Adaptive control method and device for gripper used for material conveying of labeling machine
The adaptive control method and device for labeling machines enhance accuracy and adaptability by using soft robotics and machine learning to optimize gripper paths and structures, addressing issues of inaccurate labeling and environmental interference.
Patent Information
- Application Number
- CN202411291173.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-14
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-09-14
AI Technical Summary
Existing labeling machines have low labeling accuracy and poor adaptability to different application scenarios, leading to issues such as inaccurate positioning, loose labeling, wrinkles, and reduced production efficiency due to interference factors like speed fluctuations and environmental conditions.
An adaptive control method and device for a clamping jaw in labeling machines that uses soft robot technology, multi-source monitoring, and machine learning to optimize product grabbing and release paths, considering product and label features, and environmental interference, to ensure stable and accurate labeling.
Improves labeling accuracy and adaptability by optimizing gripper structures and control strategies, reducing labeling errors and enhancing production efficiency in diverse environments.
Smart Images

Figure CN118811238B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of clamp control, and in particular to a method and a device for adaptively controlling a clamp for material collection and conveying of a labeling machine. Background Art
[0002] In modern industrial production, labeling is an important part of product packaging. However, existing labeling technology has many shortcomings in practical applications. Traditional labeling machines often use fixed labeling modes and parameter settings, which are difficult to adapt to products of different shapes, sizes and materials, as well as diversified labels. This leads to problems such as inaccurate labeling position, loose labeling, and wrinkles during the labeling process, which seriously affects the labeling accuracy and quality. In addition, existing labeling machines lack sufficient adaptability and flexibility when facing complex and changing production environments. For example, interference factors such as speed fluctuations on the production line, vibrations and dust in the environment will lead to poor labeling effects and even labeling errors, thereby reducing production efficiency and product qualification rates.
[0003] The existing technology has technical problems that the labeling machine has low labeling accuracy and poor adaptability to different application scenarios. Summary of the invention
[0004] The present application provides a method and device for adaptively controlling a clamping jaw for material collection and conveying of a labeling machine, which is used to solve the technical problems of low labeling accuracy and poor adaptability to different application scenarios of the labeling machine in the prior art.
[0005] In view of the above problems, the present application provides a method and device for adaptively controlling a clamping jaw for material collection and conveying of a labeling machine.
[0006] In a first aspect of the present application, a method for adaptively controlling a clamping jaw for a labeling machine is provided, the method comprising:
[0007] Connect the labeling machine, identify the target product and multiple associated target labels, extract product feature indicators and label feature indicators, the label feature indicators include shape features, size features, and material features; use soft robot technology, combined with the product feature indicators, design the material collection and conveying gripper structure of the labeling machine, the material collection and conveying gripper structure includes product grabbing posture and product release posture; based on the material collection and conveying gripper structure, integrate multi-source monitoring equipment for perception optimization, use machine learning models, autonomously learn and optimize the product grabbing path and product release path of the material collection and conveying gripper; through the product grabbing posture and The first material picking and conveying stability factor is obtained through the product grabbing path; the second material picking and conveying stability factor is obtained through the product release posture and the product release path; according to the label characteristic indicators and the surface characteristics of the product, combined with historical labeling examples, the label stability evaluation result in the attachment process is predicted; based on the label stability evaluation result, the negative impact of attachment is introduced through the first material picking and conveying stability factor and the second material picking and conveying stability factor, and the product grabbing posture and product grabbing path, the product release posture and product release path are adaptively optimized with the use cycle of the target product as a constraint condition to obtain the optimal control strategy.
[0008] The second aspect of the present application provides a self-adaptive control device for a clamping jaw used for material collection and conveying of a labeling machine, the device comprising:
[0009] A feature index extraction module, which is used to connect to the labeling machine, identify the target product and the associated multiple target labels, and extract product feature indicators and label feature indicators. The label feature indicators include shape features, size features, and material features. A clamp structure design module, which is used to use soft robot technology and combine the product feature indicators to design the material-grabbing and conveying clamp structure of the labeling machine. The material-grabbing and conveying clamp structure includes a product grabbing posture and a product release posture. An autonomous learning optimization module, which is based on the material-grabbing and conveying clamp structure, integrates multi-source monitoring equipment for perception optimization, and uses a machine learning model to autonomously learn and optimize the product grabbing path and product release path of the material-grabbing and conveying clamp. A material-grabbing and conveying stability factor acquisition module, which is used to design the material-grabbing and conveying clamp structure of the labeling machine. The stability factor acquisition module is used to acquire the first material handling stability factor through the product grabbing posture and the product grabbing path; to acquire the second material handling stability factor through the product releasing posture and the product releasing path; the stability assessment result prediction module is used to predict the label stability assessment result in the attachment process according to the label characteristic index and the product surface characteristics in combination with the historical labeling examples; the optimal control strategy acquisition module is based on the label stability assessment result, through the first material handling stability factor and the second material handling stability factor, introduces the negative impact of attachment, takes the use cycle of the target product as the constraint condition, and adaptively optimizes the product grabbing posture and the product grabbing path, the product releasing posture and the product releasing path to obtain the optimal control strategy.
[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:
[0011] The feature index extraction module identifies the target product and multiple associated target labels, extracts product feature indexes and label feature indexes; the gripper structure design module uses soft robot technology and combines the product feature indexes to design the material handling and conveying gripper structure of the labeling machine; the autonomous learning optimization module integrates multi-source monitoring equipment for perception optimization based on the material handling and conveying gripper structure, and uses machine learning models to autonomously learn and optimize the product grabbing path and product release path of the material handling and conveying gripper; the material handling and conveying stability factor acquisition module is used to obtain the first material handling and conveying stability factor through the product grabbing posture and product grabbing path; the second material handling and conveying stability factor is obtained through the product release posture and product release path; the stability evaluation result prediction module is used to predict the label stability evaluation result during the attachment process based on the label feature indexes and product surface characteristics, combined with historical labeling examples; the optimal control strategy acquisition module performs adaptive optimization on the product grabbing posture and product grabbing path, the product release posture and product release path, and obtains the optimal control strategy. The technical effect of improving labeling accuracy and enhancing the adaptability of the labeling machine to different application scenarios is achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0013] Figure 1 A schematic flow chart of a method for adaptively controlling a gripper for material conveying of a labeling machine provided in an embodiment of the present application;
[0014] Figure 2 This is a schematic diagram of the structure of the adaptive control device for the clamping jaws used for material collection and conveying of a labeling machine provided in an embodiment of the present application.
[0015] Explanation of the reference numerals: characteristic index extraction module 10 , gripper structure design module 20 , autonomous learning optimization module 30 , material handling and conveying stability factor acquisition module 40 , stability assessment result prediction module 50 , optimal control strategy acquisition module 60 . DETAILED DESCRIPTION
[0016] The present application provides a method and device for adaptively controlling the grippers for material conveying of a labeling machine, so as to solve the technical problems of low labeling accuracy and poor adaptability to different application scenarios of the labeling machine in the prior art.
