A flexible yarn pipe automatic replacement control system based on a humanoid robot

Abstract

This invention provides a flexible yarn bobbin automatic replacement control system based on a humanoid robot, relating to the field of textile automation technology. It includes a multi-degree-of-freedom humanoid robot with a flexible actuator mounted at the end of its robotic arm; a multi-modal sensing system integrated at the end of the robotic arm, comprising a binocular depth camera, a structured light projector, and force and torque sensors; receiving and fusing data from the multi-modal sensing system to reconstruct the dynamic pose of the target spindle in real time under vibration conditions, wherein the pose reconstruction frequency is calculated by adapting the vibration frequency to the sensor sampling frequency; based on the reconstructed spindle pose and feedback from the force and torque sensors, the end-effector pose and contact force of the robotic arm are dynamically adjusted by an impedance controller, wherein the stiffness parameter of the impedance controller is determined by calculating the ratio of the yield strength of the yarn bobbin material to the contact area.

Description

Technical Field

[0001] This invention relates to the field of textile automation technology, specifically to a flexible yarn bobbin automatic replacement control system based on a humanoid robot. Background Technology

[0002] In the textile industry, bobbin replacement is a crucial step in ensuring the continuous operation of spinning equipment. Currently, the industry mainly uses manual replacement or traditional dedicated mechanical devices and conventional robotic arm systems to achieve automatic bobbin replacement. For different machine types such as ring spinning, air-jet spinning, and winding machines, existing technologies are mostly designed with dedicated equipment for single machine models. For high-speed air-jet spinning machines (spindle speeds reaching 340,000 rpm), although attempts have been made to determine the spindle position through static positioning to complete bobbin replacement, and some systems are equipped with basic gripping and insertion mechanisms to cope with workshop environment and equipment characteristics, the overall technical solution does not fully consider the need for flexible adaptation under complex working conditions.

[0003] However, existing bobbin replacement technologies have many shortcomings in practical applications, making it difficult to meet the demands of efficient and stable production. In high-speed air-jet spinning machines, the extremely high speed of the spindle causes severe vibrations, resulting in a shift in the actual position of the spindle after static positioning. The dynamic positioning problem remains unsolved. Furthermore, the high-speed rotation of the full bobbin creates a "sticky effect" with the spindle, requiring the initial pull-out force to exceed the holding force, a requirement that the force control capabilities of existing systems cannot meet. In harsh workshop environments with oil and fly waste, the lenses of existing vision systems lack effective protection and are easily contaminated, leading to failures in bobbin or spindle recognition. Additionally, the grippers are prone to slipping when grasping bobbins with oily surfaces. Furthermore, after oil stains alter the friction between the spindle and the inner wall of the yarn tube, the traditional constant force insertion strategy often results in insufficient insertion and damage to the spindle due to insufficient force, or excessive force. To meet the needs of multi-machine compatibility, traditional special-purpose machines need to be mechanically reconfigured for the different spindle heights, diameters, and layouts of different machine types (ring spinning, air-jet spinning, and winding machines), resulting in extremely low production changeover efficiency. In environments with dense spindle positions and limited space, existing robotic arms, due to their limited arm span and lack of flexibility in motion path planning, are prone to collisions with adjacent spindles and yarn tubes, seriously affecting the stability and safety of yarn tube replacement. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a flexible automatic yarn tube replacement control system based on a humanoid robot.

[0005] A flexible yarn tube automatic replacement control system based on a humanoid robot includes a multi-degree-of-freedom humanoid robot with a flexible actuator mounted at the end of its robotic arm;

[0006] A multimodal sensing system, integrated at the end of the robotic arm, includes a visual sensor for perception and a force and torque sensor for force feedback;

[0007] The central processing unit is configured to receive and process sensor data, reconstruct the pose of the target spindle in real time, dynamically adjust the end pose and contact force of the robotic arm through impedance control based on the pose and force feedback, and plan the collision-free motion trajectory of the robotic arm in dense environments.

[0008] An adaptive flexible gripping module, serving as the flexible actuator, includes an inflatable gripper and a coaxially arranged vacuum suction cup;

[0009] A cleaning and condition self-maintenance system, including a cleaning device that acts on the vision sensor.

[0010] Preferably, the central processing unit further includes performing a full-yarn tube extraction operation, specifically:

[0011] The adaptive flexible gripping module is controlled to grip the full yarn tube with a preset force;

[0012] Control the robotic arm to apply an initial pull-out force along the spindle axis;

[0013] After sensing the relative displacement between the yarn tube and the spindle, the pulling force is switched to the holding force, and the yarn tube is unloaded in a spiral motion trajectory.

[0014] Preferably, the inner surface of the inflatable gripper is designed with micro-raised anti-slip textures:

[0015] The height of the anti-slip texture is determined by calculating 1.5 times the surface roughness of the yarn tube:

[0016] The spacing is determined by calculating three times the height of the anti-slip texture:

[0017] The hardness of the gripper material is determined by calculating that the contact stress during yarn gripping does not exceed 1 / 3 of the yarn tube's yield strength.

[0018] The central processing unit controls the clamping force of the grippers through a precision pneumatic pressure regulating valve.

[0019] The lower limit of adjustment is determined by calculating the ratio of the weight of the full yarn tube to the friction coefficient of the gripper.

[0020] Preferably, it also includes an intelligent supply pipe subsystem, which includes:

[0021] Vibration pipe management mechanism:

[0022] Its vibration frequency was determined by calculation that avoided 10% of the yarn tube resonant frequency:

[0023] The amplitude is determined by calculating 1 / 10 of the yarn tube diameter:

[0024] Mobile AGV platform, used to transport yarn tube racks between various textile equipment;

[0025] A docking system based on UWB wireless positioning is used to guide the AGV platform to accurately dock with the humanoid robot. The docking accuracy is determined by superimposing the UWB positioning error and the AGV fine-tuning accuracy.

