Parcel palletizing method and apparatus

Abstract

The application discloses a parcel stacking method and device, and relates to the technical field of computers. A specific embodiment of the method comprises the following steps: a simulation environment for simulating a stacking scene is constructed, and three-dimensional models and physical properties of a stacking container and parcels in the simulation environment are defined; state information of a placed parcel and attribute information of a parcel to be placed in the simulation environment are acquired, a target placement point of the parcel to be placed in the stacking container is generated based on the state information and the attribute information through a machine learning model; in the simulation environment, the parcel to be placed is subjected to simulated placement based on the target placement point, and the stability of a stack type after the simulated placement is evaluated through physical simulation; and a stacking control instruction containing the target placement point is output based on the evaluation result of the stability of the stack type. The embodiment can effectively ensure the stability of the stack type in the stacking scene.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a parcel palletizing method and apparatus. Background Technology

[0002] With the rapid development of logistics automation and intelligent warehousing systems, automated parcel palletizing using robots has become a key technology for improving operational efficiency and reducing labor costs. The core task of a palletizing system is to maximize the loading rate within a given container (such as a pallet or cage cart) by arranging and stacking parcels in a reasonable spatial order based on their size, weight, and other attributes.

[0003] However, most existing methods only focus on geometric constraints and mechanical accessibility, lacking an effective assessment of the stacking stability of stacked packages. In real physical environments, stacked packages may tilt or even collapse due to factors such as uneven gravity distribution, slippage between packages, and external vibrations. Existing algorithms cannot predict or prevent this, making it difficult to guarantee the stability of the stack. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a parcel palletizing method and apparatus that can effectively ensure the stability of the pallet type in a palletizing scenario.

[0005] To achieve the above objectives, according to one aspect of the present invention, a parcel palletizing method is provided, comprising:

[0006] Construct a simulation environment for simulating palletizing scenarios, in which three-dimensional models and physical properties of palletizing containers and packages are defined;

[0007] The system acquires the status information of already placed packages and the attribute information of packages to be placed in the simulation environment. Based on the status information and attribute information, it generates the target placement point of the package to be placed in the palletizing container through a machine learning model.

[0008] In the simulation environment, based on the target placement point, the package to be placed is simulated and placed, and the stability of the stack after the simulated placement is evaluated through physical simulation.

[0009] Based on the evaluation results of the stack type stability, the output includes stacking control instructions containing the target placement points.

[0010] Preferably, the stability of the stack after simulated placement is evaluated through physical simulation, including:

[0011] The positional offset and / or tilt angle of each package inside the palletizing container after placement are simulated through physical simulation calculations.

[0012] The stability of the stack is evaluated based on the calculated position offset and / or tilt angle.

[0013] Preferably, the stability assessment results of the stack type are obtained based on the calculated position offset and / or tilt angle, including:

[0014] In response to the position offset not exceeding a first preset threshold and the tilt angle not exceeding a second preset threshold, an evaluation result characterizing the stability of the stack type is obtained.

[0015] Preferably, the stability of the stack after simulated placement is evaluated through physical simulation, including:

[0016] The stability of the stacking structure after simulated placement is evaluated by simulating the physical interactions between the package to be placed and the palletizing container, as well as the physical interactions between the package to be placed and the package already placed.

[0017] Preferably, after evaluating the stability of the stack after simulated placement through physical simulation, the method further includes:

[0018] Based on the evaluation results of the stack type stability, a feedback signal is generated, which is used to train the machine learning model to adjust the generation strategy of the target placement point.

[0019] Preferably, the feedback signal is a reward value;

[0020] Based on the evaluation results of the stack stability, feedback signals are generated, including:

[0021] Get the volume of the placed packages and the volume of the palletizing container;

[0022] The reward value is calculated based on the evaluation results of the stability of the stack type, the volume of the placed packages, and the volume of the palletizing container.

[0023] Preferably, training the machine learning model includes:

[0024] Construct a loss function based on feedback signals;

[0025] The gradient descent algorithm is used to adjust the internal parameters of the machine learning model so that the loss function meets the preset conditions, and the target placement point is generated using the adjusted internal parameters.

[0026] Preferably, a simulation environment for simulating palletizing scenarios is constructed, including:

[0027] Multiple simulation sub-environments are constructed to perform parallel simulations of multiple palletizing scenarios.

