 0-1 interval; mapping the grasping state representation value to a representation value score in the 0-1 interval; weighting and summing the three scores according to preset weights to obtain a comprehensive grasping score; comparing the comprehensive score with preset grasping empty score intervals, normal single-layer score intervals, and grasping multiple score intervals, and directly outputting one of the grasping states: grasping empty, normal single-layer, or grasping multiple. In this embodiment of the invention, judging the multi-source sensing data with the initial judgment threshold includes comparing the collected contact force data, visual morphology data, and grasping state characterization value with their respective preset initial judgment thresholds to obtain three sets of independent comparison results; combining the three sets of comparison results, three types of grasping states are output: grasping empty, normal single layer, and grasping multiple layers.

[0023] S400: Based on the judgment result, execute the corresponding transfer and unloading or closed-loop error correction operation, record the data of the grabbing and abnormal error correction process, update the judgment threshold and grabbing parameters of the same type of cut piece, and carry out the feeding cycle to realize the automatic feeding closed-loop judgment of textile cut pieces.

[0024] It should be noted that this invention accurately distinguishes between empty feed, normal single-layer feed, and multiple feed states through multi-source sensing fusion judgment; it performs hierarchical closed-loop error correction and adaptive parameter adjustment for anomalies, replacing the traditional stop-and-feed method; and it combines a mechanical model and a self-learning mechanism to adapt to all types of fabrics and continuously optimize thresholds. This achieves high layer-by-layer feeding accuracy and significantly improves the continuity and intelligence level of the sewing production line.

[0025] In this embodiment of the invention, step S100 includes the following sub-steps A1-A3; In A1: Based on the fabric weight and cut area in the basic parameters, and combined with the single-point output force of the soft robotic arm, determine the minimum number of grippers involved in grasping. In A2: Based on the preset shape-preserving gripping rules, determine the gripping points of each gripper at the outer contour boundary and corner intersection of the cut piece; In A3: Based on the minimum number of grippers and the gripping points, a gripping scheme is obtained.

[0026] In the implementation of this invention, the basic parameters of the cut piece to be grasped include the fabric material, namely knitted, woven, nonwoven, weight, thickness, stiffness coefficient, friction factor, and single-layer critical buckling force.

[0027] In one optional embodiment, the minimum number of grippers required for grasping can be determined using a finite element simulation analysis method. During system initialization, the CAD outline of the fabric piece and the constitutive parameters of the fabric are entered. The finite element simulation module is called to perform grasping-transfer simulation on candidate gripper layout combinations, and the maximum deflection of the fabric piece under each number of grippers is calculated. The maximum deformation of the fabric piece obtained from the simulation is compared with the preset process allowable deformation threshold to select the minimum number of grippers that meets the deformation constraint conditions. Combined with the shape-preserving grasping rules, the grasping point layout is completed. The number of grippers is written into the control system and enters the subsequent sensor calibration and pre-grabbing stage. In another optional embodiment, the minimum number of grippers required for grasping can be determined using an orthogonal experimental iterative calibration method. This involves system initialization, inputting fabric parameters and dimensions of the cut pieces, and setting acceptable grasping failure thresholds. Several candidate gripper numbers are selected, an orthogonal experimental table is constructed, and corresponding gripper layouts are configured. Small-batch pre-grabbing tests are conducted sequentially using each group of grippers, and the frequency of failures such as missed grasping, excessive grasping, and excessive cut piece deformation is statistically analyzed for each scheme. The test results are compared to select the minimum number of grippers that meets the failure index requirements. The determined minimum number of grippers and corresponding grasping point configurations are written into the controller. Sensor threshold calibration is completed, and the pre-grabbing and multi-source sensing acquisition process begins. In this embodiment of the invention, determining the minimum number of grippers involved in grasping includes calculating the minimum number of grippers based on the mechanical load calculation formula of fabric weight - cut area - maximum output force of a single gripper.