[0026] The central processing unit can send scheduling instructions to the AGV platform to realize on-demand autonomous replenishment of yarn. The triggering condition for replenishment is determined by calculating 1 / 10 of the maximum capacity of the material rack.

[0027] Preferably, the central processing unit has a built-in equipment process parameter database, which stores the spindle layout diagram, spindle dimensions, and design pull-in / pull-out force curves of the textile equipment.

[0028] The force parameters in the design insertion / extraction force curve are determined by superimposing the viscous effect force of the spindle, the weight of the yarn tube, and the safety factor for the corresponding machine model:

[0029] Before performing the pipe replacement task, the central processing unit receives the equipment model information and automatically calls the corresponding process parameters to configure the template matching threshold of the visual recognition algorithm, the safety distance parameter of the motion planner, and the stiffness and damping parameters of the impedance controller.

[0030] The template matching threshold is determined by calculating the minimum value of image feature similarity.

[0031] Preferably, the cleaning and condition self-maintenance system further includes a self-diagnosis module, which is configured to:

[0032] The sharpness of the images acquired by the binocular depth camera is analyzed periodically. The sharpness is determined by calculating the modulation transfer function, and the value of the modulation transfer function is determined by calculating the ratio of the image edge contrast to the original contrast.

[0033] When the image clarity is below a set threshold, the jet dust removal device is automatically activated for cleaning. The set threshold is determined by calculating the modulation transfer function value corresponding to the minimum effective resolution of image recognition.

[0034] If the clarity is not restored after cleaning, a maintenance alert is sent to the monitoring system.

[0035] Preferably, the self-diagnostic module of the cleaning and condition self-maintenance system is further configured as follows:

[0036] The air pressure decay characteristics of the inflatable gripper are monitored periodically. The air pressure decay rate is determined by dividing the difference between the initial working air pressure and the residual air pressure after 10 minutes of rest by the initial air pressure.

[0037] The zero drift error of the force and torque sensors is monitored synchronously. The zero drift error is determined by the deviation between the average value of 10 consecutive samples taken by the sensor under no-load conditions and the theoretical zero value.

[0038] Preferably, the intelligent tube supply subsystem further includes a yarn tube quality pre-inspection module, which is integrated into the discharge end of the vibrating tube handling mechanism, and includes:

[0039] The visual inspection unit uses an industrial camera to capture images of the outer wall of the empty yarn tube and identifies damaged areas using an edge detection algorithm. The damage determination threshold is determined by calculating 1 / 3 of the yarn tube wall thickness.

[0040] The tension detection unit is used to detect the end tension of the fully loaded yarn bobbin. The tension detection threshold is determined by calculating 1 / 5 of the yarn breaking strength.

[0041] When the yarn tension Ftensile of a full yarn tube is detected to be less than 0.5N, it is determined to be a yarn breakage, and the central processing unit controls the rejection mechanism to remove the unqualified yarn tube.

[0042] Preferably, the multi-machine collaborative scheduling module is integrated into the central processing unit and is configured as follows:

[0043] The system receives the task status signal of the humanoid robot and the position signals of multiple mobile AGV platforms, and dynamically allocates resources based on task priority. The task priority is determined by multiplying the number of full yarn tubes of the textile equipment by the equipment downtime loss coefficient.

[0044] A collision-free transport path is planned for multiple AGV platforms, and the safe distance of the path is determined by calculation based on the coupling relationship between the width of the AGV platform and its maximum travel speed.

[0045] This invention provides a flexible automatic yarn bobbin changing control system based on a humanoid robot, which has the following advantages:

[0046] The multimodal sensing system addresses the problem of tube removal failure caused by static positioning offset and insufficient force control under high-speed vibration, significantly improving the success rate of yarn tube insertion and removal. The air jet dust removal device of the cleaning and self-maintenance system solves the recognition failure problem caused by lens contamination of the binocular depth camera. The adaptive flexible gripping module with inflatable grippers featuring anti-slip texture and preset gripping force adapted to oily environments enhances the anti-slip gripping ability of oily yarn tubes. At the same time, the dynamic insertion force strategy with adaptive impedance control solves the problem of yarn tube insertion force adaptation in oily environments.

[0047] The central processing unit has a built-in database of process parameters for multiple machine models. When switching models, there is no need to reconstruct the mechanical structure. It only needs to call the process parameters of the corresponding equipment and automatically configure the template matching threshold of the visual recognition algorithm, the safety distance parameters of the motion planner, and the stiffness and damping parameters of the impedance controller. This significantly shortens the machine model switching cycle and solves the problem of low production changeover efficiency of traditional dedicated machines. At the same time, the central processing unit calculates the motion safety distance adapted to the dense spindle environment by combining the minimum spacing between spindles, the size of the robotic arm end effector, and the dynamic compensation coefficient, and then plans the collision-free motion trajectory of the robotic arm. Combined with the flexible motion characteristics of the multi-degree-of-freedom robotic arm, collisions with adjacent spindles and yarn tubes in the dense spindle space are avoided, improving the stability and safety of yarn tube replacement and reducing the equipment collision failure rate. Attached Figure Description

[0048] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] like Figure 1 As shown, the present invention proposes a flexible yarn tube automatic replacement control system based on a humanoid robot, including a multi-degree-of-freedom humanoid robot, the end of which is equipped with a flexible actuator;

[0051] A multimodal sensing system, integrated at the end of the robotic arm, includes a visual sensor for perception and a force and torque sensor for force feedback;

[0052] The central processing unit is configured to receive and process sensor data, reconstruct the pose of the target spindle in real time, dynamically adjust the end pose and contact force of the robotic arm through impedance control based on the pose and force feedback, and plan the collision-free motion trajectory of the robotic arm in dense environments.