[0028] According to another aspect of the present invention, a parcel palletizing device is provided, comprising:

[0029] The building unit is used to build a simulation environment for simulating a palletizing scenario. The simulation environment defines the three-dimensional models and physical properties of palletizing containers and packages.

[0030] The generation unit is used to acquire the status information of the placed packages and the attribute information of the packages to be placed in the simulation environment. Based on the status information and attribute information, it generates the target placement point of the package to be placed in the palletizing container through a machine learning model.

[0031] The evaluation unit is used to simulate the placement of packages in a simulation environment based on target placement points, and to evaluate the stability of the stack structure after the simulated placement through physical simulation; and

[0032] The output unit is used to evaluate the stability of the stack type and outputs stacking control instructions containing the target placement points.

[0033] Preferably, the evaluation unit calculates the positional offset and / or tilt angle of each package within the palletized container after placement using physical simulation; based on the calculated positional offset and / or tilt angle, it obtains the evaluation result of the stability of the pallet type.

[0034] Preferably, the evaluation unit obtains an evaluation result characterizing the stability of the stack type in response to the position offset not exceeding a first preset threshold and the tilt angle not exceeding a second preset threshold.

[0035] Preferably, the evaluation unit evaluates the stability of the stacking pattern after simulated placement by simulating the physical interaction between the package to be placed and the palletizing container, as well as the physical interaction between the package to be placed and the package already placed.

[0036] Preferably, the parcel palletizing device further includes a feedback unit that generates a feedback signal based on the evaluation results of the stability of the pallet type. This feedback signal is used to train a machine learning model to adjust the generation strategy of the target placement point.

[0037] Preferably, the feedback signal is a reward value. Feedback unit: acquires the volume of the placed packages and the volume of the palletizing container; calculates the reward value based on the evaluation results of the pallet stability, the volume of the placed packages, and the volume of the palletizing container.

[0038] Preferably, training the machine learning model includes: constructing a loss function based on feedback signals; and using a gradient descent algorithm to adjust the internal parameters of the machine learning model so that the loss function meets preset conditions, so as to generate target placement points using the adjusted internal parameters.

[0039] Preferably, the building unit constructs multiple simulation sub-environments for parallel simulation of multiple palletizing scenarios.

[0040] According to another aspect of the present invention, an electronic device for parcel palletizing is provided, comprising:

[0041] One or more processors; and

[0042] Storage device for storing one or more programs.

[0043] When one or more programs are executed by one or more processors, the one or more processors implement the methods described above in the embodiments of the present invention.

[0044] According to another aspect of the present invention, a computer program is provided that, when executed by a processor, implements the methods described above in the embodiments of the present invention.

[0045] According to another aspect of the present invention, a computer-readable medium is provided having a computer program stored thereon, which, when executed by a processor, implements the methods described above in the embodiments of the present invention.

[0046] One embodiment of the above invention has the following advantages or beneficial effects: by constructing a simulation environment and using physical simulation, the stability of the stack type is evaluated, thereby overcoming the shortcomings of the prior art that only focuses on geometric accessibility and lacks stability evaluation. It can effectively predict and prevent the risk of stack collapse caused by vibration, uneven force, etc. in real physical environment, thereby effectively ensuring the stability of the stack type in the stacking scenario.

[0047] The further effects of the aforementioned unconventional alternative methods will be explained below in conjunction with specific implementation methods. Attached Figure Description

[0048] The accompanying drawings are provided to better understand the invention and are not intended to unduly limit the scope of the invention. Wherein:

[0049] Figure 1 This is a schematic diagram of the main flow of the parcel palletizing method according to an embodiment of the present invention;

[0050] Figures 2(a) and 2(b) are schematic diagrams of the structure of a pallet as a palletizing container according to an embodiment of the present invention, wherein Figure 2(a) shows the pallet in an empty state and Figure 2(b) shows the pallet in a state with packages placed on it.

[0051] Figures 3(a) and 3(b) are schematic diagrams of the structure of a cage car as a palletizing container according to an embodiment of the present invention, wherein Figure 3(a) shows the cage car in an unloaded state and Figure 3(b) shows the cage car in a fully loaded state.