[0028] Specifically, the minimum number of grippers is automatically calculated based on the multi-point layout model to complete the layout of the gripping points of the soft robotic arm; Minimum number of grippers = [fabric weight * cut piece area / maximum single gripper output force * 0.85] + 1 Among them, fabric weight (g / m) 2 The cut area can be measured or provided by the fabric manufacturer; the cut area can be provided by the CAD pattern making system; the maximum output force of a single claw refers to the maximum output clamping force of the gripper within the adjustable air pressure range, which is given by the gripper model; 0.85 is its ideal / cost-effective working coefficient; [] is the rounding function; +1 is to increase the safety factor for safety reasons.

[0029] The preset shape-preserving gripping rules include mandatory gripping points on the outer contour and corners. Gripping points are set along the outer contour boundary of the cut piece and in the area away from the edge of the contour. Gripping points are also set at the intersection of each corner of the cut piece to ensure that the cut piece maintains a flat shape during gripping and transfer, and to avoid fabric deformation, wrinkles or slippage due to local drooping or uneven force.

[0030] The sensing unit includes a force sensor and a vision unit. The initial judgment thresholds include: the contact force threshold for loose gripping, the contact force range for a single-layer fabric piece, the contact force threshold for multiple-layer fabric pieces, the visual morphology standard value, and the thickness reference value. The initial judgment thresholds can be pre-calibrated and stored in the system database based on mechanical constitutive parameters such as stiffness coefficient, friction factor, and thickness from previous experimental data or theoretical mechanical models.

[0031] After initialization, the robotic arm resets to the preset gripping position of the cut pieces stack, and the system enters standby mode.

[0032] It should be noted that by calculating the number of grippers and laying out the shape-preserving points, and using the perception threshold calibrated by the constitutive parameters, the flatness and stability of the cut pieces are ensured, the risk of deformation and slippage is reduced, and a reliable basis is provided for accurate judgment of the gripping status in the future.

[0033] In this embodiment of the invention, step S200 includes the following sub-steps B1-B4; In B1: The control soft robotic arm moves to the top of the stack of layered cut pieces according to the gripping point and performs a pre-pressing action; In B2: During the pre-pressing and grasping process, the contact force data of the soft robotic arm is collected by force sensors; In B3: Visual morphological data of the captured area of ​​the cut piece is collected through visual units; In B4: Based on the stiffness coefficient, friction factor and thickness change rate in the basic parameters, combined with the downward pressure of the pre-pressing action, the grasping state characterization value of the cut piece is obtained.

[0034] In one optional embodiment, the gripping status characterization value of the cut piece can be obtained by ultrasonic ranging and thickness measurement. The standard single-layer thickness parameters of the fabric are retrieved from the database. An ultrasonic ranging sensor is mounted on the end effector of a soft robotic arm, and sensor calibration is completed. After pre-gripping is completed and the robotic arm lifts the cut piece, the ultrasonic sensor emits ultrasonic waves into the gripping area, collects the return signal, and calculates the actual thickness of the cut piece at the gripping position as the gripping status characterization value. The actual thickness is compared with the thickness of empty material, the single-layer thickness range, and the multi-layer thickness threshold. A visual unit is used for auxiliary verification. The characterization result is output, and the system enters the state judgment branch for empty / normal single-layer / multi-layer gripping. If an anomaly is detected, the corresponding graded closed-loop error correction process is executed. After correction, ultrasonic thickness measurement is repeated to update the gripping status characterization value, and the cycle continues until the gripping is qualified. In another optional embodiment, the gripping state characterization value of the cut piece can also be obtained by a multi-point distributed thin-film pressure array acquisition method. The thin-film pressure array on the flexible finger surface is zero-point calibrated, and the standard pressure map corresponding to single-layer and multi-layer gripping under the fabric is stored. The pre-pressing and pre-gripping action is executed, and the thin-film pressure array collects the pressure data of each gripping point in real time to generate a pressure distribution map. The total pressure and pressure distribution characteristics are used as the gripping state characterization value. The controller compares the measured pressure characterization value with the preset map in the database, and performs auxiliary verification with the gripping coefficient. The gripping state determination of empty gripping, single-layer gripping, and multi-layer gripping is completed. The corresponding closed-loop error correction process is triggered. After each error correction and re-gripping, the pressure array data is re-collected to update the gripping state characterization value. The judgment is iterated until a qualified single-layer cut piece is obtained. In this embodiment of the invention, obtaining the grasping state characterization value of the cut piece includes multi-source fusion acquisition and calculation of force contact force, visual morphology, grasping coefficient and buckling force to obtain the grasping state characterization value.