[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0018] Embodiment 1
[0019] like Figure 1 As shown, the present application provides an adaptive control method for a clamping jaw used for material collection and conveying of a labeling machine, the method comprising:
[0020] Step S100: connecting to a labeling machine, identifying a target product and multiple associated target labels, and extracting product feature indicators and label feature indicators, wherein the label feature indicators include shape features, size features, and material features.
[0021] Specifically, it is necessary to connect with the labeling machine first to realize data interaction and control. Then, the target product and the multiple target labels associated with it are identified. This process involves the use of sensors and image recognition technology to obtain information about the product and the label. During the identification process, the characteristic indicators of the product will be extracted. These indicators include the shape, size, weight, color, surface texture and other characteristics of the product. These characteristic indicators will help with the subsequent processing and operation of the product. At the same time, the characteristic indicators of the label will also be extracted, with special attention paid to shape characteristics (such as whether the shape of the label is square, round or other shapes), size characteristics (length, width, height and other size information of the label), and material characteristics (whether the label is paper, plastic or other materials). These label characteristic indicators are very important for choosing the appropriate labeling method and ensuring that the label can be accurately and firmly attached to the product.
[0022] Step S200: using soft robot technology and combining the product characteristic indicators, designing the material picking and conveying clamping structure of the labeling machine, wherein the material picking and conveying clamping structure includes a product grabbing posture and a product releasing posture.
[0023] Specifically, soft robotics technology is flexible and adaptable, and can be customized according to different product characteristics. In this step, the previously extracted product feature indicators will be combined, including the shape, size, weight, surface characteristics, etc. of the product. According to the characteristics of the product, the structure of the gripper is designed so that it can effectively grasp and convey the product. The gripper structure includes product grasping posture and product release posture. =, the product grasping posture refers to the shape and movement of the gripper when grasping the product, and it is necessary to ensure that the product can be grasped stably and will not cause damage to the product. The product release posture refers to the shape and movement of the gripper when releasing the product to the specified position, and it is necessary to ensure that the product can be accurately placed at the target position. For example, if the product is irregularly shaped, the gripper needs to be designed into a structure that can adapt to the shape to ensure a firm grasp. If the surface of the product is fragile, special attention should be paid to both the grasping posture and the release posture to avoid scratches or damage to the surface of the product. By adopting soft robotics technology and combining product feature indicators to design the gripper structure, the material collection and conveying efficiency and accuracy of the labeling machine can be improved to meet the labeling needs of different types of products.
[0024] Step S300: Based on the material picking and conveying gripper structure, multi-source monitoring equipment is integrated to perform perception optimization, and a machine learning model is used to autonomously learn and optimize the product grabbing path and product releasing path of the material picking and conveying gripper.
[0025] Specifically, based on the gripper structure for material collection and conveying, monitoring devices from multiple sources are integrated, including but not limited to high-definition cameras, force sensors, position sensors, etc. These monitoring devices can collect a large amount of data of the gripper, product and surrounding environment in real time, such as the position, speed, grasping force of the gripper, the shape, position, posture of the product, and the distribution of obstacles in the surrounding environment. A reinforcement learning model is used to optimize the product grasping path and product release path of the gripper for material collection and conveying. Reinforcement learning is a machine learning method that learns the optimal strategy by interacting with the environment and based on feedback signals. The gripper structure for material collection and conveying and the multi-source monitoring device are integrated, and the data collected by the multi-source monitoring device are used as the input of the reinforcement learning model. The reinforcement learning model will output an action based on the current state (i.e., the input information), which can be the moving direction of the gripper, the grasping force, etc. The gripper will perform corresponding operations based on the action output by the model, such as moving to a specified position and grabbing the product. After executing the action, the environment will give a feedback signal, which can indicate whether the operation of the gripper is successful, such as whether the product is successfully grasped, whether it is placed in the correct position when releasing the product, etc. The reinforcement learning model will adjust its strategy based on this feedback signal to achieve better results in future operations. By continuously interacting and learning with the environment, the reinforcement learning model can gradually optimize the gripper's product grabbing path and product release path, improving the gripper's operating efficiency and accuracy. By using the reinforcement learning model, the gripper can learn and optimize autonomously in continuous practice to adapt to different products and environments, improving the performance and reliability of the labeling machine.
[0026] Step S400: Obtain a first material picking and conveying stability factor through the product grabbing posture and the product grabbing path; obtain a second material picking and conveying stability factor through the product releasing posture and the product releasing path.
[0027] Specifically, the product grasping posture refers to the specific shape and angle of the gripper when grasping the product, including the degree of opening and closing of the gripper, the location and method of contact between the gripper and the product, etc. The product grasping path is the specific trajectory of the gripper moving from the initial position to the position of grasping the product. For example, if the gripping posture of the gripper makes the contact points with the product evenly distributed, and the grasping force is moderate, and the grasping path is reasonably planned, avoiding obstacles that may cause collision or interference, and the movement is smooth and smooth, then the grasping process will be more stable. By comprehensively evaluating and calculating multiple parameters of the grasping posture and grasping path, such as the contact area between the gripper and the product, the uniformity of the grasping force, the smoothness of the path, the distance from the obstacle, etc., a numerical first material handling stability factor is obtained. The higher the value of this factor, the better the stability of the grasping process. The product release posture is the specific shape and angle of the gripper when releasing the product, such as the speed and degree of the gripper's release. The product release path is the specific trajectory of the gripper moving from the position of grasping the product to the position of releasing the product. If the release posture of the gripper can allow the product to smoothly detach from the gripper, and the release path avoids sudden turns or excessive acceleration, it ensures that the product can be accurately and stably placed in the predetermined position without falling, offset or other unstable situations. By analyzing and calculating the relevant parameters of the release posture and release path, such as the change in release speed, the straightness of the path, the deviation from the target position, etc., the second material handling and conveying stability factor is obtained to characterize the stability of the release process. In summary, through detailed analysis and quantitative calculation of the posture and path during product grasping and release, the first material handling and conveying stability factor and the second material handling and conveying stability factor are obtained, providing a clear quantitative basis for subsequent optimization and improvement.
[0028] Step S500: predicting the label stability evaluation result during the attachment process according to the label characteristic index and the product surface characteristics in combination with historical labeling examples.