[0053] An adaptive flexible gripping module, serving as the flexible actuator, includes an inflatable gripper and a coaxially arranged vacuum suction cup;

[0054] The adaptive flexible gripping module's inflatable gripper and vacuum suction cup work in a collaborative mode: when gripping yarn tubes with no oil stains on the surface, only the inflatable gripper is activated; when gripping yarn tubes with oil stains on the surface, the inflatable gripper activates the vacuum suction cup while holding the yarn tube, with the vacuum level set to -0.08MPa. The gripping reliability is enhanced through negative pressure adsorption. The opening and closing of the vacuum suction cup is controlled by the central processing unit based on the oil stain status on the yarn tube surface identified by the visual sensor.

[0055] A cleaning and condition self-maintenance system, including a cleaning device that acts on the vision sensor;

[0056] In this embodiment, the specific steps are as follows:

[0057] A flexible actuator is mounted at the end of a multi-degree-of-freedom humanoid robot arm;

[0058] A multimodal sensing system, integrated at the end of the robotic arm, includes a binocular depth camera, a structured light projector, and force and torque sensors;

[0059] Central processing unit, configured for:

[0060] (a) Receive and fuse data from the multimodal sensing system to reconstruct the dynamic pose of the target spindle in real time under vibration conditions, wherein the pose reconstruction frequency is calculated by adapting the vibration frequency to the sensor sampling frequency:

[0061] (b) Based on the reconstructed spindle pose and feedback from the force and torque sensors, the end-effector pose and contact force of the robotic arm are dynamically adjusted via an impedance controller. The stiffness parameter of the impedance controller is determined by calculating the ratio of the yield strength of the yarn tube material to the contact area.

[0062] It should be noted that the yarn tube is typically made of ABS with a yield strength σs = 45 MPa, and the contact area between the clamps and the yarn tube is S = 100 mm². 2 =1×10 -4 m 2 From stiffness K = σs × S / Δx (Δx is the maximum allowable deformation of 0.1 mm), we get K = (45 × 10 6 Pa×1×10 -4 m 2 ) / 0.0001m=4.5×10 4 N / m=45N / mm, suitable for yarn tube contact force control requirements;

[0063] (c) Plan the collision-free motion trajectory of the robotic arm in a dense spindle environment, wherein the safe distance of the trajectory is determined by superimposing a dynamic compensation coefficient on half of the difference between the minimum spacing between spindles and the end dimension of the robotic arm;

[0064] An adaptive flexible gripping module, including an inflatable gripper and a vacuum suction cup coaxial with the gripper;

[0065] A cleaning and condition self-maintenance system, including an air jet dust removal device, is applied to the binocular depth camera lens.

[0066] As an optional embodiment: the central processing unit further includes performing a full-bore extraction operation, specifically:

[0067] The adaptive flexible gripping module is controlled to grip the full yarn tube with a preset force;

[0068] Control the robotic arm to apply an initial pull-out force along the spindle axis;

[0069] After sensing the relative displacement between the yarn tube and the spindle, the pulling force is switched to the holding force, and the yarn tube is removed in a spiral motion trajectory.

[0070] In this embodiment, the specific steps are as follows:

[0071] The adaptive flexible gripping module is controlled to grip the full yarn tube with a preset force. The preset force is calculated by multiplying the ratio of the weight of the full yarn tube to the friction coefficient of the gripper by a safety factor, which is in the range of 1.0 to 1.4.

[0072] It should be noted that the normal mass of the full yarn tube is m_full = 150g = 0.15kg, and the gravity G = m_full × g = 0.15 × 9.8 = 1.47N; the coefficient of friction between the gripper and the ABS yarn tube in an oily environment is μ = 0.3; the basic gripping force F0 = G / μ = 1.47 / 0.3 = 4.9N; multiplying by a safety factor of 1.2, we get the preset force F_grip1 = 4.9 × 1.2 = 5.88N (rounded to 6N).

[0073] The robotic arm is controlled to apply an initial pull-out force along the spindle axis. This initial pull-out force is calculated by summing the viscous effect force and the holding force by a first coefficient multiple, where the first coefficient ranges from 1.2 to 1.5.

[0074] The viscous effect force is determined by the product of the dynamic friction coefficient of the high-speed rotating spindle, the spindle speed, and the contact area between the yarn tube and the spindle rod.

[0075] The dynamic friction coefficient μviscosity = 0.25 (mean value of 30 groups of rotating friction experiments in oily environments), and the spindle rotation speed n = 4 × 102 4 r / min = 666.7 r / s, contact area between yarn bobbin and spindle S = π × 30 mm × 80 mm = 7536 mm 2 =7.536×10 -3 m 2 Viscous effect force Fviscous = μviscous × n × S × k (k = 1.2 is the speed correction coefficient, based on 55 × 102). 4(Empirical fitting within the r / min speed range) yielded Fviscosity = 0.25 × 666.7 × 7.536 × 10-10 -3 ×1.2≈1.81N;

[0076] The holding force is determined by multiplying the weight of the full yarn tube by a safety factor of 1.5: F_holding = G × 1.5 = 1.47 × 1.5 ≈ 2.21 N (the 1.5 factor is to overcome the dynamic disturbance during the pulling process, verified by 200 pulling tests).

[0077] The initial pull-out force Finitial = Fviscous + 1.3 × Fmaintain = 1.81 + 1.3 × 2.21 ≈ 4.68 N (rounded to 5 N) is used to overcome the viscous effect of the high-speed rotating spindle.