[0035] Specifically, the controller drives the soft robotic arm downwards to contact the stacked cut pieces and perform a pre-pressing action. The pressing parameters are set based on a single-layer buckling force model. The minimum value must be ensured. It can be greater than the static friction between the top layer and the layer below it, which is also related to the interlayer friction coefficient of the fabric. Yes, it's related; it's the first time the fabric can be arched.

[0036] The contact force data collection includes real-time acquisition of contact force data via force sensors mounted on the end effector of the soft robotic arm during pre-pressing and grasping processes. The contact force data includes the total contact force at the end effector of the soft robotic arm and the component force at each grasping point. The collected contact force data is compared with a preset buckling force range; the minimum value of the preset buckling force range is greater than the static friction force between the topmost and next-level cut pieces. The maximum value is less than This ensures that the contact force remains within a reasonable range during normal single-layer gripping. The visual morphological data includes the visual morphological data of the area of ​​the fabric piece being grasped, acquired by a vision unit during the pre-pressing and grasping process. The vision unit includes at least one industrial camera, positioned above or slightly above the grasping area of ​​the soft robotic arm, with its field of view covering the grasping area of ​​the stacked fabric pieces. The visual morphological data includes the shape of the grasping area, edge curling, and stacking contour. When grasping two or more layers of fabric, the edges of the fabric pieces in the image will exhibit multi-layer stacking characteristics, showing a clear and identifiable difference from images of single-layer fabric pieces; when grasping empty areas, the fabric contour cannot be detected at all. After image preprocessing, edge detection, and contour extraction, the visual morphological data is compared with preset visual morphological benchmark values.

[0037] The grasping status representation value of the cut piece includes the grasping status representation value, which is represented by the grasping coefficient as follows: in, The stiffness coefficient, As the friction factor, This represents the rate of change in fabric thickness under pressure. This refers to the pressure applied by the soft finger.

[0038] Contact force data, visual morphology data, and grasping status representation values ​​are collectively used as multi-source sensing data and uploaded to the controller in real time.

[0039] It should be noted that by integrating multi-source sensing data from force perception, vision, and grasping coefficients, and fully utilizing the fabric's mechanical characteristics, the accuracy of grasping gaps, single layers, and multiple grasps is improved, the probability of fabric misjudgment is reduced, and a reliable basis for closed-loop error correction is provided. In this embodiment of the invention, step S300 includes the following sub-steps C1-C4; In C1: The contact force data is compared with the contact force threshold range in the initial judgment threshold to obtain the first comparison result; In C2: The visual morphology data is compared with the visual morphology baseline value in the initial judgment threshold to obtain the second comparison result; In C3: The capture status representation value is compared with the qualified range of the representation value in the initial judgment threshold to obtain the third comparison result; In C4: Combining the first comparison result, the second comparison result, and the third comparison result, output the current capture status as one of the following: empty capture, normal single layer, or multiple captures.

[0040] In this embodiment of the invention, the contact force threshold range includes an empty gripping threshold, a single-layer cut piece force range, and a multi-layer cut piece force threshold. The empty gripping threshold is used to distinguish whether a cut piece has been gripped. The single-layer cut piece force range corresponds to the contact force range when gripping a single-layer cut piece normally. The multi-layer cut piece force threshold is used to identify situations where the contact force exceeds the limit when gripping multiple cut pieces.