[0029] Specifically, label feature indicators such as shape, size, and material will affect the fit and firmness of the label during the attachment process. For example, larger labels require stronger adhesion to be stably attached, and labels of different materials have different sensitivities to environmental factors such as temperature and humidity. Product surface characteristics include surface roughness, flatness, material, and whether there is oil, dust, etc. A rough or uneven surface will cause the label to not fit tightly, while oil or dust will reduce the stickiness of the label. Historical labeling examples provide labeling effect data for similar labels and product combinations in the past. By analyzing these data, we can understand the probability and degree of label shedding, warping, wrinkling, etc. under a specific combination of label features and product surface characteristics. Based on the above information, a regression model is established, with label feature indicators and product surface characteristics as input variables, and quantitative indicators of label stability (such as the proportion of labels that remain intact after a certain period of time after attachment, the average length of label edge warping, etc.) as output variables. Ultimately, the label stability assessment result obtained can be a quantitative score, grade (such as high, medium, and low stability) or a specific description
[0030] Step S600: Based on the label stability evaluation result, the negative impact of adhesion is introduced through the first material picking and conveying stability factor and the second material picking and conveying stability factor, and the product grasping posture and product grasping path, product release posture and product release path are adaptively optimized with the use cycle of the target product as a constraint to obtain the optimal control strategy.
[0031] Specifically, negative effects of attachment refer to factors that can cause labels to be loosely attached, such as vibration, impact or improper pressure on the product during the material collection and conveying process. Taking the use cycle of the target product as a constraint means that the optimization process must ensure that the label can remain stably attached throughout the entire use cycle of the product. For the optimization of the product grasping posture and product grasping path, consideration will be given to adjusting the opening and closing speed of the gripper, the change curve of the grasping force, etc., to reduce the vibration and impact on the product during the grasping process. For example, if it is found that the current grasping posture and path cause the product to shake greatly during grasping, the grasping action will be slowed down and the buffer stage will be increased during grasping. For the optimization of the product release posture and product release path, emphasis will be placed on controlling the speed and angle of release to ensure that the product is accurately placed while minimizing the adverse effects on label attachment. For example, the original method of quickly releasing the product will cause the label to collide with the contact surface instantly and produce wrinkles, so it needs to be adjusted to a smoother release method. During the entire optimization process, different parameter combinations are continuously simulated and tested, and the decision tree algorithm is used to analyze the characteristics of the data to construct a tree-like structure, so as to make decisions or predictions to find the optimal control strategy that can ensure stable label attachment within the target product's usage cycle. This strategy will include the optimal product grasping posture, grasping path, product release posture and specific parameter settings for the release path, which will significantly improve the stability of label attachment.
[0032] The movement of the gripper for material collection and conveying is precisely regulated on a millimeter scale, including the speed, force and movement trajectory of grasping and releasing. Its core goal is to achieve accurate docking and stable attachment of labels to products, while minimizing the adverse effects of attachment. To this end, a highly innovative gripper structure design is adopted. For example, the gripping surface with fine texture can increase friction and effectively prevent the label from sliding during the grasping process; the clamping part made of deformable material has excellent flexibility and adaptability, and can be fine-tuned according to the shape and material of the product, significantly reducing the deviation of the label during the grasping and releasing process, while greatly reducing the risk of attachment. Through these careful designs, the adaptation process of labels and products is significantly accelerated, the attachment efficiency is greatly improved, and the consistency and stability of labeling quality are effectively ensured.
[0033] The material collection and conveying optimization strategy includes two main methods: Method 1 is passive optimization, which mainly relies on the ingenious design of the clamping structure to achieve stable material collection and conveying. For example, by carefully designing the clamping part with adaptive capabilities, it can automatically and flexibly adjust the gripping force and shape according to products of different shapes and materials, thereby achieving stable material collection and conveying without external intervention. The significant advantage of this method is that the structure is relatively simple and the operation is extremely convenient.
[0034] The second method is active optimization, which uses external control systems (such as high-precision sensors, real-time feedback algorithms, etc.) to achieve precise control of the material handling and conveying process. For example, a highly sensitive force sensor is used to monitor the contact force between the gripper and the product in real time, and the gripping strategy is dynamically adjusted through advanced algorithms to effectively respond to various adhesion conditions and product characteristics. This method has the outstanding advantages of extremely fast response speed, strong adaptability, and very significant optimization effect. In summary, through the comprehensive and integrated application of passive and active optimization strategies, not only the material handling and conveying stability and accuracy of the labeling machine are greatly improved, but also the adverse effects of adhesion problems on labeling quality are effectively reduced, thereby effectively ensuring the efficient and stable operation of the overall production line and the high consistency of product quality.
[0035] In a possible implementation, step S100 further includes:
[0036] Step S110: extracting product characteristic indicators based on target products of different forms.
[0037] Step S120: Based on the product characteristic index and in combination with the material properties of the soft robot, the gripper structure is adaptively designed with the requirement of adapting to target products of different forms.
[0038] Step S130: The product grabbing path and the product releasing path are updated through a path planning algorithm, and a sliding mode control algorithm is used to dynamically adjust the grabbing force of the material conveying gripper according to feedback information during the grabbing process.
[0039] Specifically, due to the existence of target products of different forms, a comprehensive and detailed analysis is required to extract product characteristic indicators. First, in terms of shape, it is necessary to clarify whether the product is a regular shape (such as a cube, cylinder, sphere, etc.) or an irregular shape, and describe its specific geometric characteristics, such as the contour curve and curvature change of the irregular shape. In terms of size, accurately measure the linear dimensions of the product such as length, width, height, diameter, thickness, and the curvature and angle of the curved surface. In terms of weight, use accurate weighing equipment to obtain product quality data. The center of gravity position is also an important characteristic indicator, and the coordinates of the center of gravity of the product are determined by specific methods or equipment. Surface texture characteristics include surface roughness, flatness, hardness, elasticity, etc., which are measured and evaluated by tools such as roughness measuring instruments and hardness testers. In addition, the material properties of the product (such as the type of material, strength, ductility, etc.), structural characteristics (whether there are holes, protrusions, depressions, etc.) and special characteristics (such as magnetism, conductivity, etc.) are also considered. By examining and measuring these aspects, the characteristic indicators of target products of different forms can be extracted comprehensively and accurately, providing key basic information for subsequent design, production and operation.
[0040] If the product characteristic indicators show that the product is small and irregular in shape, with sharp corners or complex curved surfaces. At this time, combined with the high flexibility and deformability of soft robot materials, the gripper structure will be designed as multiple flexible tentacle-like structures, which can bend and wrap adaptively according to the shape of the product to achieve stable grasping. If the product is large and the surface is relatively fragile, considering the softness and cushioning performance of the soft robot material, the gripper is designed as a large-area soft contact pad, which avoids damage to the product surface by evenly distributing pressure, and can adapt to changes in product size and shape. For example, for some products with special shapes or structures, such as those with cavities inside or that require grasping at a specific angle, the gripper structure will contain retractable or rotatable parts, and use the elasticity and deformation ability of the soft robot material to adjust according to the specific form of the product. The adaptive design of the gripper structure aims to make full use of the characteristics of the soft robot material to meet the limited conditions of adapting to target products of different forms, ensuring efficient, stable and non-destructive grasping operations in various situations.