[0052] Figure 4 This is a schematic diagram of multiple simulation sub-environments according to an embodiment of the present invention;

[0053] Figure 5 This is a flowchart illustrating a specific example of a parcel palletizing method according to an embodiment of the present invention;

[0054] Figure 6 This is a schematic diagram of the main modules of a parcel palletizing device according to an embodiment of the present invention;

[0055] Figure 7 This is an exemplary system architecture diagram in which embodiments of the present invention can be applied;

[0056] Figure 8 This is a schematic diagram of the structure of a computer system suitable for implementing terminal devices or servers of the present invention. Detailed Implementation

[0057] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0058] It should be noted that the collection, gathering, updating, analysis, processing, use, transmission, and storage of user personal information involved in this disclosed technical solution all comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken to prevent unauthorized access to user personal information data and to safeguard user personal information security, network security, and national security.

[0059] To collect data on the usage of our products / services, we will aggregate, analyze, and utilize technically processed user data, and share the processed statistical information with third parties. We will use secure encryption techniques and other methods to ensure that information recipients cannot re-identify specific individuals.

[0060] Figure 1 This is a schematic diagram of the main flow of a parcel palletizing method according to an embodiment of the present invention. Figure 1 As shown, the parcel palletizing method includes steps S101 to S104.

[0061] like Figure 1 As shown, in step S101, a simulation environment for simulating a palletizing scenario is constructed. The simulation environment defines the three-dimensional models and physical properties of the palletizing containers and packages.

[0062] Simulation environments used to simulate palletizing scenarios include, for example, simulation environments based on Nvidia Isaac sim, Gazebo, Unity3D, Blender, etc., constructing, for example, a 1:1 high-fidelity digital twin simulation environment.

[0063] Building the simulation environment includes constructing palletizing containers and packages within the simulation environment. Specifically, based on the containers used in actual palletizing scenarios, three-dimensional models of palletizing containers such as pallets (as shown in Figures 2(a) and (b)) and cage carts (as shown in Figures 3(a) and (b)) are constructed. Figure 2(a) shows a pallet in an empty state, and Figure 2(b) shows a pallet with packages placed on it, the pallet having a bottom surface for stacking and supporting the packages; Figure 3(a) shows a cage cart in an unloaded state, and Figure 3(b) shows a cage cart in a fully loaded state, the cage cart having a bottom surface and peripheral walls rising from the bottom surface for stacking and holding the packages. The physical properties of the palletizing containers (including dimensions, materials, coefficient of friction, etc.) are set, and the three-dimensional models of the palletizing containers are imported into the simulation environment to ensure that they are consistent with the containers in the actual palletizing scenario. In addition, a 3D model of the package was designed and constructed to ensure that its shape and size are consistent with the actual package; the physical properties of the package (including coefficient of friction, elasticity coefficient, weight, etc.) were set; and the 3D model of the package was imported into the simulation environment to ensure that it is consistent with the package in the actual palletizing scenario. By constructing the palletizing container and the package, the simulation environment is defined with the 3D model and physical properties of the palletizing container and the package.

[0064] Furthermore, multiple simulation sub-environments can be constructed for parallel simulation of multiple palletizing scenarios. That is, actual palletizing scenarios can be built and trained in parallel, supporting the independent simulation and resetting of any number of simulation sub-environments. Each simulation sub-environment is constructed separately, including the configuration of the palletizing containers and packages, absolute coordinate origin, etc. Figure 4 Examples of multiple simulation sub-environments are shown, each including a cage cart serving as a palletizing container and packages placed within it. Each simulation sub-environment independently simulates a palletizing scenario, ensuring that different sub-environments do not interfere with each other, thus enabling independent scenario simulation and ensuring the efficiency and flexibility of the training process. Running multiple simulation sub-environments simultaneously in a single training iteration significantly improves simulation efficiency and training iteration efficiency (more than 10 times faster than training with a single environment). Furthermore, each simulation sub-environment can automatically empty its palletizing container and reset to its initial state after the current reinforcement learning is completed, thereby avoiding redundancy of training data and slowdown.

[0065] like Figure 1As shown, in step S102, the status information of the placed packages and the attribute information of the packages to be placed in the simulation environment are obtained. Based on the status information and attribute information, the target placement point of the packages to be placed in the palletizing container is generated by a machine learning model.