[0041] The visual morphology baseline values ​​include the fabric outline template, single-layer morphological features, and stacked outline features, which correspond to the visual representations in three states: empty, normal single-layer, and multiple-layer. When there is no fabric outline in the visual morphology data, it corresponds to empty; when it presents single-layer edge features, it corresponds to normal single-layer; and when it presents multiple-layer stacked outlines, it corresponds to multiple-layer.

[0042] The acceptable range of the characterization value corresponds to the reasonable range of the characterization value of the crawling state during normal single-layer crawling. When the characterization value approaches zero, it corresponds to empty crawling, and when it exceeds the upper limit of the acceptable range, it corresponds to multiple crawling.

[0043] In this embodiment of the invention, after completing steps C1-C4, step S300 further includes steps C5-C7; In C5: In response to the first comparison result that the contact force data is less than the empty grasp threshold in the contact force threshold range, the second comparison result that the visual morphology data does not match the cut piece outline reference, and the third comparison result that the grasping state characterization value is close to zero, the current grasping state is determined to be empty grasping. In C6: In response to the first comparison result that the contact force data is in the single-layer force range within the contact force threshold range, the second comparison result that the visual morphology data matches the single-layer cut piece morphology benchmark, and the third comparison result that the grasping state characterization value is in the qualified range of characterization value, the current grasping state is determined to be a normal single layer. In C7: When the first comparison result is that the contact force data exceeds the multi-layer force threshold in the contact force threshold range, the second comparison result is that the visual morphology data presents a multi-layer stacked outline, and the third comparison result is that the grasping state characterization value exceeds the qualified characterization value range, the current grasping state is determined to be grasping multiple.

[0044] In this embodiment of the invention, when the contact force is less than the empty grasp threshold + the visual outline of no fabric + the grasping coefficient approaches 0, the current grasping state is determined to be an empty grasp. Triggering scenarios for the empty grasp state include: completion of fabric picking after stacking of cut pieces, grasping point offset, and fabric adhesion loss.

[0045] When the contact force falls within the single-layer cutting force range, the gripping coefficient is within the acceptable range, the visual display shows the single-layer shape, and the fabric buckling force conforms to the single-layer separation model, the current gripping state is determined to be normal single-layer gripping. The actions performed include directly entering the cutting piece transfer, shape-preserving transfer, and precise material placement process to complete a single material loading.

[0046] When the contact force is greater than the multi-layer threshold, the visual appearance of multi-layer stacked outlines, and the interlayer friction exceeds the single-layer critical value, the current grasping state is determined to be grasping multiple layers, that is, more than or equal to 2 layers of cut pieces. This is further subdivided into two-layer grasping and multi-layer grasping, distinguishing between mild adhesion (static electricity / fleece) and severe overlapping.

[0047] It should be noted that by jointly determining the grasping status through multi-dimensional conditions, the misjudgment of a single sensor is avoided, and the grasping of empty, single-layer, and multi-layer working conditions is accurately identified, providing a reliable judgment basis for subsequent hierarchical closed-loop error correction and improving the accuracy of piece grasping.

[0048] In this embodiment of the invention, step S400 includes the following sub-steps D1-D3; In D1: When the judgment result is a normal single layer, a transfer and unloading operation is performed; In D2: In response to the judgment result of missing, a missing error correction operation is performed. The missing error correction operation includes adjusting the gripping point and the downward speed and re-performing the gripping and judgment. If it is judged to be missing again, the remaining amount of the cut piece stack is checked. If the stack is empty, a material replenishment alarm is triggered. If the stack has remaining amount, the downward pressure and the gripper opening are adjusted and the gripping and judgment are performed again. In D3: In response to the judgment result of grabbing too many, the grabbing error correction operation is executed. The grabbing error correction operation includes maintaining the clamping posture and lifting, using the difference in interlayer friction to perform layer peeling, adjusting the gripper clamping force, opening distance and transfer speed and re-executing the grabbing and judgment. If the number of layer peelings reaches the preset value and it is still grabbing too many, the cut piece is released and reset.