[0041] The path planning algorithm takes into account a variety of factors, such as the layout of the workspace, the location of obstacles, and the initial and target positions of the product. For example, if the workspace is narrow, the algorithm will plan a more compact grasping path that avoids obstacles. For products with complex shapes, the algorithm will select the most suitable grasping angle and approach direction based on their geometric characteristics to ensure efficient and safe grasping. At the same time, a sliding mode control algorithm is used, which can dynamically adjust the grasping force of the material conveying jaws based on real-time feedback information during the grasping process, such as the contact force between the jaws and the product detected by the force sensor and the movement state of the jaws detected by the accelerometer. Assuming that during the grasping process, the feedback information shows that the contact force between the jaws and the product is too large, which will cause damage to the product, the sliding mode control algorithm will quickly reduce the grasping force. On the contrary, if the feedback information indicates that the contact force is too small and the product is at risk of slipping, the algorithm will increase the grasping force to ensure stable grasping. This dynamic adjustment of the grasping force can adapt to products of different weights, shapes and surface characteristics, improve the reliability and adaptability of grasping, and reduce the possibility of errors and product damage during grasping. Through adaptive design and path optimization, the labeling efficiency and quality of products with different shapes can be effectively improved, and the labeling failure rate caused by changes in product shape can be reduced.
[0042] In a possible implementation, step S110 further includes:
[0043] Step S111: Based on the product characteristic index, use the near infrared spectrum analysis algorithm to identify the material of the target product and obtain the surface characteristics of the product.
[0044] Step S112: According to the surface characteristics of the product, a fuzzy logic control algorithm is used to correct the gripping force of the material conveying clamp.
[0045] Step S113: After completing the gripping force calibration of the material picking and conveying clamp, the end point of the updated product gripping path and the starting point of the updated product release path are adjusted according to the label attachment position in combination with the force sensing algorithm, the first material picking and conveying stability factor, and the second material picking and conveying stability factor.
[0046] Specifically, based on the product characteristic indicators extracted previously. These indicators include basic information such as the shape, size, and weight of the product. Then, the near-infrared spectroscopy analysis algorithm is used to identify the material of the target product. Near-infrared spectroscopy technology obtains its spectral information by measuring the absorption, reflection, or scattering characteristics of the target product to near-infrared light. Different materials have unique absorption peaks and characteristic spectral bands in the near-infrared spectral region. The spectral data of the collected target product is compared and analyzed with the pre-established material spectrum database. For example, the database has stored typical spectral characteristics of various common materials (such as plastics, metals, wood, etc.). During the comparison process, the algorithm determines the material type of the target product based on the similarities and differences of the spectra. Based on the material recognition results, the surface characteristics of the product are further obtained. This includes surface roughness, hardness, glossiness, chemical composition, etc. For example, if the product material is identified as a certain plastic, it will be known that its surface is relatively smooth and has a lower hardness; if it is a metal material, the surface is rougher and has a higher hardness. These obtained product surface characteristic information has important guiding significance for subsequent operations such as grasping, processing, and labeling.
[0047] These surface characteristics include the roughness value of the surface, the hardness level, the description of the smoothness, etc. Then, the fuzzy logic control algorithm is used to correct the gripping force of the gripper for material handling and conveying. The working principle of the fuzzy logic control algorithm is to fuzzify the input product surface characteristic parameters. For example, the surface roughness is divided into fuzzy subsets such as "very rough", "relatively rough", "medium rough", "relatively smooth", and "very smooth". Similarly, there will be similar fuzzy divisions for hardness. Then, reasoning and judgment are carried out according to the pre-set fuzzy rules, which are based on experience and experimental data. For example, if the product surface is judged to be "very rough" and "high hardness", then the rule will indicate that the gripping force should be increased. Conversely, if the surface is "very smooth" and "low hardness", the rule will indicate that the gripping force should be reduced. In the reasoning process, the algorithm will comprehensively consider the fuzzy state of multiple surface characteristic parameters. Finally, through defuzzification, a specific gripping force correction value is obtained, so as to achieve precise adjustment of the gripping force of the gripper to adapt to products with different surface characteristics, ensure the stability and safety of gripping, and avoid damage to the product surface.
[0048] After the correction of the gripping force of the gripper for material handling is completed, it will be combined with the force sensing algorithm, which can sense the information of the force generated when the gripper contacts the product in real time, including the magnitude, direction and change trend of the force. This force sensing data is considered together with the first material handling stability factor and the second material handling stability factor mentioned above, and the label attachment position is used as an important reference point to adjust the end point of the updated product grasping path. This means that if the force sensing data shows that the contact force between the gripper and the product is unevenly distributed or not conducive to subsequent operations at the end point of the currently planned grasping path, and combined with the evaluation results of the stability factor, it is found that there is a risk of instability, then the end point position of the grasping path will be fine-tuned, such as moving slightly forward, backward, left or right, or making slight changes in height to ensure that the posture and position of the gripper when grasping the product are more conducive to subsequent operations, especially the attachment of labels. Similarly, similar adjustments will be made to the starting point of the updated product release path. By comprehensively considering force sensing information, stability factors and label attachment positions, the starting position and posture of the product when released can ensure that the label can be accurately and stably attached to the predetermined position to the greatest extent, reducing the problem of label attachment deviation or looseness caused by inaccurate release position. Through material identification and grasping force optimization, the damage rate of special material products during the labeling process is reduced, and the success rate of labeling is improved.
[0049] In a possible implementation, step S113 further includes:
[0050] Step S1131: On the target product production line, a high-speed visual recognition algorithm is used to synchronously track the target product, and a product grabbing path and a product releasing path are planned when the target product production line is in operation.
[0051] Step S1132: using a model predictive control algorithm, adjusting the motion trajectory of the product grabbing posture and the motion trajectory of the product releasing posture according to the speed change of the target product production line.
[0052] Step S1133: predicting the speed change trend of the target product production line, and optimizing the adaptability of the product grabbing path and the motion trajectory of the product grabbing posture, and the product releasing path and the motion trajectory of the product releasing posture under the running state of the target product production line.
