Method and system for optimizing dynamic stacking path planning of corrugated board finished products

Abstract

The present application relates to the technical field of corrugated board production, and specifically discloses a corrugated board finished product dynamic stacking path planning optimization method and system, real-time data is obtained through a visual recognition system, an optimal initial stacking mode is matched in combination with pallet size and order requirements, a laminated compression deformation model is established, deformation compensation is predicted in real time to adjust the placement height of a mechanical arm, path searching is performed in three-dimensional space, time is shortest, energy consumption is lowest, there is no collision and motion is smooth, the stack height is monitored in real time and an error value is generated, if the threshold is exceeded, the action is interrupted and path re-planning is triggered, and the error is fed back to the deformation model for self-learning optimization. Through the steps of combining visual recognition, stack type matching, deformation compensation, path planning, real-time monitoring and feedback optimization, the present application realizes accurate planning and optimization of the corrugated board finished product dynamic stacking path, so as to ensure that the stacking process is efficient and stable, the stack is flat and meets the order requirements.

Description

Technical Field

[0001] This invention relates to the field of corrugated cardboard production technology, and in particular to a method and system for dynamic palletizing path planning and optimization of finished corrugated cardboard products. Background Technology

[0002] In the final stage of corrugated cardboard production, the efficiency and stability of palletizing operations directly affect the smoothness of subsequent warehousing and transportation. However, the current corrugated cardboard palletizing process still faces many technical bottlenecks, which restrict the improvement of production efficiency.

[0003] Firstly, in terms of initial stacking pattern planning, traditional methods rely heavily on the experience and judgment of operators, lacking a scientific and systematic decision-making basis. Due to the diverse specifications of corrugated cardboard and varying order requirements, manually selected stacking patterns often fail to balance pallet space utilization and stack stability. While some solutions improve area utilization, insufficient interlayer friction leads to cardboard slippage, while others prioritize stability but waste pallet space due to loose layouts. Furthermore, the estimation of the pallet's maximum load capacity lacks a dynamic correlation with the edge crush strength of the cardboard, easily leading to the collapse of lower layers due to overload.

[0004] Secondly, the deformation issues inherent in corrugated cardboard remain a key concern for palletizing quality. Corrugated cardboard, composed of multiple layers of bonded paper, has limited compressive strength. Under continuous pressure, it undergoes compressive deformation, which intensifies with the number of pallet layers and the accumulation of weight on the upper layers. Adding to the complexity, this deformation is also affected by environmental factors. Changes in temperature and humidity alter the cardboard's flexibility, and prolonged palletizing operations amplify deformation due to creep. Traditional palletizing systems lack accurate deformation prediction models, and robotic arms consistently perform placement actions at a fixed height. This leads to deformation of the lower layers, causing the upper layers to gradually deviate from their preset positions, disrupting the flatness of the stack's surface, and in severe cases, causing the stack to tilt.

[0005] Therefore, there is an urgent need for a dynamic palletizing path planning and optimization method and system for finished corrugated cardboard products to solve the above problems. Summary of the Invention

[0006] The purpose of this invention is to provide a method for dynamic palletizing path planning and optimization of finished corrugated cardboard, comprising the following steps:

[0007] The system acquires real-time data of corrugated cardboard that is about to enter the palletizing station. The real-time data includes the current number of palletized layers and the total weight of the current layer.

[0008] Based on the current pallet size and order requirements, the optimal initial stacking pattern is matched from the pre-stored stacking type database, and the corresponding first layer stacking coordinate point set is generated;

[0009] A compression deformation model for corrugated cardboard is established. Based on the current number of stacked layers, the total weight of the current layer, and the compressive strength coefficient of the cardboard material, the overall deformation compensation of the current stacked layer is predicted in real time by calculation.

[0010] Starting from the current gripping position of the robotic arm and ending at the target placement point after dynamic height compensation, a path search is performed in three-dimensional space. During the robotic arm's stacking action, the actual height data of the stack is compared with the theoretical model calculated in real time to generate a height error value.

[0011] If the absolute value of the height error exceeds the preset threshold, the current action is immediately interrupted, path replanning is triggered, and the height error value is fed back to the deformation model for self-learning optimization, updating the model parameters, until the error is eliminated and execution continues.

[0012] Furthermore, the step of matching the optimal initial stacking pattern from the pre-stored pallet type database based on the current pallet size and order requirements, and generating the corresponding first layer stacking coordinate point set, includes:

[0013] Real-time measurement of the effective bearing surface length and width of the pallet to obtain the total number of cardboard layers required by the order, the maximum stacking height of a single layer, and the edge crush strength value of the cardboard;

[0014] The maximum allowable lamination load is calculated based on the maximum load-bearing area of ​​the pallet and the edge crush strength of the cardboard, combined with a preset safety factor.

[0015] Traverse the stack type database and filter candidate stacking modes that meet the requirements of single-layer stacking area utilization rate and inter-layer friction coefficient being less than preset values, and theoretical total height being less than the order limit;

[0016] Candidate solutions are evaluated using a multi-dimensional weighted score based on area utilization and friction coefficient, and the solution with the highest score is selected as the optimal stacking mode.

[0017] Using the geometric center of the pallet as a reference, a set of coordinate points is generated according to the optimal pattern arrangement rules, combined with the cardboard size and preset gap value.

[0018] Furthermore, the step of establishing a corrugated cardboard compression deformation model, which calculates and predicts the overall deformation compensation of the current stacked layer in real time based on the current number of stacked layers, the total weight of the current layer, and the compressive strength coefficient of the cardboard material, includes:

[0019] Real-time acquisition of the compressive strength coefficient of cardboard material and the average thickness of a single layer of cardboard;

[0020] The basic deformation is obtained based on the current number of stacked layers, the average thickness of a single layer of cardboard, the total weight of the current layer, and the compressive strength coefficient of the cardboard material.

[0021] Based on the duration of palletizing operations and environmental temperature and humidity data, a pre-stored creep influencing factor comparison table is called to make timely corrections to the basic deformation variables;

[0022] The actual surface height of the stacked stacks is scanned by a laser rangefinder installed above the palletizer, and the deviation from the theoretical height is calculated.

[0023] The height deviation value is fed back to the corrugated cardboard compression deformation model, and the overall deformation compensation amount is obtained by using the gradient descent method.

[0024] Furthermore, the step of performing path search in three-dimensional space, starting from the current grasping position of the robotic arm and ending at the target placement point after dynamic height compensation, includes:

[0025] A 3D grid map containing the robotic arm, the current stack, and the conveyor equipment is constructed based on the point cloud data of the depth camera. The edge of the stack is extended outward by a preset safe distance as the obstacle area, and the coordinates of all static obstacles during the movement of the robotic arm are marked.