[0049] In this embodiment of the invention, when the judgment result is a normal single layer, the cutting piece transfer trajectory, moving speed and posture are controlled according to the shape-preserving grasping strategy.

[0050] like Figure 3 The point allocation strategy and shape-preserving gripping strategy are based on a defined gripping point layout to ensure that the gripping points are evenly stressed during the picking and transfer of the fabric piece, keeping the fabric piece flat overall and preventing curling or deformation of edges and corners due to localized drooping or uneven stress. For example... Figure 4 As shown, the path and speed of movement during the transfer process are determined by the motion control system of the robotic arm. The posture refers to the specific shape of the cut piece during the transfer process, that is, the gripping point is lifted and folded, other positions are in a certain tense state, and the whole piece is transferred or flipped with the robotic arm.

[0051] During the cutting stage, a boot-shaped soft finger is used to perform a reverse pushing action. This means controlling the boot-shaped soft finger to apply a gentle pushing force from the edge of the cut piece towards the center, eliminating wrinkles generated during the grasping process, and ensuring that the cut piece is flat and aligned on the cutting table, making it easier to connect to the next sewing process.

[0052] After the material is unloaded, the robotic arm returns to the starting position for picking up the cut pieces and begins the next feeding cycle.

[0053] When the judgment result is that the grasping point is not grasped, the first round of error correction is performed. The robotic arm is controlled to slightly adjust the grasping point, offsetting it by ±5mm to 10mm from the center of the cut piece stack. At the same time, the downward speed of the soft robotic arm is reduced. Then, the pre-grabbing action is re-executed and multi-source sensing data is collected again for re-judgment.

[0054] like Figure 5 As shown, different production scenarios have different requirements for the gripping cycle time. The descent speed is related to the gripping cycle time; the descent speed is equal to the descent distance divided by the cycle time. When the gripping and separation speed is too fast, due to the air resistance effect of the fabric, an air pressure difference will be generated between the upper and lower layers of fabric at the gripping point. This air pressure difference may cause the gripped piece to fall off the gripping point or cause the next layer of fabric to be lifted and wrinkled, affecting the layer-by-layer separation effect. By reducing the descent speed, the air resistance effect can be effectively weakened, and the gripping stability can be improved.

[0055] The second round of error correction includes, if the first round of error correction determines that the material was not picked up again, initiating visual inspection of the stacking margin to determine whether the material piece stacking has been completely removed: If the visual inspection of the stack determines that it is empty, the system will issue a material replenishment alarm signal and the current round of material feeding will end. If the visual inspection of the stack determines that there is excess material, the downward pressure of the robotic arm and the gripper opening of the soft robotic arm are adjusted to reposition and grasp the material, collect data again, and make a judgment. The downward pressure is provided and adjusted by the force control system of the robotic arm; the gripper opening is provided and adjusted by the air pressure controller. Increasing the air pressure in the chamber of the soft robotic arm increases the opening, and vice versa.

[0056] The third round of error correction includes pausing the robotic arm operation if it is still determined to be a missed grab after two consecutive rounds of error correction, i.e., a total of three missed grabs. This triggers a manual review prompt, allowing the operator to confirm the stacking status and equipment condition. At the same time, the current fault parameters are recorded and stored in the self-learning database for subsequent fault warnings and parameter optimization for similar working conditions.

[0057] The maximum number of iterations for error correction is a preset value, preferably 3 rounds. After each round of error correction, data collection and status judgment are performed again. If the robot arm is still not grasped after reaching the maximum number of iterations, the robot arm is paused and a manual intervention prompt is triggered. If it is determined to be a normal single layer after any round of error correction, the error correction loop is exited and the normal single-layer transfer and unloading operation is started.