[0066] Machine learning models, such as those based on PCT (Packing Configuration Trees), DRL (Deep Reinforcement Learning), PPO (Proximal Policy Optimization), and DDPG (Deep Deterministic Policy Gradient), are suitable for the 3D online bin packing problem. For example, during each training iteration, the agent based on the machine learning model acquires the state information of the current simulation environment. This state information includes the state information of already placed packages (e.g., the size and position of the placed packages) and the attribute information of the packages to be placed (e.g., the size and weight of the packages to be placed). This state information of the simulation environment is input into the machine learning model as the basis for generating the target placement point. Based on the state information of the current simulation environment, the agent generates the coordinates of the placement point of the package to be placed (the target placement point). In addition, the agent can make a pre-judgment of the space in the palletizing container. For example, if the agent determines that there is insufficient space in the palletizing container to continue stacking the packages to be placed, it outputs a full load signal; if it determines that there is space, it outputs a guidance signal containing the coordinates of the generated placement point to guide the placement of packages in the simulation environment.

[0067] refer to Figure 1In step S103, in the simulation environment, based on the target placement point, simulated placement of the package to be placed is performed, and the stability of the stack after simulated placement is evaluated through physical simulation. Specifically, for example, according to the output guidance signal containing the coordinates of the generated placement point, a package placement simulation is performed in the simulation environment, and the stability of the stack is evaluated after each placement action. Physical simulation, for example, simulates the physical interaction between the package to be placed and the palletizing container and the placed packages to evaluate the stability of the stack after simulated placement. For example, using the PhysX physics engine, by calculating contact force, torque, and material deformation in real time, physical simulation is performed on real physical interactions, such as collision interactions between packages and collision interactions between packages and palletizing containers (pallets or cages), to evaluate the stability of the stack after simulated placement. In addition to simulating the physical interactions between packages and between packages and palletizing containers, environmental factors such as transportation vibration and mechanical impact can also be simulated to cause package slippage, package tipping, etc., which change the stability of the package. Any simulation situation in which the stability of the package can be evaluated through physical simulation falls within the protection scope of this invention.

[0068] In this embodiment, the PhysX physics engine is used for physical simulation. It is understood that those skilled in the art can also choose other physics simulation engines capable of rigid body dynamics simulation, collision detection, and contact force calculation, such as, but not limited to, the physics systems in simulation platforms like Bullet Physics, Open Dynamics Engine (ODE), Havok Physics, and MuJoCo, all of which can achieve the physical simulation-based evaluation of the stability of the stack after simulated placement, as described in this invention.

[0069] Furthermore, physical simulation can be used to calculate the positional offset and / or tilt angle of each package within the simulated palletizing container after placement. Based on the calculated positional offset and / or tilt angle, an evaluation result of the pallet's stability is obtained. Specifically, the positional offset and / or tilt angle of each package within the simulated palletizing container are calculated, and the stability of the pallet is determined by comparing the positional offset and / or tilt angle with preset thresholds. For example, if the positional offset does not exceed a first preset threshold (e.g., 10 cm) and the tilt angle does not exceed a second preset threshold (15 degrees), the pallet is considered stable, and an evaluation result characterizing the stability of the pallet is obtained; conversely, if the positional offset exceeds the first preset threshold or the tilt angle exceeds the second preset threshold, the pallet is considered unstable, and an evaluation result characterizing the instability of the pallet is obtained. The target placement point corresponding to the evaluation result characterizing the stability of the pallet is output as a stacking control command to guide the stacking in actual package stacking scenarios.

[0070] Furthermore, the criteria for determining stack stability are not limited to both positional offset not exceeding a first preset threshold and tilt angle not exceeding a second preset threshold, i.e., simultaneously satisfying both positional offset and tilt angle conditions. Alternatively, it can be either satisfying only one of them, i.e., positional offset not exceeding the first preset threshold or tilt angle not exceeding the second preset threshold. However, simultaneously satisfying both positional offset and tilt angle conditions can more reliably ensure stack stability.

[0071] Furthermore, it is understood that the above comparison with the preset threshold is only a preferred determination method. The stability determination conditions of the present invention are not limited to this. For example, stability can also be determined by comparing with a dynamic threshold that is adaptive to the package attributes; or it can be determined based on the convergence trend of the position offset and / or tilt angle over the simulation time. Any determination rule that can determine whether the stack type is stable based on the position offset and tilt angle falls within the protection scope of the present invention.

[0072] It is also understood that the above-described evaluation using positional offset and / or tilt angle is only a preferred embodiment. The stability evaluation using physical simulation of this invention is not limited to this. For example, other physical quantities output by the physical simulation can also be used for evaluation, including but not limited to: the relationship between tangential force and friction between packages, the overturning moment of the packages, and the stress or pressure distribution on the contact surface. Any physical quantity or index that can be obtained based on physical simulation for determining the stability of the stack falls within the protection scope of this invention.