[0078] After sensing the relative displacement between the yarn tube and the spindle, the pulling force is switched to the holding force, and the yarn tube is smoothly unloaded using a spiral motion trajectory. The pitch of the spiral motion trajectory is determined by calculating half the lead of the thread on the inner wall of the yarn tube. The conventional thread lead on the inner wall of the yarn tube is 4mm, and half (2mm) is taken to avoid thread jamming. The rotational angular velocity is determined by calculating the ratio of the maximum linear velocity at the end of the robotic arm to the radius of the yarn tube: the maximum linear velocity at the end of the robotic arm v=50mm / s, the radius of the yarn tube r=15mm, and the angular velocity ω=v / r=50 / 15≈3.33rad / s (taken as πrad / s).

[0079] As an optional embodiment: the central processing unit is configured to perform the insertion of an empty yarn tube as follows:

[0080] The robotic arm is controlled to guide the empty yarn tube above the spindle, and coarse positioning is performed based on visual servoing. The accuracy of the coarse positioning is determined by superimposing the ranging error of the binocular depth camera and the accuracy of the structured light projector: binocular depth camera ranging error ≤ 0.5 mm, structured light projector accuracy ≤ 0.3 mm, and superimposed error ≤ √(0.5 mm). 2 +0.3 2 )≈0.58mm (take ±1mm);

[0081] Initiate the force control search strategy: Control the robotic arm to descend along the spindle axis at a downward velocity, which is calculated and determined through the coupling relationship between the kinetic energy threshold of the empty yarn tube and the mass and gravitational potential energy of the empty yarn tube.

[0082] The kinetic energy threshold is determined by the energy value corresponding to the elastic deformation limit of the hollow yarn tube material: the elastic deformation limit of the hollow yarn tube ABS material ε=0.002, the elastic modulus E=2.3GPa, and the volume V=π×15. 2 ×80=56548mm 3 =5.6548×110 -5 m 3 Deformation energy threshold E = 0.5 × E × ε2 ×V=0.5×2.3×10 9 Pa×(0.002) 2 ×5.6548×10 -5 m 3 ≈0.001J;

[0083] Because the downward speed value is too high, in actual applications, the downward speed is limited to within 50mm / s by the damping mechanism at the end of the robotic arm to prevent the empty yarn tube from hitting the spindle.

[0084] The mass of the empty yarn tube is m_empty = 30g = 0.03kg, the downward probe distance is h = 5mm (initial gap), and the gravitational potential energy is Ep = m_empty × g × h = 0.03 × 9.8 × 0.005 ≈ 0.00147J. From the total energy E_total = E_threshold + Ep = 0.001 + 0.00147 = 0.00247J = 0.5 × m_empty × v_probe², we get v_probe = √(2 × 0.00247 / 0.03) ≈ 12.8mm / s (take 13mm / s).

[0085] Simultaneously, the contact force is monitored using the force and torque sensors;

[0086] When the contact force is detected to exceed the threshold for a preset time, it indicates that the bottom of the yarn tube has contacted the spindle disc:

[0087] The threshold is determined by the sum of the weight of the empty yarn tube and the additional resistance of the oil stain: the weight of the empty yarn tube G_empty = m_empty × g = 0.03 × 9.8 = 0.294 N, the additional resistance of the oil stain F_resistance = 1.2 N (mean value of 50 oil stain environment insertion experiments), and the threshold F_threshold = 0.294 + 1.2 ≈ 1.494 N (take 1.5 N).

[0088] The preset time is determined by the ratio of the descent speed to the yarn tube wall thickness: yarn tube wall thickness t = 3mm, preset time t_set = t / v_exploration = 3 / 13 ≈ 0.23s (take 0.2s).

[0089] Trigger fine insertion sequence;

[0090] In the fine insertion sequence, an adaptive impedance control model based on force feedback is used to dynamically adjust the downward insertion force and the radial correction force:

[0091] The downward insertion force is calculated by the sum of the basic insertion force and the friction compensation: Basic insertion force Fbase = Gempty × 1.5 = 0.294 × second coefficient ≈ 0.44 N (the second coefficient is to overcome the initial insertion resistance, and the specific value range is 1.3 to 1.7); Friction compensation ΔF = |μactualμstandard| × Fpositive (μactual is the actual friction coefficient, measured by the force sensor; μstandard = 0.15 is the standard friction coefficient without oil contamination; Fpositive is the normal force of the gripper, taken as 3 N), therefore ΔF = |μactual0.15| × 3;

[0092] The radial correction force is determined by multiplying the radial deviation by the stiffness coefficient: the radial deviation Δr is detected visually, and the stiffness coefficient K_radius = 0.5 N / mm (fitted from 20 sets of deviation compensation experiments), therefore F_radius = K_radius × Δr;

[0093] This is to compensate for the changes in friction and coaxiality deviation caused by oil contamination until the yarn tube is installed in place.

[0094] As an optional embodiment: the inner surface of the inflatable gripper is designed with slightly raised anti-slip textures.

[0095] The height of the anti-slip texture is determined by calculating 1.5 times the surface roughness of the yarn tube: the surface roughness of the ABS yarn tube Ra=0.3μm, and the height h=1.5×0.3=0.45μm (take 0.5μm).

[0096] The spacing is determined by calculating 3 times the height of the anti-slip texture: spacing d = 3 × 0.5 = 1.5 μm (take 2 μm), which is suitable for the friction requirements in oily environments;

[0097] The material hardness of the grippers is determined by calculating that the contact stress during yarn tube gripping does not exceed 1 / 3 of the yarn tube's yield strength: yarn tube yield strength σs = 45 MPa, allowable contact stress σ_allowable = 45 / 3 = 15 MPa; gripper contact area S = 100 mm². 2 The maximum permissible clamping force Fmax = σ_permissible × S = 15 × 10 6 Pa×1×10 -4 m 2 =150N; the material is Shore A2530 (HG / T28152011), and its fatigue limit is 30N (verified by 1000 clamping tests, there is no permanent deformation at this hardness). Therefore, the upper limit of the actual clamping force is 30N, taking into account both flexibility and anti-slip properties.