[0058] When the judgment result indicates multiple layers are being grasped, the robotic arm maintains its current gripping posture and slightly raises the layer by a preset height, preferably 1cm to 2cm. Utilizing the difference in interlayer friction between the stacked fabric pieces, combined with the fabric's buckling characteristics, the excess bottom layer is passively peeled off under the combined action of gravity and friction. This process is based on a three-stage model of layer-by-layer separation. The three stages are as follows: the compression stage involves the soft robotic arm applying downward pressure, causing a relative misalignment tendency between the upper and lower layers; the arching stage involves the fabric pieces forming localized arches under buckling force, reducing the interlayer adhesion area; and the lifting stage involves the robotic arm lifting the fabric, causing the excess bottom layer pieces to fall back into the stack under gravity. Through this three-stage sequential evolution, the passive separation of multiple layers of fabric pieces is achieved.

[0059] The gripping parameters are dynamically adjusted based on the current fabric type, including reducing the clamping force of the soft robotic arm, optimizing the gripping opening distance, and adjusting the initial transfer speed. Based on key mechanical parameters of commonly used fabrics, such as stiffness and coefficient of friction, corresponding correction values ​​are retrieved from a preset gripping parameter library, or fine-tuned based on historical gripping data of similar fabrics. If the excessive gripping is due to electrostatic adhesion, a weak static electricity elimination module is activated.

[0060] While maintaining the grasping state, collect contact force data and visual morphology data again, and re-determine the number of grasping layers.

[0061] Repeat the layer separation-parameter correction-state judgment process until it is determined to be a normal single layer. Then exit the multi-grab error correction loop and switch to the normal single-layer transfer and unloading operation. If the number of consecutive layer separations reaches the preset value, preferably 3 times, and it is still determined to be multi-grab, it indicates that the fabric is heavily hooked. At this time, the robot releases all the cut pieces and resets. After changing the flexible gripper configuration, the gripping process is re-executed.

[0062] In this embodiment of the invention, after completing steps D1-D3, step S400 also includes steps D4-D6; In D4: Record and store data on the capture and error correction process, and update the judgment threshold and capture parameters for the same type of cut piece based on the stored process data; In D5: Count the frequency of missing or multiple captures within a predetermined batch. When the frequency of the fault exceeds the preset alarm threshold, a maintenance prompt signal is issued. In D6: Complete the transfer and unloading of the current cut piece, control the robot arm to reset to the gripping start position of the cut piece stack, and enter the next round of feeding cycle.

[0063] In this embodiment of the invention, the system records data for each grasping and error correction process. This process data includes fabric material information, grasping force value, grasping point, action speed, error type, and corresponding error correction parameters. After each grasping cycle is completed, the process data is stored in a self-learning database, and the corresponding judgment threshold and grasping parameters are dynamically updated based on accumulated data from similar fabric pieces for subsequent grasping judgments of the same type of fabric pieces.

[0064] The system counts the frequency of missed or excessive grabbing within a predetermined batch. When the frequency of the fault exceeds the preset alarm threshold, the preset alarm threshold is set according to the situation, and a maintenance prompt signal is issued to prompt the operation and maintenance personnel to check the status of the robot arm and sensors or adjust the grabbing strategy.

[0065] Once completed, the robotic arm resets to the stacking and gripping starting position and enters the next feeding cycle, thus realizing a closed-loop judgment for the entire process of automatic feeding of textile pieces.

[0066] Example 2: The above is an illustrative scheme of an automatic textile piece feeding closed-loop judgment method according to this embodiment. It should be noted that the technical solution of this automatic textile piece feeding closed-loop judgment system belongs to the same concept as the technical solution of the above-described automatic textile piece feeding closed-loop judgment method. Details not described in detail in the technical solution of the automatic textile piece feeding closed-loop judgment system in this embodiment can be found in the description of the above-described automatic textile piece feeding closed-loop judgment method.