[0073] After evaluating the stability of the stack after simulated placement through physical simulation, the process may further include: generating a feedback signal based on the evaluation results of the stack stability, which is used to train a machine learning model to adjust the generation strategy of the target placement points. Specifically, the feedback signal may include a reward value (reward function). The volume of the placed packages and the volume of the palletizing container can be obtained; the reward value is calculated based on the evaluation results of the stack stability, the volume of the placed packages, and the volume of the palletizing container. Specifically, if the stability evaluation results indicate that the stack is stable, the total volume of all placed packages is calculated, and the ratio of the total volume of all placed packages to the volume of the palletizing container is calculated as the reward value; on the other hand, if the stability evaluation results indicate that the stack is unstable or placement has failed, the reward value is set to zero. The feedback signal containing the reward value is fed back to the machine learning model to iteratively optimize the placement strategy, for example, using a gradient descent algorithm.

[0074] It is understood that the above-mentioned method of obtaining feedback signals by calculating reward values ​​based on evaluation results, package volume, and palletizing container volume is only a preferred method. The method of generating feedback signals in this invention is not limited to this. For example, a binary (success / failure) signal feedback method can be used, generating a success signal (e.g., +1) if the stack is stable after placement, and a failure signal (e.g., -1 or 0) if it collapses or cannot be placed, to feed this feedback into the model for training. Furthermore, a multi-level stability level signal feedback method can be used, classifying stability into levels such as "very stable," "basically stable," "critically stable," and "unstable" based on the severity of positional offset and tilt angle, and assigning different feedback signal values ​​(e.g., +2, +1, 0, -1) for model training. Therefore, any method that can transform the stability evaluation results obtained from physical simulation into feedback information that can be used to train a machine learning model falls within the scope of protection of this invention.

[0075] It is also understood that the above calculation of the reward value (the ratio of the total volume of all placed packages to the volume of the palletizing container, or zero) is only one preferred method. The invention is not limited to this. For example, the reward value can be calculated based on the expected usability of the remaining space within the palletizing container after the placement action. For instance, the reward can be defined as a negative value of a certain mass of the remaining space after placement; the larger the remaining space, the higher its usable potential energy, but the lower the reward. Furthermore, when the pallet structure becomes unstable, the reward value can be assigned a negative value as a clear penalty signal. Any function or rule that can combine the stability assessment results with space utilization (loading rate) and transform them into a scalar for evaluating the quality of the placement action falls within the scope of protection of this invention.

[0076] By feeding feedback signals into a machine learning model, the model can be trained. Specifically, a loss function can be constructed based on the feedback signal (e.g., reward value); the gradient descent algorithm is used to adjust the intrinsic parameters of the machine learning model so that the loss function satisfies preset conditions, thereby generating target placement points using the adjusted intrinsic parameters. The preset conditions, for example, are to minimize the function value (minimum error or optimal point). Specifically, the gradient (slope) of the loss function with respect to the intrinsic parameters of the machine learning model is calculated; the intrinsic parameters of the model are adjusted in the opposite direction of the gradient to minimize the loss function. Gradient descent, through the sparsity of reward values ​​(non-zero for successful placement, zero for failed placement), guides the agent to find the optimal solution between exploration and exploitation.

[0077] By using feedback signals from evaluation results based on space utilization / loading rate (the ratio of total package volume to container volume) and stability, a machine learning model is trained, constructing a data closed loop of "simulation-optimization-re-simulation." This allows for continuous optimization of the reinforcement learning-based stacking strategy based on data such as loading rate and stack collapse events. Furthermore, space utilization is used as a positive reward, and the reward is reset to zero if the simulation detects stack collapse (excessive offset or tilt) or full load, thus forcibly avoiding high-risk stacking. Through multi-objective optimization based on loading rate and stability, the agent is driven to balance loading rate and stability, thereby maximizing loading rate while ensuring stack stability.

[0078] In addition to the evaluation results of loading rate and stability, indicators such as position offset and / or tilt angle, contact area between packages, and number of packages calculated through physical simulation can be used as reward functions (reward values) to feed into the machine learning model for training, so as to further optimize decision-making.