[0053] Specifically, when the target product is on the production line, a high-speed visual recognition algorithm will be introduced to play a key role. The high-speed visual recognition algorithm can capture the real-time image information of the target product with extremely high frequency and accuracy. Through the rapid analysis and processing of these continuous images, the algorithm can accurately identify the key features of the target product such as the position, shape, and orientation. In the process of synchronous tracking of the target product, the algorithm can update the dynamic information of the target product on the production line in real time, including its moving speed, displacement change, and posture adjustment. Based on these real-time and accurate tracking data, it begins to plan the product grabbing path and product release path under the operation state of the target product production line. For example, if the target product is moving at a constant speed along a straight conveyor belt, the algorithm will calculate when, where, at what angle and speed the gripper needs to approach the product to achieve grabbing. At the same time, it will also plan where and how the gripper releases the product after processing or processing is completed to ensure that the product can continue to flow on the production line accurately. For example, when the speed of the production line fluctuates or the position of the product has a certain degree of randomness, the high-speed visual recognition algorithm will respond quickly and dynamically adjust the planning of the grabbing path and release path to adapt to this uncertainty. In short, this step achieves accurate tracking of the target product through a high-speed visual recognition algorithm, and plans a reasonable and efficient grasping path and release path based on this, laying the foundation for subsequent grasping and releasing operations.
[0054] The model predictive control algorithm will be used to deal with the changes in the speed of the target product production line. The model predictive control algorithm will first establish a mathematical model of the speed change of the target product production line. This model will take into account a variety of factors, such as the mechanical structure of the production line, the characteristics of the drive system, load changes, and previously observed speed change patterns. Based on this model, the algorithm can predict the changes in the production line speed over a period of time in the future. Then, according to the predicted changes in the production line speed, the motion trajectory of the product grasping posture and the motion trajectory of the product release posture are adjusted. For example, if it is predicted that the production line speed is about to increase, in order to ensure that the grasping and releasing actions can keep pace with the rhythm of the production line, the algorithm will shorten the grasping and releasing action time, speed up the action speed, and adjust the start time and acceleration curve of the action so that the gripper can complete the grasping and releasing operations more quickly and accurately. On the contrary, if it is predicted that the production line speed is about to slow down, the algorithm will extend the action time and slow down the action speed to avoid grasping or releasing the product before it reaches the correct position, thereby ensuring the accuracy and stability of the operation. In this way, the model predictive control algorithm can adapt to the dynamic changes in production line speed in real time, optimize the motion trajectory of product grasping and releasing postures, and improve the efficiency and quality of the production process.
[0055] The speed change trend of the target product production line is predicted, taking into account a variety of factors, such as the equipment operation status of the production line, the current production task volume, the supply of raw materials, and the past speed change rules. Based on the prediction of the speed change trend, the product grabbing path, the motion trajectory of the product grabbing posture, the product release path, and the motion trajectory of the product release posture under the operation state of the target product production line are optimized for fitness. Fitness optimization means adjusting these paths and trajectories to better match the upcoming production line speed changes. For example, if it is predicted that the production line speed will gradually increase, the length of the product grabbing path will be shortened to make the grabbing action more direct and rapid; at the same time, the motion trajectory of the product grabbing posture is adjusted to increase the acceleration to complete the grabbing action faster. Similar optimization will be performed on the motion trajectory of the product release path and posture to ensure that the product can be released quickly and accurately to the designated position on the production line. For another example, if it is predicted that the production line speed will fluctuate greatly, the flexibility of the path and trajectory will be increased so that the grabbing and release actions can maintain a certain stability and accuracy under different speed conditions. Through this adaptability optimization, the efficiency and reliability of the entire production process can be improved, and production delays, product damage or operational errors caused by changes in production line speed can be reduced. Through high-speed identification and tracking and adaptive control systems, the efficiency and accuracy of labeling on high-speed production lines can be improved, and labeling failures caused by changes in production line speed can be reduced.
[0056] In a possible implementation, step S300 further includes:
[0057] Step S310: using an environment perception algorithm to identify interference factors in the environment, wherein the interference factors include vibration and dust.
[0058] Step S320: Based on the interference factors, an adaptive compensation mechanism is set, and the adaptive compensation mechanism is used to dynamically adjust the gripping force of the material-taking and conveying clamping claw.
[0059] Step S330: Based on the label stability evaluation result, after starting the adaptive compensation mechanism, a new labeling task is synchronously sent to the labeling machine.
[0060] Specifically, the environmental perception algorithm will combine a variety of sensors and data acquisition technologies. For vibration detection, accelerometers or vibration sensors will be used to measure parameters such as vibration frequency, amplitude and direction. These sensors can capture vibration signals generated by equipment operation, mechanical collision or other sources. For dust identification, optical sensors, particle sensors or dust concentration monitors are used. Optical sensors can determine the presence and concentration of dust by detecting the scattering or absorption of light; particle sensors can directly measure the size and number of particles in the air. By processing and analyzing the data collected by these sensors in real time, the environmental perception algorithm can accurately identify interference factors such as vibration and dust in the environment. This timely and accurate identification of interference factors provides a key prerequisite for the subsequent adoption of corresponding compensation and response measures.
[0061] Based on the interference factors identified earlier, such as vibration and dust, an adaptive compensation mechanism is set. The main function of this adaptive compensation mechanism is to adjust the gripping force of the gripper for material handling in real time and dynamically according to the interference in the environment. If the vibration in the environment is strong, it will cause unstable contact between the gripper and the product. At this time, the adaptive compensation mechanism will increase the gripping force of the gripper to enhance the firmness of the grip and prevent the product from falling off or shifting during the vibration. If there is a lot of dust in the environment, the friction between the gripper and the product surface will be reduced, affecting the stability of the grip. In this case, the adaptive compensation mechanism will increase the gripping force accordingly to ensure that the product can be reliably gripped. Alternatively, when strong intermittent vibration suddenly occurs, such as the vibration shock caused by the startup of large equipment nearby, the adaptive compensation mechanism will respond quickly, greatly increase the gripping force in a short time, and then return to normal level after the vibration weakens. By dynamically adjusting the gripping force in this way, the adaptive compensation mechanism can effectively deal with interference factors in the environment and improve the stability and reliability of the material handling process.
[0062] First, based on the previously obtained label stability evaluation results, this evaluation result is a comprehensive evaluation of the firmness, position accuracy, flatness, etc. of the label under the current environment and operating conditions. After the adaptive compensation mechanism is activated, it means that the gripping force of the gripper for material collection and conveying has been adjusted accordingly to the interference factors in the environment (such as vibration, dust, etc.) to optimize the gripping and conveying process. Then, a new labeling task will be sent to the labeling machine synchronously, and this new labeling task will be updated and adjusted according to the current situation. For example, if the label stability evaluation results show that the previous labeling effect is not ideal because the posture or position of the product has changed after the gripping force adjustment, the new labeling task will include fine-tuning the labeling position, changing the labeling pressure, or resetting the labeling speed. For another example, if environmental interference factors have a potential impact on the attachment of the label, the new labeling task will add some additional inspection steps or adopt more stringent quality control standards. Through such synchronous operation, it can ensure that the labeling machine receives the new task instructions that match it in time after making adjustments in the material collection and conveying link, thereby improving the efficiency and quality of the entire labeling process and ensuring that the label can be stably and accurately attached to the product. Through anti-interference path planning and adaptive compensation mechanism, the labeling quality in complex environments is ensured, and labeling failures caused by environmental factors are reduced.