[0026] A path optimization algorithm is used to search for paths in a 3D grid map. The path optimization algorithm simultaneously optimizes the dual objective function of the shortest time and the lowest energy consumption. The energy consumption calculation includes the power consumption of the motors of each joint of the robotic arm and the standby power consumption of the servo system.

[0027] During the movement of the robotic arm along the planned path, the changes in the height of the stack and the status of the surrounding environment are monitored in real time. When the distance between the preset path and the obstacle is less than the safety threshold, local path replanning is immediately triggered, and a new collision-free path is generated using the dynamic window method.

[0028] The discrete path point sequence is fitted into a continuous and smooth motion trajectory using cubic spline interpolation.

[0029] The system monitors the motor current values ​​of each joint of the robotic arm in real time. When the current exceeds the preset ratio of the rated value, it automatically reduces the movement speed and replans the path.

[0030] Furthermore, the step of generating a height error value by comparing the actual height data of the stack with the theoretically calculated height during the stacking operation performed by the robotic arm includes:

[0031] A 3D laser scanner installed above the palletizing station performs a panoramic scan of the upper surface of the pallet at a preset sampling frequency to obtain point cloud data containing spatial coordinates. A filtering algorithm is then used to remove noise points and extract effective height data.

[0032] Plane fitting is performed on the processed point cloud data to calculate the average height, maximum height deviation and surface flatness index of the stack surface. The surface flatness is obtained by calculating the root mean square deviation between each measurement point and the fitted plane.

[0033] Based on the current number of stacked layers, the nominal thickness of a single layer of cardboard, and historical data on deformation compensation, the theoretical height value is dynamically calculated. The historical data on deformation compensation includes the average deviation between the actual height and the theoretical height recorded during the previous stacking process, in order to obtain the height error value.

[0034] Furthermore, the step of immediately interrupting the current action, triggering path replanning, and feeding the height error value back to the deformation model for self-learning optimization and updating the model parameters until the error is eliminated before continuing execution includes:

[0035] Based on the absolute value of the altitude error, the error is divided into three levels, each triggering a different response mechanism.

[0036] Among them, the first-level error only records the deviation data and updates the historical mean database;

[0037] Second-level error triggers local path replanning and adjusts deformation compensation coefficient;

[0038] Level 3 error immediately interrupts the operation and initiates a comprehensive inspection procedure, replans the path, and feeds the height error value back to the deformation model for self-learning optimization, updates the model parameters, and continues execution only after the error is eliminated;

[0039] Record pressure distribution data, ambient temperature and humidity, and cardboard batch information in the three-level error events to establish a fault knowledge base based on case-based reasoning.

[0040] Furthermore, this invention also discloses a dynamic palletizing path planning and optimization system for finished corrugated cardboard products, comprising:

[0041] The acquisition module is used to acquire real-time data of corrugated cardboard that is about to enter the palletizing station. The real-time data includes the current number of palletized layers and the total weight of the current layer.

[0042] The generation module is used to match the optimal initial stacking mode from the pre-stored stacking type database according to the current pallet size and order requirements, and generate the corresponding first layer stacking coordinate point set;

[0043] A module is established to create a compression deformation model for corrugated cardboard. Based on the current number of stacked layers, the total weight of the current layer, and the compressive strength coefficient of the cardboard material, the overall deformation compensation of the current stacked layer is predicted in real time by calculation.

[0044] The search module is used to perform path search in three-dimensional space, starting from the current grasping position of the robotic arm and ending at the target placement point after dynamic height compensation.

[0045] The comparison module is used to compare the actual height data of the stack with the theoretical model calculated during the stacking action of the robotic arm, and generate a height error value.

[0046] The learning module is used to immediately interrupt the current action, trigger path replanning, and feed the height error value back to the deformation model for self-learning optimization, update the model parameters, and continue execution until the error is eliminated.

[0047] Furthermore, the establishment module includes:

[0048] The first acquisition unit is used to acquire the compressive strength coefficient of the cardboard material and the average thickness of a single layer of cardboard in real time.

[0049] The second acquisition unit is used to acquire basic deformation based on the current number of stacked layers, the average thickness of a single layer of cardboard, the total weight of the current layer, and the compressive strength coefficient of the cardboard material.

[0050] The correction unit is used to call a pre-stored creep influence factor comparison table to make timely corrections to the basic deformation based on the duration of the palletizing operation and the ambient temperature and humidity data.

[0051] The calculation unit is used to scan the actual surface height of the stacked stacks using a laser rangefinder sensor mounted above the palletizer, and calculate the deviation between the actual height and the theoretical height.

[0052] The third acquisition unit is used to feed back the height deviation value to the corrugated cardboard compression deformation model and obtain the overall deformation compensation amount using the gradient descent method.

[0053] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described dynamic palletizing path planning optimization method for corrugated cardboard finished products.

[0054] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described method for optimizing the dynamic palletizing path of corrugated cardboard finished products.

[0055] The beneficial effects of this application are as follows:

[0056] Firstly, this invention provides a systematic solution to the traditional challenges of corrugated cardboard stacking. In the initial stacking planning, a scientific matching model balances pallet space utilization and stack stability, avoiding the drawbacks of manual experience-based selection. This reduces space waste and lowers the risk of stack slippage. Regarding cardboard deformation, the dynamic deformation compensation model comprehensively considers factors such as the number of layers, weight, time, and environment. Combined with real-time feedback, it precisely adjusts the placement height of the robotic arm, effectively ensuring the flatness of the stack's surface and resolving the tilting problem caused by accumulated deformation.

[0057] Secondly, this invention also achieves synergistic optimization of time, energy consumption, safety, and smoothness of movement, avoiding collision risks, reducing energy consumption, shortening single stacking time, eliminating the impact of movement on the stack, and enabling rapid response to environmental changes. Furthermore, the error handling mechanism achieves full-process error control through hierarchical response and closed-loop optimization, which avoids the accumulation of small errors, reduces unnecessary downtime, lowers the failure rate and the probability of recurrence of similar errors, and significantly improves the continuity, efficiency, stability, and accuracy of palletizing operations. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of a method flow proposed in an embodiment of this application.

[0059] Figure 2 This is a schematic diagram of the system structure proposed in an embodiment of the present invention.

[0060] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0061] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0062] like Figure 1 As shown, this application provides a method for optimizing the dynamic palletizing path planning of finished corrugated cardboard, including the following steps:

[0063] S1, by using a vision recognition system installed upstream of the conveyor line, real-time data of the corrugated cardboard that is about to enter the palletizing station is obtained in real time. The real-time data includes the current number of palletized layers and the total weight of the current layer.