[0079] It is understood that the above-described method of constructing a loss function and updating model parameters using gradient descent is a preferred approach for training using feedback signals. However, this invention is not limited to this. For example, evolutionary algorithms (such as CMA-ES) can be used to encode policy model parameters as individuals, directly using feedback signals as fitness functions, and optimizing the policy through iterative evolution of the population; Monte Carlo Tree Search (MCTS) can also be used, generating possible action sequences through forward simulation at each decision time, and using feedback signals generated by physical simulation to evaluate the value of different sequences, thereby selecting the optimal action online. Any algorithm that can receive stability and load rate feedback and adjust or generate better palletizing decisions accordingly falls within the scope of protection of this invention.

[0080] refer to Figure 1 In step S104, based on the evaluation results of the stack stability, a stacking control command containing the target placement point is output. Specifically, the target placement point corresponding to the evaluation results characterizing the stack stability is used as a control signal and output to a stacking device, such as a robotic arm, to control the stacking device to place the packages to be placed in the actual stacking scenario to the target placement point, thereby ensuring the stability of the stack after the packages are stacked.

[0081] Figure 5 This is a flowchart illustrating a specific example of a parcel palletizing method according to an embodiment of the present invention. For example... Figure 5As shown, the digital twin simulation environment is first built, including: constructing palletizing containers (e.g., pallets, cage carts), constructing packages (with attributes such as friction coefficient, elasticity coefficient, and weight), and building parallel training scenarios, supporting parallel independent simulation and resetting of any number of sub-environments. Then, reinforcement learning agent training is performed, including acquiring state information such as the size and position of already placed packages, and the size and weight of packages to be placed. Based on the state information, a machine learning model generates the coordinates of the placement points for the packages to be placed. Based on the state information of the stack (e.g., including placement point coordinates), the Isaac Sim physics engine is used to simulate the state of the packages after placement, actually calculating the positional offset and tilt angle of each package after placement, thereby evaluating the stability of the stack. The stack evaluation results are fed back to the reinforcement learning training process (machine learning model) for training. Reward feedback includes a reward function that is the ratio of package volume to container volume if the stack is determined to be stable after placement; otherwise, the reward function is 0 if placement is impossible or the stack collapses after placement. A loss function is constructed using the above reward function, and the internal parameters of the machine learning model are adjusted to minimize the loss function, thereby optimizing the generation strategy of the target placement points.

[0082] The parcel palletizing method according to embodiments of the present invention assesses the stability of the pallet type by constructing a simulation environment and utilizing physical simulation. This overcomes the shortcomings of existing technologies that only focus on geometric accessibility and lack stability assessment. It can effectively predict and prevent the risk of pallet collapse caused by vibration, uneven stress, etc., in real physical environments, thereby effectively ensuring the stability of the pallet type in palletizing scenarios. Furthermore, by utilizing a feedback mechanism to construct a data closed loop of "simulation-optimization-re-simulation," the palletizing strategy based on machine learning models can be continuously optimized.

[0083] Figure 6 This is a schematic diagram of the main modules of a parcel palletizing device according to an embodiment of the present invention. Figure 6 As shown, the parcel palletizing device 600 includes: a building unit 601, a generating unit 602, an evaluation unit 603, and an output unit 604.

[0084] The construction unit 601 is used to construct a simulation environment for simulating palletizing scenarios. The simulation environment defines three-dimensional models and physical properties of palletizing containers and packages. Preferably, the construction unit 601 can construct multiple simulation sub-environments for parallel simulation of multiple palletizing scenarios.

[0085] The generation unit 602 is used to obtain the status information of the placed packages and the attribute information of the packages to be placed in the simulation environment. Based on the status information and attribute information, it generates the target placement point of the package to be placed in the palletizing container through a machine learning model.

[0086] The evaluation unit 603 is used to perform simulated placement of packages to be placed in a simulation environment based on the target placement point, and to evaluate the stability of the stack after simulated placement through physical simulation.

[0087] Specifically, the evaluation unit 603 can use physical simulation to calculate the positional offset and / or tilt angle of each package inside the palletized container after placement; based on the calculated positional offset and / or tilt angle, the evaluation result of the stability of the pallet type is obtained.

[0088] In addition, the evaluation unit 603 can obtain an evaluation result characterizing the stability of the stack type in response to the position offset not exceeding a first preset threshold and the tilt angle not exceeding a second preset threshold.