[0063] In a possible implementation, step S500 further includes:
[0064] Step S510: Identify various defects that occur during the labeling process, classify them, and obtain a labeling defect classification set.
[0065] Step S520: Based on the attachment defect classification set, attaching defect tracing is performed in combination with the historical labeling instance, and the attachment defect cause is analyzed to determine the attachment defect factor.
[0066] Step S530: Based on the attachment defect factor, the labeling parameters of the labeling machine are adjusted by feedback control, and the labeling parameters include labeling speed and labeling pressure.
[0067] Specifically, high-definition image recognition technology is used to take photos of the labeled products, and then image processing algorithms are used to detect whether there are obvious appearance defects such as wrinkles, bubbles, and damage on the label surface. At the same time, position detection sensors are used to determine whether the label has shifted on the product, or whether the edge of the label is not accurately aligned with the edge of the product. As for the degree of fit between the label and the product, a pressure sensor will be used to detect whether there are areas that are not fully fitted. After identifying these defects, they will be classified according to certain rules and standards. For example, according to the type of defect, it can be divided into appearance defect categories (such as wrinkles, bubbles, and damage), position defect categories (offset, misalignment), fit defect categories (incomplete fit), etc., and finally a set of attachment defect classifications is obtained.
[0068] Work is carried out based on the previously obtained attachment defect classification set, which contains the classification of various defects that occur in the current labeling process. Then, these classifications will be combined with historical labeling instances to trace attachment defects. The historical labeling instances record similar defects that occurred in past labeling operations and the relevant operating parameters, environmental conditions and other information at that time. By comparing the current defect classification with the defect conditions in the historical instances, the common factors that cause these defects are analyzed. For example, if the current label offset defect is similar to the offset defect caused by the decline in equipment accuracy in a certain period in history, then the focus will be on whether the equipment accuracy is the cause of the current defect. In the analysis process, a variety of factors are considered comprehensively, such as the working parameters of the labeling machine (labeling speed, pressure, temperature, etc.), the characteristic changes of the label material, the state of the product surface (flatness, roughness, cleanliness, etc.), and the conditions of the operating environment (humidity, temperature, dust content, etc.). Through comprehensive and in-depth comparison and analysis, the specific factors that cause the current attachment defects are finally determined.
[0069] Based on the identified attachment defect factors, the labeling parameters of the labeling machine will be adjusted by feedback control. If the attachment defect factors indicate that the labeling pressure is too high, resulting in label wrinkles or loose fitting, the labeling pressure will be reduced accordingly. For example, the original labeling pressure is 30N, and according to the feedback control, it is adjusted to 25N or lower to improve the labeling effect. If the attachment defect factors show that the labeling speed is too fast, causing the label position to shift or the attachment to be uneven, the labeling speed will be slowed down. For example, the labeling speed is reduced from the original 60 products per minute to 50 products per minute, making the labeling process more stable and accurate. Through this real-time feedback control and dynamic adjustment of labeling parameters (such as labeling speed and labeling pressure), problems in the labeling process can be corrected in time, the labeling quality can be gradually optimized, and the recurrence of attachment defects can be reduced. Through online quality inspection and feedback control, the labeling process can be adjusted in real time to improve the qualified rate of labeling and reduce the production of unqualified products.
[0070] In a possible implementation, step S600 further includes:
[0071] Step S610: Utilize the Internet of Things technology to collect data during the labeling process and upload it to the cloud.
[0072] Step S620: Analyze the uploaded data through a big data analysis algorithm to extract key factors associated with labeling efficiency, including a product conveying speed of a target product and a label supply speed corresponding to a plurality of target labels associated with the target product.
[0073] Step S630: Based on the key factors, a decision tree algorithm is used to identify potential efficiency bottlenecks in the labeling process and generate optimization decisions.
[0074] Specifically, IoT technology will acquire data by deploying various sensors and intelligent monitoring devices in labeling equipment and related links. For example, speed sensors are installed on the conveyor belt to monitor the moving speed of the target product; pressure sensors are set at key parts of the labeling machine to collect pressure data applied during labeling; and position sensors are used to determine whether the label is accurately pasted. These sensors and monitoring devices will convert the collected data into digital signals in real time and transmit the data to the local data collection terminal through network connections (such as Wi-Fi, Bluetooth, 4G / 5G, etc.). The local terminal will perform preliminary processing and integration of these data, remove outliers, perform data compression, etc., to reduce the amount of data and improve transmission efficiency. The processed labeling process data will be uploaded to the cloud server. The cloud provides powerful storage and computing capabilities, which can safely and stably save a large amount of data and prepare for subsequent analysis and processing.
[0075] When the data of the labeling process is uploaded to the cloud, big data analysis algorithms will be used to deeply explore the value of these data. Big data analysis algorithms have a powerful ability to process and analyze massive amounts of data. They will conduct a comprehensive and detailed analysis of the large amount of labeling process data uploaded. During the analysis process, the focus is on extracting the key factors that are closely related to labeling efficiency. Among them, the product conveying speed of the target product is a key factor. If the product conveying speed is unstable or too slow, the labeling machine will wait for the product, thereby reducing the overall labeling efficiency. For example, when the product conveying speed is reduced from 50 per minute to 30, the labeling machine will have a long idle time, affecting the number of products that are labeled per unit time. The label supply speed corresponding to the multiple target labels associated with the target product is also crucial. If the label supply speed cannot keep up with the product conveying speed, it will cause interruption or delay in the labeling process. For example, if the label supply speed is 40 per minute and the product conveying speed is 50 per minute, there will be insufficient label supply, affecting the labeling efficiency. Through the precise analysis of these data, the specific impact of these key factors on the labeling efficiency can be accurately quantified, providing strong data support for subsequent optimization decisions.