[0064] S2, based on the current pallet size and order requirements, matches the optimal initial stacking mode from the pre-stored stacking type database and generates the corresponding first layer stacking coordinate point set;

[0065] S3. Establish a corrugated cardboard compression deformation model. Based on the current number of stacked layers, the total weight of the current layer, and the compressive strength coefficient of the cardboard material, calculate and predict the overall deformation compensation amount of the current stacked layer in real time. This compensation amount is used to dynamically adjust the descent target height of the robotic arm end effector to ensure that the upper surface of the stack remains flat after each placement.

[0066] S4, starting from the current gripping position of the robotic arm and ending at the target placement point after dynamic height compensation, performs path search in three-dimensional space. The path planning must simultaneously meet the following constraints: shortest time, lowest energy consumption, no collision with the stack and surrounding equipment, and smooth movement of each joint of the robotic arm without abrupt changes.

[0067] S5, during the process of the robotic arm performing the stacking action, compares the actual height data of the stack in real time with the height calculated by the theoretical model to generate a height error value;

[0068] S6. If the absolute value of the height error exceeds the preset threshold, the current action is immediately interrupted, path replanning is triggered, and the height error value is fed back to the deformation model for self-learning optimization, updating the model parameters, and execution continues until the error is eliminated.

[0069] As described in steps S1-S6 above, the present invention combines visual recognition, stack type matching, deformation compensation, path planning, and real-time monitoring and feedback optimization to achieve precise planning and optimization of the dynamic palletizing path of finished corrugated cardboard products, so as to ensure that the palletizing process is efficient and stable, and that the stack is flat and meets the order requirements.

[0070] A systematic solution has been developed to address the traditional challenges of corrugated cardboard stacking, demonstrating significant technological effectiveness. In initial stacking planning, a scientific matching model balances pallet space utilization with stack stability, avoiding the drawbacks of manual experience-based selection. This reduces space waste and lowers the risk of stack slippage. Regarding cardboard deformation, a dynamic deformation compensation model comprehensively considers factors such as the number of layers, weight, time, and environment. Combined with real-time feedback, it precisely adjusts the robotic arm's placement height, effectively ensuring the flatness of the stack's surface and resolving the tilting problem caused by accumulated deformation.

[0071] It also achieves synergistic optimization of time, energy consumption, safety, and smoothness of movement, avoiding collision risks, reducing energy consumption, shortening single stacking time, eliminating the impact of movement on the stack, and responding quickly to environmental changes. Furthermore, the error handling mechanism can achieve full-process error control through hierarchical response and closed-loop optimization, which not only avoids the accumulation of small errors but also reduces unnecessary downtime, lowers the failure rate and the probability of the same error recurring, and significantly improves the continuity, efficiency, stability, and accuracy of palletizing operations.

[0072] Corrugated cardboard is relatively soft, and during stacking, as the number of layers increases and weight accumulates, the lower layers are prone to compression deformation. Without compensation, this can lead to uneven placement of the upper layers, affecting the overall stability of the stack. Simultaneously, robotic arms must balance efficiency, energy consumption, and safety during stacking, avoiding collisions with surrounding equipment or already stacked items. Traditional stacking methods often lack dynamic adjustment mechanisms, making it difficult to handle these complex situations and achieve efficient, accurate, and stable stacking operations. This invention addresses this by acquiring data in real-time, dynamically matching stack types, establishing a deformation model for compensation, optimizing path planning, and combining real-time monitoring and feedback adjustments.

[0073] A vision recognition system installed upstream of the conveyor line acquires real-time data on corrugated cardboard that is about to enter the palletizing station. This data includes the current number of palletized layers and the total weight of the current layer. This step provides basic data support for subsequent palletizing planning and deformation compensation, ensuring that subsequent steps can be adjusted based on the actual situation.

[0074] In one embodiment, the step of matching the optimal initial stacking pattern from a pre-stored pallet type database based on the current pallet size and order requirements, and generating the corresponding first-layer stacking coordinate point set, includes:

[0075] S21, measures the effective bearing surface length and width of the pallet in real time to obtain the total number of cardboard layers required by the order, the maximum stacking height of a single layer, and the edge crush strength value of the cardboard;

[0076] S22, based on the maximum load-bearing area of ​​the pallet and the edge crush strength of the cardboard, combined with a preset safety factor, calculates the maximum allowable lamination load;

[0077] S23, Traverse the stack type database and filter candidate stacking modes that meet the requirements of single-layer stacking area utilization rate and inter-layer friction coefficient being less than preset values, and theoretical total height being less than the order limit;

[0078] S24, perform multi-dimensional weighted scoring on candidate schemes based on area utilization rate and friction coefficient, and select the scheme with the highest score as the optimal stacking mode;

[0079] S25, using the geometric center of the pallet as a reference, generates a set of coordinate points according to the optimal pattern arrangement rules, combined with the cardboard size and preset gap value.

[0080] As described in steps S21-S25 above, by measuring pallet parameters in real time, selecting the optimal initial stacking mode in combination with order requirements, and generating a precise set of coordinate points for the first layer of stacking, the scientific planning of the initial layout of corrugated cardboard stacking is achieved, so as to ensure the stability of the stack, improve space utilization, and meet order constraints.

[0081] In corrugated cardboard palletizing operations, the effective load-bearing area of ​​pallets is limited. Orders have specific requirements regarding the total number of layers, the height of each layer, and the edge crush strength of the cardboard. Simultaneously, the stacking pattern directly affects the stability of the stack—insufficient interlayer friction coefficients can easily lead to cardboard slippage, while low area utilization results in wasted pallet space. Traditional palletizing often relies on manual experience to select stacking patterns, lacking a systematic consideration of pallet load-bearing capacity, order constraints, and stacking efficiency, frequently leading to problems such as wasted space, unstable stacks, or failure to meet order requirements. This step constructs a scientific initial stacking planning system through systematic parameter collection, constraint calculation, pattern selection, and coordinate generation, specifically addressing the aforementioned problems.

[0082] First, the effective load-bearing surface length and width of the pallet are measured in real time to obtain the total number of cardboard layers, maximum stacking height per layer, and edge crush strength of the cardboard required by the order. The pallet size data determines the available load-bearing space, while the order parameters clarify the boundary conditions for stacking. For example, if the order requires a maximum stacking height of 30cm per layer, the stacking patterns selected later must strictly meet this limitation. These data provide the basis for matching subsequent stacking patterns.

[0083] Based on the pallet's maximum load-bearing area and the edge crush strength of the cardboard, the maximum allowable lamination load is calculated using a preset safety factor. The calculation formula is: Maximum lamination load = (Pattern's maximum load-bearing area × Cardboard edge crush strength) × Preset safety factor (e.g., 0.8). This clarifies the maximum pressure each layer of cardboard can withstand from a mechanical perspective, preventing excessive deformation of the lower layer due to excessive pressure on the upper layer. For example, for cardboard with an edge crush strength of 5000 N / m, if the pallet's load-bearing area is 1.2 m², with a safety factor of 0.8, the maximum lamination load is 1.2 × 5000 × 0.8 = 4800 N, ensuring the lower layer of cardboard remains within its safe load-bearing range.