[0089] The evaluation unit 603 can evaluate the stability of the stacking structure after simulated placement by simulating the physical interaction between the package to be placed and the palletizing container, as well as the physical interaction between the package to be placed and the package already placed.

[0090] The output unit 604 is used to evaluate the stability of the stack type and outputs stacking control instructions containing the target placement point.

[0091] In addition, the parcel palletizing device 600 may also include a feedback unit that generates a feedback signal based on the evaluation results of the stability of the pallet type. This feedback signal is used to train a machine learning model to adjust the generation strategy of the target placement point.

[0092] Feedback signals can be, for example, reward values. The feedback unit can obtain the volume of the placed packages and the volume of the palletizing container; based on the evaluation results of the pallet stability, the volume of the placed packages, and the volume of the palletizing container, it calculates the reward value.

[0093] In addition, training a machine learning model may include: constructing a loss function based on feedback signals; and using a gradient descent algorithm to adjust the internal parameters of the machine learning model so that the loss function meets preset conditions, so as to generate target placement points using the adjusted internal parameters.

[0094] The parcel palletizing device according to embodiments of the present invention assesses the stability of the pallet arrangement by constructing a simulation environment and utilizing physical simulation. This overcomes the shortcomings of existing technologies that only focus on geometric accessibility and lack stability assessment. It can effectively predict and prevent the risk of pallet collapse caused by vibration, uneven stress, etc., in real physical environments, thereby effectively ensuring the stability of the pallet arrangement in palletizing scenarios. Furthermore, by utilizing a feedback mechanism to construct a data closed loop of "simulation-optimization-re-simulation," the palletizing strategy based on machine learning models can be continuously optimized.

[0095] Figure 7An exemplary system architecture 700 is shown that can be applied to the parcel palletizing method or parcel palletizing apparatus of the present invention.

[0096] like Figure 7 As shown, system architecture 700 may include terminal devices 701, 702, and 703, a network 704, and a server 705. Network 704 serves as the medium for providing communication links between terminal devices 701, 702, and 703 and server 705. Network 704 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0097] Users can use terminal devices 701, 702, and 703 to interact with server 705 via network 704 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 701, 702, and 703, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0098] Terminal devices 701, 702, and 703 can be various electronic devices with displays and web browsing capabilities, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0099] Server 705 can be a server providing various services, such as a backend management server supporting shopping websites browsed by users using terminal devices 701, 702, and 703 (for example only). The backend management server can analyze and process data such as received product information query requests, and feed back the processing results (such as target push information, product information - for example only) to the terminal devices.

[0100] It should be noted that the parcel palletizing method provided in this embodiment of the invention is generally executed by server 705, and correspondingly, the parcel palletizing device is generally installed in server 705.

[0101] It should be understood that Figure 7 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0102] The following is for reference. Figure 8 It shows a schematic diagram of the structure of a computer system 800 suitable for implementing a terminal device of the present invention. Figure 8 The terminal device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0103] like Figure 8As shown, the computer system 800 includes a central processing unit (CPU) 801, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 802 or programs loaded from storage section 808 into random access memory (RAM) 803. The RAM 803 also stores various programs and data required for the operation of the system 800. The CPU 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0104] The following components are connected to I / O interface 805: an input section 806 including a keyboard, mouse, etc.; an output section 807 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to I / O interface 805 as needed. A removable medium 811, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 810 as needed so that computer programs read from it can be installed into storage section 808 as needed.

[0105] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 809, and / or installed from removable medium 811. When the computer program is executed by central processing unit (CPU) 801, it performs the functions defined above in the system of this invention.

[0106] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0107] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0108] The units described in the embodiments of the present invention can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor can be described as including a building unit, a generating unit, an evaluation unit, and an output unit. The names of these units do not necessarily limit the unit itself; for example, the output unit can also be described as "a unit that outputs stacking control instructions containing target placement points based on the evaluation results of the stack stability."

[0109] In another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiments; or it may exist independently and not assembled into the device. The computer-readable medium carries one or more programs, which, when executed by the device, cause the device to include: constructing a simulation environment for simulating a palletizing scenario, the simulation environment defining three-dimensional models and physical properties of palletizing containers and packages; acquiring state information of placed packages and attribute information of packages to be placed in the simulation environment; generating target placement points for the packages to be placed in the palletizing container based on the state information and attribute information using a machine learning model; performing simulated placement of the packages to be placed in the simulation environment based on the target placement points, and evaluating the stability of the pallet type after simulated placement through physical simulation; and outputting a palletizing control command containing the target placement points based on the evaluation result of the pallet type stability.