[0067] This embodiment of a closed-loop judgment system for automatic feeding of textile cut pieces includes: The grasping planning module is used to obtain the basic parameters of the cut piece to be grasped, determine the grasping scheme based on the basic parameters and the preset grasping planning model, and set the initial judgment threshold of the sensing unit. The pre-grabbing and data acquisition module is used to perform pre-grabbing actions according to the grasping scheme and acquire multi-source sensing data; The grasping state judgment module is used to judge the multi-source sensing data with the initial judgment threshold to obtain the judgment result of the current grasping state; The graded response and closed-loop judgment module is used to perform corresponding transfer and unloading or closed-loop error correction operations based on the judgment result, record the data of the capture and abnormal error correction process, update the judgment threshold and capture parameters of the same type of cut pieces, and perform the feeding cycle to realize the automatic feeding closed-loop judgment of textile cut pieces.

[0068] This embodiment also provides a computer device applicable to a closed-loop judgment method for automatic feeding of textile cut pieces, including: The system includes a memory and a processor. The memory stores computer-executable instructions, and the processor executes these instructions to implement a closed-loop judgment method for automatic feeding of textile cut pieces as described in the above embodiments.

[0069] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a closed-loop judgment method for automatic feeding of textile pieces as proposed in the above embodiment.

[0070] The storage medium proposed in this embodiment belongs to the same inventive concept as the closed-loop judgment method for automatic feeding of textile pieces proposed in the above embodiment. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0071] Based on the above description of the implementation methods, those skilled in the art will clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computing device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0072] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A closed-loop judgment method for automatic feeding of textile cut pieces, characterized in that, include: Obtain the basic parameters of the cut piece to be grasped, determine the grasping scheme based on the basic parameters and the preset grasping planning model, and set the initial judgment threshold of the sensing unit. Perform pre-grabbing actions according to the grasping scheme and acquire multi-source sensing data; The multi-source sensing data is compared with the initial judgment threshold to obtain the judgment result of the current grasping state; Based on the judgment result, the corresponding transfer and unloading or closed-loop error correction operation is executed, and the data of the grabbing and abnormal error correction process is recorded. The judgment threshold and grabbing parameters of the same type of cut pieces are updated, and the feeding cycle is carried out to realize the automatic feeding closed-loop judgment of textile cut pieces.

2. The closed-loop judgment method for automatic feeding of textile cut pieces as described in claim 1, characterized in that, Based on the aforementioned basic parameters and the preset crawling planning model, the crawling scheme is determined as follows: Based on the fabric weight and cut area in the basic parameters, and combined with the single-point output force of the soft robotic arm, determine the minimum number of grippers involved in grasping. According to the preset shape-preserving gripping rules, the gripping points of each gripper are determined at the outer contour boundary and corner intersection of the cut piece; Based on the minimum number of grippers and the gripping points, a gripping plan is obtained.

3. The closed-loop judgment method for automatic feeding of textile cut pieces as described in claim 2, characterized in that, Performing pre-grabbing actions according to the grasping scheme and acquiring multi-source sensing data includes: The control software robotic arm moves to the top of the stack of layered cut pieces according to the gripping point and performs a pre-pressing action; During the pre-pressing and grasping process, the contact force data of the soft robotic arm is collected by force sensors; Visual morphological data of the captured area of ​​the cut piece is collected through visual units; Based on the stiffness coefficient, friction factor, and thickness change rate in the basic parameters, and combined with the downward pressure of the pre-pressing action, the grasping state characterization value of the cut piece is obtained.

4. The closed-loop judgment method for automatic feeding of textile cut pieces as described in claim 1 or 3, characterized in that, The judgment result of the current grasping state is obtained by comparing the multi-source sensing data with the initial judgment threshold. The contact force data is compared with the contact force threshold range in the initial judgment threshold to obtain the first comparison result; The visual morphology data is compared with the visual morphology baseline value in the initial judgment threshold to obtain the second comparison result; The capture status representation value is compared with the qualified range of the representation value in the initial judgment threshold to obtain the third comparison result; Based on the combined results of the first, second, and third comparisons, the current capture status is output as one of the following: empty capture, normal single-layer capture, or multiple captures.