[0084] The system iterates through the pallet type database, filtering for candidate stacking patterns that meet the following criteria: single-layer stacking area utilization rate ≥ 85%, inter-layer friction coefficient > 0.4 and less than a preset value (e.g., 0.6), and theoretical total height less than the order height limit. An area utilization rate ≥ 85% significantly improves pallet space utilization and reduces idle space. An inter-layer friction coefficient controlled between 0.4 and 0.6 ensures sufficient friction between layers to prevent slippage while avoiding excessive friction that could lead to difficulties in handling. A theoretical total height meeting order requirements directly satisfies delivery standards. For example, if the order height limit is 2m, the selected patterns must have a theoretical total height less than 2m to ensure that the palletized structure meets transportation or storage requirements.

[0085] Candidate solutions are scored using a multi-dimensional weighted evaluation based on area utilization rate and friction coefficient. The scoring formula is: Overall Score = (Area Utilization Rate × Weight 1) + (Friction Coefficient × Weight 2), where Weight 1 and Weight 2 are set according to order priority. If the order prioritizes space utilization, Weight 1 can be set to 0.6 and Weight 2 to 0.4; otherwise, the weight allocation is adjusted. The solution with the highest score is selected as the optimal stacking mode. This quantitative scoring method objectively compares the merits of different modes, avoiding subjective judgment bias. For example, a candidate mode with an area utilization rate of 90% and a friction coefficient of 0.5 would score 90 × 0.6 + 0.5 × 100 × 0.4 = 54 + 20 = 74 points according to the above weights. If another mode scores lower, the former is selected as the optimal solution.

[0086] Using the geometric center of the pallet as a reference, a set of coordinate points is generated according to the optimal arrangement rules, combined with the cardboard size and a preset gap value (e.g., 2mm). The preset gap value is used to compensate for cardboard size errors and the effects of environmental expansion, ensuring that no compression deformation occurs during stacking. For example, for cardboard with dimensions of 100cm×80cm, on a 120cm×100cm pallet, with the center as the origin, according to the optimal arrangement rule of 2 cards horizontally and 1 cardboard vertically, coordinate points (-50cm, -40cm) and (50cm, -40cm) can be generated, ensuring that the cardboard is neatly arranged on the pallet and providing position guidance for subsequent precise stacking by the robotic arm.

[0087] In one embodiment, the step of establishing a corrugated cardboard compression deformation model, which involves calculating and predicting the overall deformation compensation of the current stacked layer in real time based on the current number of stacked layers, the total weight of the current layer, and the compressive strength coefficient of the cardboard material, includes:

[0088] S31, real-time acquisition of the compressive strength coefficient of the cardboard material and the average thickness of a single layer of cardboard;

[0089] S32 is based on the theory of lamination deformation and obtains the basic deformation based on the current number of stacked layers, the average thickness of a single layer of cardboard, the total weight of the current layer, and the compressive strength coefficient of the cardboard material.

[0090] S33, based on the duration of the palletizing operation and the ambient temperature and humidity data, call the pre-stored creep influence factor comparison table to make timely corrections to the basic deformation variables;

[0091] S34, by using a laser rangefinder sensor installed above the palletizer, scans the actual surface height of the stacked stack and calculates the deviation value between it and the theoretical height;

[0092] S35 feeds back the height deviation value to the corrugated cardboard compression deformation model and uses the gradient descent method to obtain the overall deformation compensation amount.

[0093] As described in steps S31-S35 above, by establishing a corrugated cardboard compression deformation model, and combining real-time parameters to calculate and dynamically correct the overall deformation compensation amount, the precise adjustment of the target height of the robotic arm end effector's descent is achieved, ensuring that the upper surface of the stack remains flat after each placement.

[0094] Corrugated cardboard has a certain degree of flexibility. During the stacking process, as the number of stacked layers increases and the weight of the upper layers accumulates, the lower layers of cardboard will undergo compression deformation due to continuous pressure. The degree of deformation intensifies with the extension of stacking time and changes in ambient temperature and humidity. If this is not compensated for, when the robotic arm places a new cardboard at a fixed height, the height deviation caused by the deformation of the lower layers will result in the new cardboard being placed unevenly, thus affecting the stability of the entire stack.

[0095] First, the compressive strength coefficient of the cardboard material (calibrated by a material pressure testing machine) and the average thickness of a single layer of cardboard are obtained in real time. The compressive strength coefficient of the cardboard material is obtained by pre-calibrating the material pressure testing machine. For example, if a batch of corrugated cardboard is subjected to a pressure test, its compressive strength coefficient is measured to be 3.2 MPa. The average thickness of a single layer of cardboard is obtained by measuring the thickness of 10 sheets of cardboard in the same batch and taking the average value. For example, if the average value is 5 mm, these parameters provide the basic data for subsequent deformation calculation.

[0096] Based on the theory of lamination deformation, and combining the current number of stacked layers, the average thickness of a single layer, the total weight of the current layer, and the compressive strength coefficient of the cardboard material, the basic deformation is obtained. The calculation formula is as follows:

[0097] ;

[0098] in, Represents the basic form variable. This formula represents the total weight of the current layer, x represents the current number of stacked layers, p represents the compressive strength coefficient of the cardboard material, and c represents the average thickness of a single layer of cardboard. From a mechanical perspective, this formula quantifies the initial deformation of multi-layer cardboard under pressure. For example, if 5 layers are currently stacked and the total weight of the current layer is 200N, substituting the above parameters, the basic deformation is calculated as: (200 × 5) / (3.2 × ... (×0.005) = 1000 / 16000 = 0.0625m = 6.25cm, thus clarifying the theoretical basic compression amount.

[0099] Based on the duration of the palletizing operation and ambient temperature and humidity data, a pre-stored creep influence factor reference table is used to make a time-sensitive correction to the basic deformation value. Corrugated cardboard exhibits creep characteristics, meaning that the deformation gradually increases over time, and this phenomenon is exacerbated by high humidity environments. For example, when the operation lasts for 2 hours and the ambient humidity is 60%, the corresponding creep influence factor in the reference table is 1.2. Therefore, the corrected deformation value = 6.25 × 1.2 = 7.5 cm, making the deformation value calculation more consistent with the deformation pattern under actual conditions.

[0100] A laser rangefinder sensor installed above the palletizer scans the actual surface height of the stacked goods and calculates the deviation from the theoretical height. The laser rangefinder sensor can accurately measure the height of various points on the surface of the stack. For example, if the theoretical height is calculated to be 100cm, and the actual average measured height is 98cm, then the height deviation is 98-100=-2cm. This deviation reflects the difference between the theoretical calculation and the actual deformation, providing realistic feedback for model correction.