[0076] Based on the key factors such as the product delivery speed and label supply speed of the target product identified above, the decision tree algorithm is introduced to conduct in-depth and accurate analysis and decision making. The decision tree algorithm is a powerful data-driven analysis tool. It will gradually build a hierarchical decision tree structure based on the different numerical ranges and combinations of key factors. Each branch node of the tree represents the judgment condition for a key factor, and the leaf node corresponds to the final decision result. When analyzing the labeling process, the specific values of the key factors collected in real time will be input into the already constructed decision tree. For example, if the product delivery speed reaches 120 per minute, and the label supply speed is only 80 per minute, the decision tree will make judgments along a specific branch path according to the pre-set judgment rules. Assume that it is set in the decision tree that when the product delivery speed exceeds the label supply speed by a certain proportion, the label supply link is considered to be a potential efficiency bottleneck. In the above example, since the product delivery speed greatly exceeds the label supply speed, the algorithm will determine that there is a problem in the label supply link, which is a potential efficiency bottleneck in the current labeling process. Once the efficiency bottleneck is identified, the decision tree algorithm will generate targeted optimization decisions based on its internal logic and rules. If it is determined that the bottleneck is caused by the inability to keep up with the label supply speed, the generated optimization decisions include: immediately purchasing and installing new label supply equipment to increase supply capacity; upgrading and transforming the existing label supply equipment to improve its operating efficiency; replanning the storage and transportation routes of labels to reduce the time loss in the intermediate links, thereby speeding up the label supply speed. For example, if the analysis finds that the product conveying speed and label supply speed match well, but the average labeling operation time of the labeling machine exceeds the preset reasonable range, the decision tree will determine that the operation settings of the labeling machine or the performance of the equipment itself is the efficiency bottleneck, and the corresponding optimization decisions are: recalibrating the parameters of the labeling machine, such as adjusting the labeling pressure and optimizing the motion trajectory of the labeling head; performing comprehensive maintenance and maintenance on the labeling machine, replacing worn parts; and even considering replacing more advanced and efficient labeling machine models. Through this detailed and precise analysis and decision-making process, it is possible to provide a practical and operational guidance plan for improving the efficiency of the labeling process, thereby effectively solving potential problems and improving the production efficiency and quality of the entire labeling process.
[0077] Embodiment 2
[0078] Based on the same inventive concept as the adaptive control method for the clamping jaws used for material feeding and conveying of the labeling machine in the above-mentioned embodiment, Figure 2 As shown, the present application provides a self-adaptive control device for a clamping jaw used for material feeding and conveying of a labeling machine. The device and method embodiments in the present application are based on the same inventive concept. The device includes:
[0079] The feature index extraction module 10 is used to connect to the labeling machine, identify the target product and the associated multiple target labels, and extract product feature indicators and label feature indicators. The label feature indicators include shape features, size features, and material features.
[0080] The clamp structure design module 20 is used to design the material picking and conveying clamp structure of the labeling machine by using soft robot technology and combining the product characteristic indicators. The material picking and conveying clamp structure includes a product grabbing posture and a product releasing posture.
[0081] The autonomous learning optimization module 30 is based on the structure of the material picking and conveying clamp, integrates multi-source monitoring equipment for perception optimization, and adopts a machine learning model to autonomously learn and optimize the product grabbing path and product releasing path of the material picking and conveying clamp.
[0082] The material picking and conveying stability factor acquisition module 40 is used to acquire a first material picking and conveying stability factor through a product grabbing posture and a product grabbing path; and to acquire a second material picking and conveying stability factor through a product releasing posture and a product releasing path.
[0083] The stability evaluation result prediction module 50 is used to predict the label stability evaluation result during the attachment process according to the label characteristic index and the product surface characteristics in combination with historical labeling examples.
[0084] The optimal control strategy acquisition module 60 is based on the label stability evaluation result, introduces the negative impact of adhesion through the first material picking and conveying stability factor and the second material picking and conveying stability factor, and takes the use cycle of the target product as a constraint condition to adaptively optimize the product grasping posture and product grasping path, product release posture and product release path to obtain the optimal control strategy.
[0085] Furthermore, the characteristic index extraction module 10 also includes:
[0086] A product characteristic index extraction unit is used to extract product characteristic indexes based on target products of different forms.
[0087] The gripper structure adaptive design unit is used to perform gripper structure adaptive design based on the product characteristic indicators and the material properties of the soft robot, with the adaptation to target products of different forms as the limiting condition.
[0088] The gripping force adjustment unit of the clamping jaws is used to update the product gripping path and the product release path through a path planning algorithm, and adopt a sliding mode control algorithm to dynamically adjust the gripping force of the clamping jaws for material conveying according to feedback information during the gripping process.
[0089] Furthermore, the product characteristic index extraction unit also includes:
[0090] A product surface characteristic acquisition unit is used to identify the material of a target product based on the product characteristic index and to acquire the product surface characteristic by using a near infrared spectrum analysis algorithm.
[0091] A gripping force correction unit is used to correct the gripping force of the material conveying clamp according to the surface characteristics of the product by using a fuzzy logic control algorithm.
[0092] A path adjustment unit is used to adjust the end point of the updated product grasping path and the starting point of the updated product release path according to the label attachment position after completing the grasping force calibration of the material picking and conveying clamp, combined with the force sensing algorithm, and the first material picking and conveying stability factor and the second material picking and conveying stability factor.
[0093] Furthermore, the path adjustment unit further includes:
[0094] The target product tracking unit is used to synchronously track the target product on the target product production line using a high-speed visual recognition algorithm, and to plan a product grabbing path and a product releasing path when the target product production line is in operation.
[0095] The motion trajectory adjustment unit is used to use a model predictive control algorithm to adjust the motion trajectory of the product grasping posture and the motion trajectory of the product releasing posture according to the speed change of the target product production line.
[0096] The motion trajectory fitness optimization unit is used to predict the speed change trend of the target product production line, and to optimize the fitness of the product grasping path and the motion trajectory of the product grasping posture, the product release path and the motion trajectory of the product release posture under the operation state of the target product production line.
[0097] Furthermore, the autonomous learning optimization module 30 also includes:
[0098] The interference factor identification unit is used to use an environment perception algorithm to identify interference factors in the environment, and the interference factors include vibration and dust.
[0099] A gripping force adjustment unit is provided, wherein the gripping force adjustment unit sets an adaptive compensation mechanism based on the interference factor, and the adaptive compensation mechanism is used to dynamically adjust the gripping force of the material-taking and conveying clamping claw.
[0100] A labeling task sending unit, wherein the labeling task sending unit synchronously sends a new labeling task to the labeling machine after starting the adaptive compensation mechanism based on the label stability evaluation result.
[0101] Furthermore, the stability assessment result prediction module 50 also includes:
[0102] The defect classification set acquisition unit is used to identify various defects occurring during the labeling process, classify them, and acquire a labeling defect classification set.
[0103] The attachment defect factor determination unit performs attachment defect tracing based on the attachment defect classification set and in combination with the historical labeling instance, and analyzes the attachment defect cause to determine the attachment defect factor.
[0104] A labeling parameter adjustment unit, wherein the labeling parameter adjustment unit adjusts the labeling parameters of the labeling machine in a feedback control manner based on the attachment defect factor, and the labeling parameters include labeling speed and labeling pressure.
[0105] Furthermore, the optimal control strategy acquisition module 60 also includes:
[0106] The labeling data collection unit is used to collect data during the labeling process using the Internet of Things technology and upload it to the cloud.