[0098] The central processing unit controls the clamping force of the grippers through a precision pneumatic pressure regulating valve.

[0099] The lower limit of adjustment is determined by calculating the ratio of the weight of the full yarn tube to the friction coefficient of the gripper: F_lower = G / μ = 1.47 / 0.3 = 4.9N (take 5N);

[0100] The upper limit for adjustment is set at 30N based on the fatigue limit.

[0101] The method is determined based on the strength of the yarn tube material and the reliability of gripping, so as to achieve reliable gripping of yarn tubes with oil stains on the surface and avoid pinching.

[0102] As an optional embodiment: Vibration tube handling mechanism:

[0103] Its vibration frequency is determined by avoiding the resonance frequency of the yarn tube by 10%: yarn tube stiffness k=500N / m (actual measurement of ABS material), mass m_empty=0.03kg, resonance frequency fres≈20.6Hz. In order to avoid resonance, the vibration frequency is taken to be below 90% or above 110% of the resonance frequency, that is, the safe frequency range is ≤18.5Hz or ≥22.7Hz. In practical applications, 22.7Hz to 50Hz is used.

[0104] The amplitude is determined by calculating 1 / 10 of the yarn tube diameter: the standard diameter of the yarn tube is 30mm, and the amplitude A = 30 / 10 = 3mm (take 13mm), which is determined based on the diameter, material and resonance characteristics of the yarn tube;

[0105] Mobile AGV platform, used to transport yarn tube racks between various textile equipment;

[0106] A docking system based on UWB wireless positioning is used to guide the AGV platform to accurately dock with the humanoid robot. The docking accuracy is determined by superimposing the UWB positioning error and the AGV fine-tuning accuracy: UWB positioning error ≤ 50mm, AGV fine-tuning accuracy ≤ 10mm, and superimposed error ≤ √(50mm). 2 +10 2 )≈51mm (after two fine adjustments, it reached ±10mm, verified by 30 docking experiments).

[0107] The docking system based on UWB wireless positioning consists of two stages: coarse positioning and secondary fine-tuning. Coarse positioning is achieved by UWB with an error of ≤51mm. Secondary fine-tuning uses a binocular depth camera at the end of the humanoid robot arm to identify the positioning QR code on the AGV platform. The central processing unit controls the stepper motor of the AGV platform to perform position compensation based on the QR code recognition result, with a compensation accuracy of ±0.5mm, ultimately achieving a docking accuracy of ±10mm.

[0108] The central processing unit can send scheduling instructions to the AGV platform to achieve on-demand autonomous replenishment of yarn. The triggering condition for replenishment is determined by calculating 1 / 10 of the maximum capacity of the material rack. The maximum capacity of the material rack is 100 yarns, and the triggering threshold is 100 × 1 / 10 = 10 yarns, which is determined based on the number of remaining yarn tubes in the material rack.

[0109] As an optional embodiment: the central processing unit has a built-in equipment process parameter database, which stores the spindle layout diagram, spindle dimensions, and design pull-out force curve of the textile equipment.

[0110] The force parameters in the designed insertion / extraction force curve are determined by superimposing the viscous effect force of the spindle, the weight of the yarn bobbin, and the safety factor for the corresponding machine model: for example, if the spindle speed of ring spinning is 15,000 rpm, the viscous effect force Fviscous' = 0.25 × 250 × 7.536 × 102 -3 ×1.2≈0.56N, holding force F_hold'=1.47×1.5≈2.21N, initial pull-out force F_initial'=0.56+1.3×2.21≈3.43N (verified by actual measurement on 10 ring spinning machines).

[0111] Before performing the pipe replacement task, the central processing unit receives the equipment model information and automatically calls the corresponding process parameters to configure the template matching threshold of the visual recognition algorithm, the safety distance parameter of the motion planner, and the stiffness and damping parameters of the impedance controller.

[0112] The template matching threshold is determined by calculating the minimum image feature similarity. The minimum image feature extraction matching degree is ≥85% (tested with 500 spindle images of different models, achieving a matching accuracy of over 85% and a correctness rate of 99.2%), therefore the threshold is set to 85%.

[0113] It enables switching between different models without the need for mechanical reconstruction.

[0114] As an optional embodiment: the cleaning and condition self-maintenance system further includes a self-diagnostic module, which is configured to:

[0115] The sharpness of the images acquired by the binocular depth camera is analyzed periodically. The sharpness is determined by the modulation transfer function (MTF). The MTF value is determined by the ratio of the image edge contrast to the original contrast: original contrast C0 = 255 (difference between black and white pixels), acquired image edge contrast C = 76.5, MTF = C / C0 = 0.3.

[0116] When the image clarity is lower than a set threshold, the jet dust removal device is automatically activated for cleaning. The set threshold is determined by calculating the modulation transfer function value corresponding to the minimum effective resolution of image recognition: the minimum effective resolution of image recognition is 0.1mm, which corresponds to an MTF value of 0.3, so the threshold is set to 0.3.

[0117] If the clarity is not restored after cleaning, a maintenance alert is sent to the monitoring system.

[0118] As an optional embodiment: the self-diagnostic module of the cleaning and condition self-maintenance system is further configured to:

[0119] The air pressure decay characteristics of the inflatable gripper are monitored periodically. The air pressure decay rate is determined by dividing the difference between the initial working air pressure and the residual air pressure after 10 minutes of rest by the initial air pressure.

[0120] The initial working air pressure P0 = 0.3MPa (suitable for gripping force requirements of grippers with a Shore A hardness of 25-30), and the maximum allowable attenuation rate α_allow = 5% (verified by 100 continuous gripping experiments, when the attenuation rate is ≤5%, the gripping force fluctuation is ≤0.5N, which meets the gripping accuracy); when the measured attenuation rate α_actual = (P0 - P_residual) / P0 > 5%, the air pressure compensation or gripper replacement warning will be automatically triggered.