[0110] According to the technical solution of this invention, by constructing a simulation environment and utilizing physical simulation, the stability of the stack type is evaluated. This overcomes the shortcomings of existing technologies that only focus on geometric accessibility and lack stability assessment. It can effectively predict and prevent the risk of stack collapse caused by vibration, uneven stress, etc., in a real physical environment, thereby effectively ensuring the stability of the stack type in the palletizing scenario. In addition, by utilizing a feedback mechanism to construct a data closed loop of "simulation-optimization-re-simulation", the stacking strategy based on the machine learning model can be continuously optimized.

[0111] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A parcel palletizing method, characterized in that, include: A simulation environment is constructed to simulate a palletizing scenario, wherein the simulation environment defines three-dimensional models and physical properties of palletizing containers and packages; The status information of the placed packages and the attribute information of the packages to be placed in the simulation environment are obtained. Based on the status information and the attribute information, the target placement point of the package to be placed in the palletizing container is generated by a machine learning model. In the simulation environment, based on the target placement point, the package to be placed is simulated and placed, and the stability of the stack after the simulated placement is evaluated through physical simulation. Based on the evaluation results of the stability of the stack type, a stacking control command containing the target placement point is output.

2. The parcel palletizing method according to claim 1, characterized in that, The method of evaluating the stability of the stack after simulated placement through physical simulation includes: The positional offset and / or tilt angle of each package within the palletizing container after placement are simulated through physical simulation calculations. Based on the calculated position offset and / or tilt angle, the stability assessment result of the stack type is obtained.

3. The parcel palletizing method according to claim 2, characterized in that, The evaluation result of the stability of the stack type is obtained based on the calculated position offset and / or tilt angle, including: In response to the position offset not exceeding a first preset threshold and the tilt angle not exceeding a second preset threshold, an evaluation result characterizing the stability of the stack type is obtained.

4. The parcel palletizing method according to claim 1, characterized in that, The method of evaluating the stability of the stack after simulated placement through physical simulation includes: The stability of the stacking pattern after simulated placement is evaluated by simulating the physical interaction between the package to be placed and the palletizing container, as well as the physical interaction between the package to be placed and the package already placed.

5. The parcel palletizing method according to claim 1, characterized in that, After evaluating the stability of the stack after simulated placement through physical simulation, the following is also included: Based on the evaluation results of the stability of the stack type, a feedback signal is generated, which is used to train the machine learning model to adjust the generation strategy of the target placement point.

6. The parcel palletizing method according to claim 5, characterized in that, The feedback signal is a reward value; The evaluation results based on the stability of the stack type generate a feedback signal, including: Obtain the volume of the placed packages and the volume of the palletizing container; The reward value is calculated based on the evaluation results of the stability of the stack type, the volume of the placed packages, and the volume of the palletizing container.

7. The parcel palletizing method according to claim 5, characterized in that, Training the machine learning model includes: A loss function is constructed based on the feedback signal; The gradient descent algorithm is used to adjust the internal parameters of the machine learning model so that the loss function meets preset conditions, so as to generate the target placement point using the adjusted internal parameters.

8. The parcel palletizing method according to claim 1, characterized in that, The simulation environment for simulating palletizing scenarios includes: Multiple simulation sub-environments are constructed to perform parallel simulations of multiple palletizing scenarios.

9. A parcel palletizing device, characterized in that, include: The building unit is used to build a simulation environment for simulating a palletizing scenario, wherein the simulation environment defines three-dimensional models and physical properties of palletizing containers and packages; The generation unit is used to acquire the status information of the placed packages and the attribute information of the packages to be placed in the simulation environment, and to generate the target placement point of the package to be placed in the palletizing container based on the status information and the attribute information through a machine learning model. An evaluation unit is used in the simulation environment to perform simulated placement of the package to be placed based on the target placement point, and to evaluate the stability of the stack after simulated placement through physical simulation. as well as The output unit is used to output a stacking control command containing the target placement point based on the evaluation result of the stability of the stack type.

10. An electronic device for parcel palletizing, characterized in that, include: One or more processors; as well as Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-8.

11. A computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-8.

12. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-8.

Images