[0101] The height deviation value is fed back to the corrugated cardboard compression deformation model, and the overall deformation compensation amount is obtained using the gradient descent method. The gradient descent method iteratively adjusts the model parameters so that the calculated values ​​gradually approach the actual values. For example, based on a deviation value of -2cm, after gradient descent iterative calculation, the overall deformation compensation amount is determined to be 9.5cm. This compensation amount is used to adjust the descent target height of the robotic arm's end effector to ensure that the upper surface of the stack remains flat after the new cardboard is placed, avoiding the stack tilting or collapsing due to deformation.

[0102] In one embodiment, the step of performing path search in three-dimensional space, starting from the current grasping position of the robotic arm and ending at the target placement point after dynamic height compensation, includes:

[0103] S41, based on the point cloud data of the depth camera, constructs a three-dimensional grid map containing the robotic arm, the current stack, and the conveyor equipment, extends the edge of the stack outward by a preset safe distance as the obstacle area, and marks the coordinates of all static obstacles during the movement of the robotic arm;

[0104] S42 uses a path optimization algorithm to search for paths in a 3D grid map. The path optimization algorithm simultaneously optimizes the dual objective function of the shortest time and the lowest energy consumption. The energy consumption calculation includes the power consumption of the motors of each joint of the robotic arm and the standby power consumption of the servo system.

[0105] S43: During the movement of the robotic arm along the planned path, the changes in the height of the stack and the status of the surrounding environment are monitored in real time. When the distance between the preset path and the obstacle is less than the safety threshold, the local path replanning is immediately triggered, and a new collision-free path is generated using the dynamic window method.

[0106] S44, the discrete path point sequence is fitted into a continuous and smooth motion trajectory by cubic spline interpolation to ensure that the rate of change of acceleration of each joint of the robotic arm is continuous and to avoid impact and vibration during the motion.

[0107] The S45 monitors the motor current values ​​of each joint of the robotic arm in real time. When the current exceeds the preset ratio of the rated value, it automatically reduces the movement speed and replans the path to ensure that the system power consumption is always within a safe range.

[0108] As described in steps S41-S45 above, the present invention ensures that the path simultaneously meets the constraints of shortest time, lowest energy consumption, no collision and smooth movement by performing robotic arm path search and optimization in three-dimensional space, thereby achieving efficient, safe and stable operation of the robotic arm in the process of corrugated cardboard palletizing.

[0109] In corrugated cardboard palletizing operations, the movement path of the robotic arm directly affects palletizing efficiency, energy consumption, and operational safety. Inadequate path planning can lead to problems such as collisions between the robotic arm and the stack or surrounding equipment, impact vibrations during movement affecting stack stability, excessive energy consumption, or excessively long operation times.

[0110] First, a 3D grid map containing the robotic arm, the current stack, and the conveyor equipment is constructed based on depth camera point cloud data. A predetermined safe distance (e.g., 5cm) is extended outwards from the stack edge as an obstacle area, and the coordinates of all static obstacles during the robotic arm's movement are marked. The depth camera can collect 3D point cloud information of the surrounding environment in real time. Through rasterization, the space is divided into several small cubic units, each marked as either free space or an obstacle. For example, the area where the stack is located and the surrounding 5cm grid are marked as obstacles, providing clear environmental boundaries for subsequent path searching and ensuring that the robotic arm maintains a safe distance from obstacles during movement.

[0111] A path optimization algorithm is used to search for paths in a 3D grid map. This algorithm simultaneously optimizes a dual objective function: minimizing time and energy consumption. Energy consumption calculation includes the power consumption of the motors at each joint of the robotic arm and the standby power consumption of the servo system. The calculation formula is as follows:

[0112] ;

[0113] Among them, the Indicates total energy consumption, the Indicates the power consumption of the joint motor, the Indicates the duration of motion, the This indicates the standby power consumption of the servo system. The time objective function is based on the ratio of the robotic arm's movement distance to its average speed. The dual objective is balanced by weighted coefficients. For example, if the time weight is set to 0.6 and the energy weight to 0.4, the algorithm will prioritize paths with shorter time and lower energy consumption when searching for paths. For example, if a certain path has a slightly longer movement distance, it can reduce the number of joint starts and stops and has lower total energy consumption. After weighted calculation, it may be selected as a better path.

[0114] During the movement of the robotic arm along the planned path, the changes in the height of the stack and the surrounding environment are monitored in real time. When the distance between the preset path and an obstacle is detected to be less than a safety threshold (e.g., 3cm), local path replanning is immediately triggered, and a new collision-free path is generated using the dynamic window method. The dynamic window method samples possible velocity combinations in the velocity space to simulate the movement trajectory of the robotic arm in a short period of time, and selects the collision-free trajectory that meets the dynamic constraints as the local path. For example, when it is suddenly detected that the space below the original path is insufficient due to the increased height of the stack caused by deformation, the dynamic window method can generate a new local path that detours upwards within 0.5 seconds to avoid collision.

[0115] The discrete path point sequence is fitted into a continuous and smooth motion trajectory using cubic spline interpolation. Cubic spline interpolation constructs a piecewise cubic polynomial, ensuring that the trajectories between adjacent path points are not only positionally continuous, but also that their first derivative (velocity) and second derivative (acceleration) are continuous. This ensures the continuous rate of change of acceleration at each joint of the robotic arm, avoiding impacts and vibrations during movement. For example, for path points (0,0,0), (1,1,1), and (2,0,2), after cubic spline interpolation, the robotic arm's velocity and acceleration between these points will transition smoothly without abrupt changes, reducing disturbance to the already stacked objects.

[0116] The system monitors the motor current values ​​of each joint of the robotic arm in real time. When the detected current exceeds a preset percentage (e.g., 80%) of the rated value, it automatically reduces the movement speed and replans the path to ensure that the system power consumption remains within a safe range. Motor current is positively correlated with load torque. Excessive current indicates that the joint load is too high. Reducing the speed can decrease torque output. For example, if the rated current of a joint motor is 10A, when the detected current reaches 8.5A, the system automatically reduces the joint's movement speed by 20% and replans the path to reduce the load and prevent motor overload damage.

[0117] In one embodiment, the step of generating a height error value by comparing the actual height data of the stack with the theoretically calculated height during the stacking action performed by the robotic arm includes:

[0118] S51 uses a 3D laser scanner installed above the palletizing station to perform a panoramic scan of the upper surface of the pallet at a preset sampling frequency, obtain point cloud data containing spatial coordinates, and use a filtering algorithm to remove noise points and extract effective height data.

[0119] S52, perform plane fitting on the processed point cloud data, and calculate the average height value, maximum height deviation and surface flatness index of the stack surface. The surface flatness is obtained by calculating the root mean square deviation between each measurement point and the fitted plane.