[0121] The zero-drift error of the force and torque sensors is monitored synchronously. The zero-drift error is determined by calculating the deviation between the average value of 10 consecutive samples taken by the sensor under no-load conditions and the theoretical zero value.

[0122] The theoretical zero value Fzero = 0N, and the allowable zero drift error ΔFallowable = ±0.1N. When ΔFactual > ±0.1N, the sensor's automatic calibration program is started. The standard weight used for calibration has a mass mstandard = 1kg (generating a standard force Fstandard = 9.8N). The measured value is corrected through a deviation compensation algorithm to ensure the accuracy of force feedback.

[0123] As an optional embodiment: the intelligent tube supply subsystem further includes a yarn tube quality pre-inspection module, which is integrated into the discharge end of the vibrating tube handling mechanism, and includes:

[0124] The visual inspection unit uses an industrial camera to capture images of the outer wall of the empty yarn tube and identifies damaged areas using an edge detection algorithm. The damage determination threshold is determined by calculating 1 / 3 of the yarn tube wall thickness.

[0125] The unified damage assessment criteria are depth and area. It is further supplemented that "the damage assessment of the yarn tube adopts a dual threshold of depth and area: the damage depth does not exceed 1 / 3 (1 mm) of the yarn tube wall thickness, and the area of ​​a single damage item does not exceed 3.14 mm." 2 At the same time, the total damaged area shall not exceed 5% of the total area of ​​the outer wall of the yarn tube. If any threshold is exceeded, the yarn tube shall be judged as unqualified.

[0126] The standard wall thickness of the empty yarn tube is t=3mm, the maximum allowable damage depth is t_damaged = t / 3 = 1mm, and the maximum allowable damage area is S_damaged = π × (1mm). 2 =3.14mm 2 (Verification through an installation experiment with 300 damaged yarn tubes showed that when the damage exceeded this threshold, the insertion failure rate was ≥15%).

[0127] The tension detection unit is used to detect the end tension of the fully loaded yarn bobbin. The tension detection threshold is determined by calculating 1 / 5 of the yarn breaking strength.

[0128] The conventional yarn breaking strength F_break = 2.5 N, and the allowable minimum tension F_allow = F_break / 5 = 0.5 N; when it is detected that the actual yarn tension F_actual of the full yarn tube < 0.5 N, it is determined that the yarn is broken, and the central processing unit controls the rejection mechanism to remove the unqualified yarn tube.

[0129] Among them, the detection efficiency of the quality pre-inspection module is adapted to the discharging speed of the vibrating yarn tube sorting mechanism, and the detection time consumption ≤ 0.2 s / root (the discharging speed of the vibrating yarn tube sorting mechanism is 5 roots / s, and synchronous detection is realized through a parallel image processing algorithm).

[0130] As an optional embodiment, the multi-robot collaborative scheduling module is integrated into the central processing unit and is configured as:

[0131] Receive the task status signal of the humanoid robot and the position signals of multiple mobile AGV platforms, and dynamically allocate resources based on the task priority. The task priority is calculated and determined by the product of the number of full yarn tubes of the textile equipment and the equipment downtime loss coefficient;

[0132] The equipment downtime loss coefficient k_loss = 1.2 (the high-speed air-jet spinning machine loses about 20 yuan per minute of downtime, and for ring spinning, k_loss = 0.8), the number of full yarn tubes n_full, and the priority P = n_full × k_loss; when P ≥ 6 (for example, when n_full = 5 and k_loss = 1.2, P = 6), trigger multi-robot collaborative tube replacement;

[0133] The equipment downtime loss coefficient k_loss is the ratio of the economic loss per minute of downtime of the textile equipment to the benchmark loss (10 yuan / minute). The high-speed air-jet spinning machine loses about 20 yuan per minute of downtime, so k_loss = 20 / 10 = 1.2; the ring spinning machine loses about 8 yuan per minute of downtime, so k_loss = 8 / 10 = 0.8; the winding machine loses about 15 yuan per minute of downtime, so k_loss = 1.5

[0134] Plan the collision-free transportation paths of multiple AGV platforms. The path safety distance is calculated and determined by the coupling relationship between the width of the AGV platform and the maximum driving speed.

[0135] The width of the AGV platform W = 500 mm, the maximum driving speed v_AGV = 0.5 m / s, and the safety distance d_safe = W / 2 + v_AGV × 100 mm·s / m = 250 mm + 50 mm = 300 mm (100 mm·s / m is the speed compensation coefficient. After 20 multi-AGV intersection experiments, the collision-free probability at this distance is 100%);

[0136] Establish communication with the PLC control system of the textile equipment, receive the spindle start-stop signals of the equipment, and send the equipment start permission signal after the tube replacement is completed to achieve seamless connection between the tube replacement process and the textile production process.

[0137] Working principle:

[0138] S1. The multimodal sensing system is used to locate and track the dynamic pose of the target spindle in real time under vibration environment. The positioning frequency is determined by calculating 10 times the vibration frequency: the spindle vibration frequency is 50Hz, the positioning frequency = 50 × 10 = 500Hz, which is adapted to the pose change rate under vibration environment.

[0139] S2. Plan the collision-free motion trajectory of the robotic arm, move the adaptive flexible gripping module above the full yarn tube, and the planning of the motion trajectory is determined by the A* algorithm combined with the spatial distribution of dense spindles: spindle coordinates (xi, yi), robotic arm kinematic parameters (joint angle range, arm length), and the minimum path of the cost function is searched by the A* algorithm. The spatial distribution parameters are determined by superimposing the spindle coordinates and the robotic arm kinematic parameters: motion range = arm length + spindle spacing, which is adapted to the spatial distribution of dense spindles;

[0140] S3. Combining visual and force feedback, the fully loaded yarn tube is pulled out and moved to a designated container in a compliant control manner. The force control parameters of the compliant control are calculated and determined by the initial pull-out force (5N) and the holding force (2.21N).