5. The closed-loop judgment method for automatic feeding of textile cut pieces as described in claim 4, characterized in that, Based on the combined results of the first, second, and third comparisons, the current capture status is output as one of the following: empty capture, normal single-layer capture, or multiple captures. In response to the first comparison result that the contact force data is less than the empty grasp threshold in the contact force threshold range, the second comparison result that the visual morphology data does not match the cut piece outline reference, and the third comparison result that the grasping state characterization value is close to zero, the current grasping state is determined to be empty grasping. In response to the first comparison result that the contact force data is within the single-layer force range of the contact force threshold range, the second comparison result that the visual morphology data matches the single-layer cut piece morphology benchmark, and the third comparison result that the grasping state characterization value is within the qualified range of characterization value, the current grasping state is determined to be a normal single layer. In response to the first comparison result that the contact force data exceeds the multi-layer force threshold in the contact force threshold range, the second comparison result that the visual morphology data presents a multi-layer stacked outline, and the third comparison result that the grasping state characterization value exceeds the qualified characterization value range, the current grasping state is determined to be grasping multiple objects.

6. The closed-loop judgment method for automatic feeding of textile cut pieces as described in claim 5, characterized in that, Based on the judgment result, the corresponding transfer unloading or closed-loop error correction operation includes: When the judgment result is a normal single layer, the transfer and unloading operation is executed; In response to the judgment result of missing the grab, a missing grab error correction operation is performed. The missing grab error correction operation includes adjusting the grab point and the downward speed and re-executing the grab and judgment. If it is judged to be missing again, the remaining amount of the cut piece stack is checked. If the stack is empty, a material replenishment alarm is triggered. If the stack has remaining amount, the downward pressure and the gripper opening are adjusted and the grab and judgment are performed again. In response to the judgment result of "grabbing too many", a "grabbing too many" error correction operation is performed. The "grabbing too many" error correction operation includes maintaining the clamping posture and lifting, using the difference in interlayer friction to perform layer peeling, adjusting the gripper clamping force, opening distance and transfer speed, and re-performing the gripping and judgment. If the "grabbing too many" result is still obtained after the number of layer peelings reaches the preset value, the cut piece is released and reset.

7. The closed-loop judgment method for automatic feeding of textile cut pieces as described in claim 6, characterized in that, Record data on the data capture and error correction process, update the judgment threshold and capture parameters for similar cut pieces, and perform the material loading cycle, including: Record and store data on the capture and error correction process, and update the judgment threshold and capture parameters for the same type of cut pieces based on the stored process data; The system counts the frequency of missed or excessive captures within a predetermined batch. When the frequency of the fault exceeds the preset alarm threshold, a maintenance prompt signal is issued. After completing the transfer and unloading of the current cut piece, control the robotic arm to reset to the starting position of the cut piece stack and enter the next round of feeding cycle.

8. A closed-loop judgment system for automatic feeding of textile cut pieces, using the closed-loop judgment method for automatic feeding of textile cut pieces as described in any one of claims 1-7, characterized in that, include: The grasping planning module is used to obtain the basic parameters of the cut piece to be grasped, determine the grasping scheme based on the basic parameters and the preset grasping planning model, and set the initial judgment threshold of the sensing unit. The pre-grabbing and data acquisition module is used to perform pre-grabbing actions according to the grasping scheme and acquire multi-source sensing data; The grasping state judgment module is used to judge the multi-source sensing data with the initial judgment threshold to obtain the judgment result of the current grasping state; The graded response and closed-loop judgment module is used to perform corresponding transfer and unloading or closed-loop error correction operations based on the judgment result, record the data of the capture and abnormal error correction process, update the judgment threshold and capture parameters of the same type of cut pieces, and perform the feeding cycle to realize the automatic feeding closed-loop judgment of textile cut pieces.

9. A computer device, characterized in that, include: The system includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the closed-loop judgment method for automatic feeding of textile cut pieces as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by a processor, implements the steps of the automatic feeding closed-loop judgment method for textile cut pieces according to any one of claims 1 to 7.

Images