[0107] A labeling efficiency-related factor extraction unit is used to analyze the uploaded data through a big data analysis algorithm to extract key factors related to labeling efficiency, wherein the key factors include a product conveying speed of a target product and a label supply speed corresponding to a plurality of target labels associated with the target product.
[0108] The optimization decision generation unit adopts a decision tree algorithm based on the key factors to identify potential efficiency bottlenecks in the labeling process and generate optimization decisions.
[0109] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0110] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.
[0111] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.
Claims
1. The method for adaptively controlling the gripper for material conveying of a labeling machine is characterized in that: The method comprises: Connecting to a labeling machine, identifying a target product and multiple associated target labels, and extracting product feature indicators and label feature indicators, wherein the label feature indicators include shape features, size features, and material features; Using soft robot technology and combining the product characteristic indicators, the material picking and conveying gripper structure of the labeling machine is designed, and the material picking and conveying gripper structure includes a product grabbing posture and a product releasing posture; Based on the gripper structure for material retrieving and conveying, multi-source monitoring equipment is integrated for perception optimization, and a machine learning model is adopted to autonomously learn and optimize the product grabbing path and product releasing path of the gripper for material retrieving and conveying; The first material picking and conveying stability factor is obtained through the product grabbing posture and the product grabbing path; the second material picking and conveying stability factor is obtained through the product releasing posture and the product releasing path; According to the label characteristic index and the surface characteristics of the product, combined with historical labeling examples, predict the label stability evaluation result during the attachment process; Based on the label stability evaluation result, the negative effect of adhesion is introduced through the first material retrieving and conveying stability factor and the second material retrieving and conveying stability factor, and the product grabbing posture and product grabbing path, product release posture and product release path are adaptively optimized with the use cycle of the target product as a constraint condition to obtain the optimal control strategy; Among them, the target product and the associated multiple target tags are identified, and the product feature indicators and tag feature indicators are extracted, including: Extract product characteristic indicators based on target products of different forms; According to the product characteristic indicators, combined with the material characteristics of the soft robot, the gripper structure is adaptively designed with the adaptation to target products of different forms as the limiting condition; The product grabbing path and product releasing path are updated through the path planning algorithm, and the sliding mode control algorithm is used to dynamically adjust the grabbing force of the material conveying gripper according to the feedback information during the grabbing process; Among them, based on target products of different forms, product characteristic indicators are extracted, including: Based on the product characteristic indicators, the material of the target product is identified using a near-infrared spectroscopy analysis algorithm to obtain the surface characteristics of the product; According to the surface characteristics of the product, a fuzzy logic control algorithm is used to correct the gripping force of the gripper for material collection and conveying; After completing the gripping force calibration of the material picking and conveying clamp, combined with the force sensing algorithm, and the first material picking and conveying stability factor and the second material picking and conveying stability factor, the end point of the updated product gripping path and the starting point of the updated product release path are adjusted according to the label attachment position.
2. The method for adaptively controlling the gripper for material feeding and conveying of a labeling machine according to claim 1, characterized in that: The method further comprises: On the target product production line, a high-speed visual recognition algorithm is used to synchronously track the target product, and the product grabbing path and product release path are planned when the target product production line is in operation; Using a model predictive control algorithm, according to the speed change of the target product production line, the motion trajectory of the product grabbing posture and the motion trajectory of the product releasing posture are adjusted; Predict the speed change trend of the target product production line, and optimize the adaptability of the product grabbing path and the motion trajectory of the product grabbing posture, the product releasing path and the motion trajectory of the product releasing posture under the running state of the target product production line.
3. The method for adaptively controlling the gripper for material feeding and conveying of a labeling machine according to claim 2, characterized in that: Based on the material-taking and conveying gripper structure, multi-source monitoring equipment is integrated to perform perception optimization, and the method further includes: Using environmental perception algorithms to identify interference factors in the environment, including vibration and dust; Based on the interference factors, an adaptive compensation mechanism is set, and the adaptive compensation mechanism is used to dynamically adjust the gripping force of the material-taking and conveying clamping claw; Based on the label stability evaluation result, after the adaptive compensation mechanism is started, a new labeling task is synchronously sent to the labeling machine.
4. The method for adaptively controlling the gripper for material feeding and conveying of a labeling machine according to claim 3, characterized in that: According to the label characteristic index and the product surface characteristics, combined with historical labeling examples, the label stability evaluation result during the attachment process is predicted, and the method further includes: Identify and classify various defects that occur during the labeling process to obtain a classification set of labeling defects; Based on the attachment defect classification set, attaching defect tracing is performed in combination with the historical labeling instance, and the attachment defect causes are analyzed to determine the attachment defect factors; Based on the attachment defect factors, the labeling parameters of the labeling machine are adjusted by feedback control, and the labeling parameters include labeling speed and labeling pressure.
5. The method for adaptively controlling the gripper for material feeding and conveying of a labeling machine according to claim 1, characterized in that: The method comprises: Use IoT technology to collect data during the labeling process and upload it to the cloud; By using a big data analysis algorithm, the uploaded data is analyzed to extract key factors associated with labeling efficiency, including a product conveying speed of a target product and a label supply speed corresponding to a plurality of target labels associated with the target product; Based on the key factors, a decision tree algorithm is used to identify potential efficiency bottlenecks in the labeling process and generate optimized decisions.
6. The self-adaptive control device for the gripper used for material conveying of the labeling machine is characterized in that: For executing the method according to any one of claims 1 to 5, the device comprises: A feature index extraction module, which is used to connect to a labeling machine, identify a target product and multiple associated target labels, and extract product feature indicators and label feature indicators, wherein the label feature indicators include shape features, size features, and material features; A clamp structure design module, wherein the clamp structure design module is used to design a material picking and conveying clamp structure of the labeling machine by using soft robot technology and combining the product characteristic indicators, wherein the material picking and conveying clamp structure includes a product grabbing posture and a product releasing posture; An autonomous learning optimization module, which is based on the gripper structure for material retrieving and conveying, integrates multi-source monitoring equipment for perception optimization, and adopts a machine learning model to autonomously learn and optimize the product grabbing path and product releasing path of the gripper for material retrieving and conveying; A material picking and conveying stability factor acquisition module, which is used to acquire a first material picking and conveying stability factor through a product grabbing posture and a product grabbing path; and acquire a second material picking and conveying stability factor through a product releasing posture and a product releasing path; A stability assessment result prediction module, which is used to predict the label stability assessment result during the attachment process based on the label characteristic index and the product surface characteristics in combination with historical labeling examples; The optimal control strategy acquisition module is based on the label stability evaluation result, introduces the negative impact of adhesion through the first material picking and conveying stability factor and the second material picking and conveying stability factor, and takes the use cycle of the target product as a constraint condition to adaptively optimize the product grasping posture and product grasping path, product release posture and product release path to obtain the optimal control strategy.