[0141] S4. Obtain an empty yarn tube from the tube supply subsystem, and again combine visual and force feedback to accurately insert the empty yarn tube into the target spindle using force control search and adaptive impedance strategy. The parameters of the force control search are calculated and determined by the downward probe speed (13mm / s) and the contact force threshold (1.5N).

[0142] S5. Throughout the process, the cleaning and self-maintenance system ensures the continuous and reliable operation of the sensing system. The self-maintenance cycle is determined by the ratio of lens contamination rate to cleaning effect duration: lens contamination rate vcontamination = 0.1MTF / 60s (experimentally measured), cleaning effect duration teffect = 120s (the time after cleaning when MTF is maintained above 0.3), and safety margin coefficient k = 2.

[0143] The maintenance cycle calculation process is as follows: using the effective runtime as the base, first calculate the ratio of the MTF threshold to the MTF critical value, then multiply this ratio by the effective runtime, and finally divide the resulting product by the lens contamination rate to obtain the self-maintenance cycle. Here, the MTF threshold is set to 0.3, which is the minimum effective MTF value required for image recognition; the MTF critical value is set to 0.2, which is the minimum MTF value to trigger camera lens cleaning operations; and the lens contamination rate is a decrease of 0.1 in the MTF value every 60 seconds. Substituting the parameters into the calculation, with an effective runtime of 120 seconds, first calculate the ratio of 0.3 to 0.2 to 1.5, then multiply 1.5 by 120 to get 180, and finally divide 180 by 0.1 to get 1200 seconds, meaning that a camera lens sharpness check is performed every 1200 seconds.

[0144] S41. During the insertion phase, the robotic arm is controlled to descend along the spindle axis at a first speed, which is calculated and determined by the descent speed (13 mm / s), while the contact force in the Z-axis direction is continuously monitored.

[0145] S42. If the contact force is less than the threshold, which is determined by calculation using the contact force threshold (1.5N), then the downward probe continues;

[0146] S43. If the contact force increases sharply and exceeds the threshold in a short period of time, the short period of time is determined by the ratio of the contact force change rate to the yarn tube deformation response time: the contact force change rate is ≥50N / s (determined as a collision), the yarn tube deformation response time is 0.01s, and the short period of time is 0.01×(50 / 1.5)≈0.33s (taken as 0.3s). It is determined that a radial deviation collision has occurred, and the robotic arm is controlled to move a small distance in the opposite direction of the collision force to compensate. The small distance is determined by 1 / 3 of the maximum estimated radial deviation: the maximum estimated radial deviation is 1mm, and the small distance is 1 / 3≈0.33mm (taken as 0.3mm). Then, the probe is tried again.

[0147] S44. If the contact force steadily reaches the threshold, it is determined that the bottom of the yarn tube has contacted the spindle. Switch to the impedance control mode and apply a constant insertion force until the yarn tube is installed in place. The constant insertion force is calculated and determined by the downward insertion force (0.44N+ΔF).

[0148] Before removing the full yarn tube, add a pre-loosening step:

[0149] The robotic arm is controlled to clamp the yarn tube. The clamping force is determined by the ratio of the viscous force between the yarn tube and the spindle to the friction coefficient of the gripper: viscous force F_viscous = 1.81N, friction coefficient 0.3, clamping force F_clamping = 1.81 / 0.3 ≈ 6.03N (take 6N), which is determined based on the degree of adhesion between the yarn tube and the spindle.

[0150] A small reciprocating rotational torque is applied around the spindle axis. The magnitude of the rotational torque is determined by multiplying the static friction torque between the yarn tube and the spindle by a safety factor of 1.2.

[0151] The static friction torque M_static = μ_static × F_clamp × r_spin (μ_static = 0.4, F_clamp = 6 N, r_spin = 15 mm = 0.015 m) = 0.4 × 6 × 0.015 = 0.036 N·m;

[0152] The rotational torque M_rotation = 0.036 × 1.2 = 0.0432 N·m (take 0.04 N·m), which is compatible with the static friction torque of the yarn tube and the spindle.

[0153] The angle of the reciprocating rotation is determined by calculating half of the yarn tube winding fly waste fracture limit angle: the fracture limit angle is 10° (measured in 100 fly waste winding experiments), angle = 10 / 2 = 5°; the number of cycles is determined by calculating the number of cycles required to break the viscosity effect: 3 cycles can break the viscosity (80 pull-out experiments verified a success rate of 98%), which is determined based on the need to avoid yarn tube winding fly waste fracture, to break the viscosity effect caused by oil or high-speed rotation, and then perform the axial pull-out action.

[0154] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that many variations, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A flexible yarn bobbin automatic changing control system based on a humanoid robot, characterized in that, This includes multi-degree-of-freedom humanoid robots, whose robotic arms are equipped with flexible actuators at the end. A multimodal sensing system, integrated at the end of the robotic arm, includes a visual sensor for perception and a force and torque sensor for force feedback; The central processing unit is configured to receive and process sensor data, reconstruct the pose of the target spindle in real time, dynamically adjust the end pose and contact force of the robotic arm through impedance control based on the pose and force feedback, and plan the collision-free motion trajectory of the robotic arm in dense environments. The adaptive flexible gripping module includes an inflatable gripper and a coaxially arranged vacuum suction cup.

2. The flexible yarn bobbin automatic replacement control system based on a humanoid robot according to claim 1, characterized in that, The central processing unit also includes performing a full-yarn tube extraction operation, specifically: The adaptive flexible gripping module is controlled to grip the full yarn tube with a preset force; Control the robotic arm to apply an initial pull-out force along the spindle axis; After sensing the relative displacement between the yarn tube and the spindle, the pulling force is switched to the holding force, and the yarn tube is unloaded in a spiral motion trajectory.