[0120] S53. Based on the current number of stacked layers, the nominal thickness of a single layer of cardboard, and historical data on deformation compensation, the theoretical height value is dynamically calculated. The historical data on deformation compensation includes the average deviation between the actual height and the theoretical height recorded during the previous stacking process, in order to obtain the height error value.

[0121] As described in steps S51-S53 above, the present invention generates a height error value by real-time monitoring of the actual height of the stack and comparing it with the height calculated by the theoretical model. This provides a precise basis for subsequent path replanning and deformation model optimization, ensuring the accuracy of stack height control during the stacking process.

[0122] During the corrugated cardboard stacking process, the actual stack height may deviate from the theoretically calculated height due to differences in cardboard material, deviations in deformation model predictions, and changes in environmental factors. If this deviation is not monitored and quantified in a timely manner, the robotic arm will continue to place cardboard at the incorrect height, eventually leading to problems such as stack tilting and collapse.

[0123] First, a 3D laser scanner installed above the palletizing station performs a panoramic scan of the pallet's upper surface at a preset sampling frequency (e.g., 10Hz) to acquire point cloud data containing spatial coordinates. A filtering algorithm (e.g., Gaussian filtering) is then used to remove noise points and extract valid height data. The 3D laser scanner can quickly acquire dense 3D coordinates of the pallet surface. Gaussian filtering smooths out abnormal points caused by environmental interference (such as dust or changes in lighting). For example, after filtering 1000 point cloud data points obtained from the scan, 20 noise points deviating from the normal range are removed, retaining 980 valid height values, providing a reliable data foundation for subsequent calculations.

[0124] The processed point cloud data is fitted with a plane to calculate the average height, maximum height deviation, and surface flatness index of the stack surface. The surface flatness is obtained by calculating the root mean square deviation between each measurement point and the fitted plane. The plane fitting constructs an ideal plane that is closest to the actual surface using the least squares method. For example, after fitting, the average height is 120cm, the maximum height deviation (the difference between a point and the average height) is +3cm, and the root mean square deviation is 1.2cm. The above indexes quantify the flatness of the stack surface. If the root mean square deviation is too large, it indicates that the surface is not flat and needs adjustment.

[0125] Based on the current number of stacked layers, the nominal thickness of a single layer of cardboard, and historical deformation compensation data, the theoretical height is dynamically calculated. The historical deformation compensation data includes the average deviation between the actual height and the theoretical height recorded during previous stacking processes, used to obtain the height error value. The calculation formula is as follows:

[0126] ;

[0127] Among them, the Indicates the theoretical height value, the This indicates the current number of stacked layers. Indicates the nominal thickness of a single-layer cardboard, the This indicates the amount of deformation compensation in the early stage. The height error value represents the average deviation. The height error value is calculated as: Actual Average Height Value - Theoretical Height Value. For example, if 10 layers have been stacked, each layer has a nominal thickness of 5cm, the accumulated deformation compensation is 8cm, and the average deviation is 0.5cm, then the theoretical height value is 10 × 5 + 8 + 0.5 = 58.5cm. If the actual average height value is 59.2cm, then the height error value is 59.2 - 58.5 = 0.7cm. The height error value directly reflects the degree of deviation between the actual and theoretical heights, providing a basis for determining whether to trigger path replanning.

[0128] In one embodiment, the step of immediately interrupting the current action, triggering path replanning, and feeding the height error value back to the deformation model for self-learning optimization and updating the model parameters until the error is eliminated before continuing execution includes:

[0129] S61 divides the error into three levels based on the absolute value of the altitude error, triggering different response mechanisms for each level.

[0130] S62, where the first-level error only records the deviation data and updates the historical mean database;

[0131] S63, Level 2 error triggers local path replanning and adjusts deformation compensation coefficient;

[0132] S64, Level 3 error immediately interrupts the operation and initiates a full inspection procedure, replans the path, and feeds the height error value back to the deformation model for self-learning optimization, updates the model parameters, and continues execution only after the error is eliminated;

[0133] S65 records pressure distribution data, ambient temperature and humidity, and cardboard batch information in three-level error events, and establishes a fault knowledge base based on case-based reasoning.

[0134] As described in steps S61-S65 above, the present invention achieves dynamic correction of height deviation during the palletizing process by hierarchically responding to height error values ​​and combining path replanning with deformation model self-learning optimization, ensuring that the pallet remains flat and improving the stability and fault tolerance of the palletizing system.

[0135] During the corrugated cardboard stacking process, the actual stack height inevitably deviates from the theoretical height due to factors such as cardboard deformation characteristics, environmental interference, and model prediction bias. If the accumulated error exceeds a certain range, it will cause subsequent stacked cardboard to continuously deviate from the preset position, ultimately leading to risks such as stack tilting and collapse. Traditional technologies lack a graded error handling mechanism and often adopt a uniform shutdown and adjustment method, which either affects efficiency due to overreaction or amplifies risks due to insufficient adjustment.

[0136] First, based on the absolute value of the height error, the error is divided into three levels, each triggering a different response mechanism. The level division is determined based on preset thresholds. For example, the level 1 error threshold is set to ≤1cm, the level 2 error to 1-3cm, and the level 3 error to ≥3cm. Different levels correspond to different levels of risk, providing clear standards for differentiated handling.

[0137] The first-level error only records deviation data and updates the historical average database. When the error value is within 1cm, it indicates that the deviation is small and does not pose a threat to the stability of the stack. There is no need to interrupt the operation. For example, if a height error value of 0.8cm is detected, the system will automatically store the data in the historical average database for deviation correction in subsequent theoretical height calculations. By accumulating data, the system continuously optimizes the basic parameters of the theoretical model and improves the long-term prediction accuracy.

[0138] Secondary errors trigger local path replanning and adjust deformation compensation coefficients. When the error value is between 1-3cm, it indicates that the deviation has affected the local flatness of the stack and needs to be corrected without interrupting the current layer of stacking. For example, when the error value is 2cm, the system immediately triggers local path replanning of the robotic arm, adjusts the placement path of subsequent cardboard using the dynamic window method, and substitutes the error value into the deformation model to adjust the compensation coefficient (e.g., adjust the original compensation coefficient of 1.2 to 1.3) to make the placement height compensation of the next cardboard more accurate and gradually eliminate the deviation.

[0139] Level 3 errors immediately halt operations and initiate a comprehensive inspection procedure. The path is replanned, and the height error value is fed back to the deformation model for self-learning optimization, updating model parameters until the error is eliminated before resuming execution. When the error value reaches 3cm or above, it indicates a significant risk of stack tilting, requiring emergency handling. For example, when the error value is 4cm, the system immediately stops the robotic arm's movement and initiates a comprehensive inspection, including manual verification of cardboard damage and pallet misalignment. After eliminating anomalies, the entire stacking path is replanned, and the 4cm error value is input into the deformation model. The creep influence factor, compressive strength coefficient, and other parameters in the model are iteratively optimized using the gradient descent method. For example, the humidity influence factor is adjusted from 1.2 to 1.25, enabling the model to more accurately predict the degree of deformation under high humidity conditions. After the model parameters are updated and the error simulation is eliminated, the palletizing operation is restarted.