3. The flexible yarn bobbin automatic replacement control system based on a humanoid robot according to claim 1, characterized in that, The central processing unit also includes performing the insertion of an empty yarn tube, specifically: The robotic arm is controlled to guide the empty yarn tube above the spindle, and coarse positioning is performed based on visual servoing. The accuracy of the coarse positioning is determined by superimposing the ranging error of the binocular depth camera and the accuracy of the structured light projector. Initiate the force control search strategy: Control the robotic arm to descend along the spindle axis at a downward velocity, which is calculated and determined through the coupling relationship between the kinetic energy threshold of the empty yarn tube and the mass and gravitational potential energy of the empty yarn tube. The kinetic energy threshold is determined by the energy value corresponding to the elastic deformation limit of the hollow yarn tube material: Simultaneously, the contact force is monitored using the force and torque sensors; When the contact force is detected to exceed the threshold for a preset time, it indicates that the bottom of the yarn tube has contacted the spindle disc: The threshold is determined by the sum of the weight of the empty yarn tube and the additional resistance from the oil stains: Trigger fine insertion sequence; In the fine insertion sequence, an adaptive impedance control model based on force feedback is used to dynamically adjust the downward insertion force and the radial correction force: The downward insertion force is determined by the sum of the basic insertion force and the frictional compensation.

4. The flexible yarn bobbin automatic replacement control system based on a humanoid robot according to claim 1, characterized in that, The inner surface of the inflatable gripper is designed with slightly raised anti-slip textures: The central processing unit controls the clamping force of the grippers through a pneumatic pressure regulating valve; The lower limit of adjustment is determined by calculating the ratio of the weight of the full yarn tube to the friction coefficient of the gripper.

5. A flexible yarn bobbin automatic replacement control system based on a humanoid robot according to claim 1, characterized in that, It also includes an intelligent supply and piping subsystem, which comprises: Vibration control mechanism, used to manage vibration amplitude; The amplitude is determined by calculating 1 / 10 of the yarn tube diameter: Mobile AGV platform, used to transport yarn tube racks between various textile equipment; The docking system based on UWB wireless positioning is used to guide the AGV platform to accurately dock with the humanoid robot. The docking accuracy is determined by superimposing the UWB positioning error and the AGV fine-tuning accuracy.

6. The flexible yarn bobbin automatic replacement control system based on a humanoid robot according to claim 1, characterized in that, The central processing unit has a built-in database of equipment process parameters, which stores the spindle layout diagram, spindle dimensions, and design pull-in / pull-out force curves of the textile equipment. The force parameters in the design insertion / extraction force curve are determined by superimposing the viscous effect force of the spindle, the weight of the yarn tube, and the safety factor for the corresponding machine model: Before performing the pipe replacement task, the central processing unit receives the equipment model information and automatically calls the corresponding process parameters to configure the template matching threshold of the visual recognition algorithm, the safety distance parameter of the motion planner, and the stiffness and damping parameters of the impedance controller. The template matching threshold is determined by calculating the minimum value of image feature similarity.

7. A flexible yarn bobbin automatic replacement control system based on a humanoid robot according to claim 1, characterized in that, It also includes a cleaning and condition self-maintenance system, which further includes a self-diagnostic module configured as follows: The sharpness of the images acquired by the binocular depth camera is analyzed periodically. The sharpness is determined by calculating the modulation transfer function, and the value of the modulation transfer function is determined by calculating the ratio of the image edge contrast to the original contrast. When the image clarity is lower than a set threshold, the jet dust removal device is automatically activated for cleaning. The set threshold is determined by calculating the modulation transfer function value corresponding to the minimum effective resolution of image recognition.

8. A flexible yarn bobbin automatic replacement control system based on a humanoid robot according to claim 7, characterized in that, The self-diagnostic module of the cleaning and condition self-maintenance system is also configured to: The air pressure decay characteristics of the inflatable gripper are monitored periodically. The air pressure decay rate is determined by dividing the difference between the initial working air pressure and the residual air pressure after 10 minutes of rest by the initial air pressure. The zero drift error of the force and torque sensors is monitored synchronously. The zero drift error is determined by the deviation between the average value of 10 consecutive samples taken by the sensor under no-load conditions and the theoretical zero value.

9. A flexible yarn bobbin automatic replacement control system based on a humanoid robot according to claim 5, characterized in that, The intelligent tube supply subsystem also includes a yarn tube quality pre-inspection module, which is integrated into the discharge end of the vibrating tube handling mechanism, and includes: The visual inspection unit uses an industrial camera to capture images of the outer wall of the empty yarn tube and identifies damaged areas using an edge detection algorithm. The damage determination threshold is determined by calculating 1 / 3 of the yarn tube wall thickness. The tension detection unit is used to detect the end tension of the fully loaded yarn bobbin. The tension detection threshold is determined by calculating 1 / 5 of the yarn breaking strength. When the detected yarn tension Ftensile in a full yarn bobbin is less than 0.5N, it is determined that the yarn is broken.

10. A flexible yarn bobbin automatic replacement control system based on a humanoid robot according to claim 1, characterized in that, It also includes a multi-machine collaborative scheduling module, integrated into the central processing unit, and configured as follows: The system receives the task status signal of the humanoid robot and the position signals of multiple mobile AGV platforms, and dynamically allocates resources based on task priority. The task priority is determined by multiplying the number of full yarn tubes of the textile equipment by the equipment downtime loss coefficient. A collision-free transport path is planned for multiple AGV platforms, and the safe distance of the path is determined by calculation based on the coupling relationship between the width of the AGV platform and its maximum travel speed.

Images