[0140] Record pressure distribution data, ambient temperature and humidity, and cardboard batch information for Level 3 error events to establish a fault knowledge base based on case-based reasoning. Pressure distribution data is acquired through pressure sensors installed at the bottom of the pallet, ambient temperature and humidity are collected through workshop temperature and humidity sensors, and cardboard batch information is associated with material identifiers in the order. For example, recording a Level 3 error event: uneven pressure distribution (maximum pressure point 250N), ambient humidity 70%, cardboard batch A032. After this information is stored in the knowledge base, when encountering the same or similar working conditions later, the system can quickly recall historical processing solutions, shorten error handling time, and improve the system's fault response capability.

[0141] In one embodiment, the present invention also discloses a dynamic palletizing path planning and optimization system for finished corrugated cardboard products, comprising:

[0142] The acquisition module 1 is used to acquire real-time data of the corrugated cardboard that is about to enter the palletizing station. The real-time data includes the current number of palletized layers and the total weight of the current layer.

[0143] The generation module 2 is used to match the optimal initial stacking mode from the pre-stored stacking type database according to the current pallet size and order requirements, and generate the corresponding first layer stacking coordinate point set;

[0144] Module 3 is established to create a compression deformation model for corrugated cardboard. Based on the current number of stacked layers, the total weight of the current layer, and the compressive strength coefficient of the cardboard material, the overall deformation compensation of the current stacked layer is predicted in real time by calculation.

[0145] Search module 4 is used to perform path search in three-dimensional space, starting from the current grasping position of the robotic arm and ending at the target placement point after dynamic height compensation.

[0146] Comparison module 5 is used to compare the actual height data of the stack with the theoretical model calculated during the stacking action of the robotic arm to generate a height error value.

[0147] Learning module 6 is used to immediately interrupt the current action, trigger path replanning, and feed the height error value back to the deformation model for self-learning optimization, update the model parameters, and continue execution after the error is eliminated.

[0148] In one embodiment, the establishment module includes:

[0149] The first acquisition unit is used to acquire the compressive strength coefficient of the cardboard material and the average thickness of a single layer of cardboard in real time.

[0150] The second acquisition unit is used to acquire basic deformation based on the current number of stacked layers, the average thickness of a single layer of cardboard, the total weight of the current layer, and the compressive strength coefficient of the cardboard material.

[0151] The correction unit is used to call a pre-stored creep influence factor comparison table to make timely corrections to the basic deformation based on the duration of the palletizing operation and the ambient temperature and humidity data.

[0152] The calculation unit is used to scan the actual surface height of the stacked stacks using a laser rangefinder sensor mounted above the palletizer, and calculate the deviation between the actual height and the theoretical height.

[0153] The third acquisition unit is used to feed back the height deviation value to the corrugated cardboard compression deformation model and obtain the overall deformation compensation amount using the gradient descent method.

[0154] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described dynamic palletizing path planning optimization method for corrugated cardboard finished products.

[0155] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described method for optimizing the dynamic palletizing path of corrugated cardboard finished products.

[0156] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0157] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0158] The above description is merely a preferred embodiment of the present invention and does not limit the scope of this application. Any equivalent results or equivalent process transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the scope of protection of this application.

Claims

1. A method for optimizing the dynamic palletizing path of finished corrugated cardboard, characterized in that, Includes the following steps: The system acquires real-time data of corrugated cardboard that is about to enter the palletizing station. The real-time data includes the current number of palletized layers and the total weight of the current layer. Based on the current pallet size and order requirements, the optimal initial stacking pattern is matched from the pre-stored stacking type database, and the corresponding first layer stacking coordinate point set is generated; A compression deformation model for corrugated cardboard is established. Based on the current number of stacked layers, the total weight of the current layer, and the compressive strength coefficient of the cardboard material, the overall deformation compensation of the current stacked layer is calculated and predicted in real time. The specific steps include: Real-time acquisition of the compressive strength coefficient of cardboard material and the average thickness of a single layer of cardboard; The basic deformation is obtained based on the current number of stacked layers, the average thickness of a single layer of cardboard, the total weight of the current layer, and the compressive strength coefficient of the cardboard material. Based on the duration of palletizing operations and environmental temperature and humidity data, a pre-stored creep influencing factor comparison table is called to make timely corrections to the basic deformation variables; The actual surface height of the stacked stacks is scanned by a laser rangefinder installed above the palletizer, and the deviation from the theoretical height is calculated. The height deviation value is fed back to the corrugated cardboard compression deformation model, and the overall deformation compensation amount is obtained by using the gradient descent method. Starting from the current gripping position of the robotic arm and ending at the target placement point after dynamic height compensation, a path search is performed in three-dimensional space. The specific steps include: A 3D grid map containing the robotic arm, the current stack, and the conveyor equipment is constructed based on the point cloud data of the depth camera. The edge of the stack is extended outward by a preset safe distance as the obstacle area, and the coordinates of all static obstacles during the movement of the robotic arm are marked. A path optimization algorithm is used to search for paths in a 3D grid map. The path optimization algorithm simultaneously optimizes the dual objective function of the shortest time and the lowest energy consumption. The energy consumption calculation includes the power consumption of the motors of each joint of the robotic arm and the standby power consumption of the servo system. During the movement of the robotic arm along the planned path, the changes in the height of the stack and the status of the surrounding environment are monitored in real time. When the distance between the preset path and the obstacle is less than the safety threshold, local path replanning is immediately triggered, and a new collision-free path is generated using the dynamic window method. The discrete path point sequence is fitted into a continuous and smooth motion trajectory using cubic spline interpolation. The system monitors the motor current values ​​of each joint of the robotic arm in real time. When the current exceeds the preset ratio of the rated value, it automatically reduces the movement speed and replans the path. During the stacking process, the actual height data of the stack is compared with the height calculated by the theoretical model in real time to generate a height error value. If the absolute value of the height error exceeds a preset threshold, the current action is immediately interrupted, path replanning is triggered, and the height error value is fed back to the deformation model for self-learning optimization, updating the model parameters, until the error is eliminated and execution continues. The specific steps include: Based on the absolute value of the altitude error, the error is divided into three levels, each triggering a different response mechanism. Among them, the first-level error only records the deviation data and updates the historical mean database; Second-level error triggers local path replanning and adjusts deformation compensation coefficient; Level 3 error immediately interrupts the operation and initiates a comprehensive inspection procedure, replans the path, and feeds the height error value back to the deformation model for self-learning optimization, updates the model parameters, and continues execution only after the error is eliminated; Record pressure distribution data, ambient temperature and humidity, and cardboard batch information in the three-level error events to establish a fault knowledge base based on case-based reasoning.

2. The method for dynamic palletizing path planning and optimization of corrugated cardboard finished products according to claim 1, characterized in that, The step of matching the optimal initial stacking pattern from the pre-stored pallet type database based on the current pallet size and order requirements, and generating the corresponding first layer stacking coordinate point set, includes: Real-time measurement of the effective bearing surface length and width of the pallet to obtain the total number of cardboard layers required by the order, the maximum stacking height of a single layer, and the edge crush strength value of the cardboard; The maximum allowable lamination load is calculated based on the maximum load-bearing area of ​​the pallet and the edge crush strength of the cardboard, combined with a preset safety factor. Traverse the stack type database and filter candidate stacking modes that meet the requirements of single-layer stacking area utilization rate and inter-layer friction coefficient being less than preset values, and theoretical total height being less than the order limit; Candidate solutions are evaluated using a multi-dimensional weighted score based on area utilization and friction coefficient, and the solution with the highest score is selected as the optimal stacking mode. Using the geometric center of the pallet as a reference, a set of coordinate points is generated according to the optimal pattern arrangement rules, combined with the cardboard size and preset gap value.

3. The method for dynamic palletizing path planning and optimization of corrugated cardboard finished products according to claim 1, characterized in that, The step of generating a height error value by comparing the actual height data of the stack with the theoretically calculated height during the stacking operation performed by the robotic arm includes: A 3D laser scanner installed above the palletizing station performs a panoramic scan of the upper surface of the pallet at a preset sampling frequency to obtain point cloud data containing spatial coordinates. A filtering algorithm is then used to remove noise points and extract effective height data. Plane fitting is performed on the processed point cloud data to calculate the average height, maximum height deviation and surface flatness index of the stack surface. The surface flatness is obtained by calculating the root mean square deviation between each measurement point and the fitted plane. Based on the current number of stacked layers, the nominal thickness of a single layer of cardboard, and historical data on deformation compensation, the theoretical height value is dynamically calculated. The historical data on deformation compensation includes the average deviation between the actual height and the theoretical height recorded during the previous stacking process, in order to obtain the height error value.

4. A dynamic palletizing path planning and optimization system for finished corrugated cardboard products, characterized in that: include: The acquisition module is used to acquire real-time data of corrugated cardboard that is about to enter the palletizing station. The real-time data includes the current number of palletized layers and the total weight of the current layer. The generation module is used to match the optimal initial stacking mode from the pre-stored stacking type database according to the current pallet size and order requirements, and generate the corresponding first layer stacking coordinate point set; A module is established to create a compression deformation model for corrugated cardboard. Based on the current number of stacked layers, the total weight of the current layer, and the compressive strength coefficient of the cardboard material, the module calculates and predicts the overall deformation compensation of the current stacked layer in real time. Specific steps include: Real-time acquisition of the compressive strength coefficient of cardboard material and the average thickness of a single layer of cardboard; The basic deformation is obtained based on the current number of stacked layers, the average thickness of a single layer of cardboard, the total weight of the current layer, and the compressive strength coefficient of the cardboard material. Based on the duration of palletizing operations and environmental temperature and humidity data, a pre-stored creep influencing factor comparison table is called to make timely corrections to the basic deformation variables; The actual surface height of the stacked stacks is scanned by a laser rangefinder installed above the palletizer, and the deviation from the theoretical height is calculated. The height deviation value is fed back to the corrugated cardboard compression deformation model, and the overall deformation compensation amount is obtained by using the gradient descent method. The search module is used to perform path searching in three-dimensional space, starting from the current grasping position of the robotic arm and ending at the target placement point after dynamic height compensation. Specific steps include: A 3D grid map containing the robotic arm, the current stack, and the conveyor equipment is constructed based on the point cloud data of the depth camera. The edge of the stack is extended outward by a preset safe distance as the obstacle area, and the coordinates of all static obstacles during the movement of the robotic arm are marked. A path optimization algorithm is used to search for paths in a 3D grid map. The path optimization algorithm simultaneously optimizes the dual objective function of the shortest time and the lowest energy consumption. The energy consumption calculation includes the power consumption of the motors of each joint of the robotic arm and the standby power consumption of the servo system. During the movement of the robotic arm along the planned path, the changes in the height of the stack and the status of the surrounding environment are monitored in real time. When the distance between the preset path and the obstacle is less than the safety threshold, local path replanning is immediately triggered, and a new collision-free path is generated using the dynamic window method. The discrete path point sequence is fitted into a continuous and smooth motion trajectory using cubic spline interpolation. The system monitors the motor current values ​​of each joint of the robotic arm in real time. When the current exceeds the preset ratio of the rated value, it automatically reduces the movement speed and replans the path. The comparison module is used to compare the actual height data of the stack with the theoretical model calculated during the stacking action of the robotic arm, and generate a height error value. The learning module is used to immediately interrupt the current action and trigger path replanning if the absolute value of the height error exceeds a preset threshold. It then feeds the height error value back to the deformation model for self-learning optimization, updating the model parameters until the error is eliminated, at which point execution continues. Specific steps include: Based on the absolute value of the altitude error, the error is divided into three levels, each triggering a different response mechanism. Among them, the first-level error only records the deviation data and updates the historical mean database; Second-level error triggers local path replanning and adjusts deformation compensation coefficient; Level 3 error immediately interrupts the operation and initiates a comprehensive inspection procedure, replans the path, and feeds the height error value back to the deformation model for self-learning optimization, updates the model parameters, and continues execution only after the error is eliminated; Record pressure distribution data, ambient temperature and humidity, and cardboard batch information in the three-level error events to establish a fault knowledge base based on case-based reasoning.

5. The dynamic palletizing path planning and optimization system for corrugated cardboard products according to claim 4, characterized in that, The establishment module includes: The first acquisition unit is used to acquire the compressive strength coefficient of the cardboard material and the average thickness of a single layer of cardboard in real time. The second acquisition unit is used to acquire basic deformation based on the current number of stacked layers, the average thickness of a single layer of cardboard, the total weight of the current layer, and the compressive strength coefficient of the cardboard material. The correction unit is used to call a pre-stored creep influence factor comparison table to make timely corrections to the basic deformation based on the duration of the palletizing operation and the ambient temperature and humidity data. The calculation unit is used to scan the actual surface height of the stacked stacks using a laser rangefinder sensor mounted above the palletizer, and calculate the deviation between the actual height and the theoretical height. The third acquisition unit is used to feed back the height deviation value to the corrugated cardboard compression deformation model and obtain the overall deformation compensation amount using the gradient descent method.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 3.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

Images