Body robot material picking system based on digital twin platform and configuration method

Abstract

The invention relates to the field of production line material sorting, in particular to a digital twin platform-based material picking system with a robot, which comprises a vehicle order unit, a digital twin platform and a logistics material picking unit. The vehicle order unit is connected to the logistics material picking unit and provides order information of vehicles for the logistics material picking unit. And the logistics sorting unit generates logistics circulation information from the order information to guide logistics sorting work in a real environment. Meanwhile, the logistics material picking unit further sends the logistics configuration information and the logistics circulation information to the digital twin platform. And the data twin platform generates a simulation environment based on the logistics configuration information, and then completes simulation of a real environment under the driving of the logistics circulation information. The invention also comprises a configuration method. According to the method, the key factor searching efficiency is improved, the difficulty and the operation process during scheme adjustment are reduced, and iterative optimization of the scheme can be promoted. Meanwhile, the layout of twin pictures in the parameter optimization process can guide the real field layout.

Description

Technical Field

[0001] This invention relates to the field of production line material sorting, and in particular to an embodied robot picking system and configuration method based on a digital twin platform. Background Technology

[0002] With the development trend of intelligent manufacturing and flexible production, the material sorting process on the production line, as a core node connecting production planning and workshop execution, directly determines the response speed and operational efficiency of the production line through its level of automation and intelligence. Especially in discrete manufacturing scenarios with multiple varieties and small batches, such as automobiles and 3C electronics, the demand for personalized customization of vehicle orders has surged, the types of materials are complex, the delivery routes are varied, and the picking cycle time fluctuates greatly. Traditional picking models are no longer suitable for the flexible needs of modern production lines.

[0003] Currently, material picking on production lines mainly relies on two modes: one is manual picking, where operators pick materials one by one from the storage area or line-side warehouse according to the order list. This method is not only labor-intensive and the picking accuracy is easily affected by the operator's condition, but it is also difficult to synchronize with the dynamically changing production rhythm, which can easily lead to waiting for materials or material backlog on the production line. The other is the traditional automated picking mode, which uses fixed-track robotic arms and AGVs (automated guided vehicles) combined with preset programs to complete the picking task. The movement path and operation logic of this type of system are fixed. When orders change, production line layout is adjusted, or material storage locations are changed, technicians need to rewrite the control program and debug hardware parameters. This not only has a long configuration cycle and high transformation cost, but also lacks the ability to autonomously perceive and adaptively adjust to abnormal situations on site. For example, in scenarios such as material placement deviation or pallet deformation, picking failure or material damage is very likely to occur.

[0004] Meanwhile, with the application of digital twin technology in intelligent manufacturing, some production lines have begun to try to build virtual simulation models. However, existing digital twin applications are mostly limited to production line status monitoring and data visualization, failing to deeply integrate the robot's embodied perception and autonomous decision-making capabilities. Embodied robots have the advantages of multi-sensor fusion perception and autonomous motion planning, but their application in complex material picking scenarios faces pain points such as low accuracy of mapping between virtual models and physical entities, lag in interaction between environmental perception data and the twin platform, and difficulty in optimizing multi-device collaborative scheduling strategies. Specifically, the motion status and perception data of the physical robot cannot be fed back to the digital twin platform in real time, resulting in the virtual model being unable to accurately replicate the dynamic changes of the physical scene; the optimization instructions from the twin platform are also difficult to quickly send to the robot's execution end, causing a break in the "virtual-physical" closed-loop collaboration, and failing to fully realize the simulation, pre-visualization, and optimization decision-making value of digital twins.

[0005] In summary, the current field of production line material picking urgently needs a new picking system that can integrate the advantages of digital twin technology and embodied robots to achieve full-process collaboration of virtual simulation planning, physical precision execution, and real-time data closed loop. This would solve the problems of poor flexibility, weak adaptability, and low configuration efficiency of traditional picking modes, and meet the picking needs of multiple orders and fast response in discrete manufacturing scenarios. Summary of the Invention

[0006] The purpose of this invention is to provide an embodied robot picking system and configuration method based on a digital twin platform, which mainly solves the problems existing in the prior art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is to provide an embodied robot picking system based on a digital twin platform, characterized in that it includes a vehicle order unit, a digital twin platform, and a logistics picking unit;

[0008] The vehicle order unit is connected to the logistics picking unit and provides vehicle order information to the logistics picking unit. The logistics picking unit generates logistics flow information from the order information to guide the logistics sorting work in the real environment. At the same time, the logistics picking unit also sends logistics configuration information and logistics flow information to the digital twin platform. The data twin platform generates a simulation environment based on the logistics configuration information, and then completes the simulation of the real environment under the drive of the logistics flow information.

[0009] Furthermore, the digital twin platform includes a material configuration module, a robot configuration module, an AGV configuration module, a task scheduling module, a simulation module, and an artificial intelligence module;

[0010] The material configuration module is used to set the parameters of the simulation rack, the robot configuration module is used to set the parameters of the simulation robot, and the AGV configuration module is used to configure the parameters of the simulation AGV in the simulation environment. They are all configured based on the logistics configuration information. The simulation module realizes the simulation environment based on the configurations provided by the material configuration module, the robot configuration module, and the AGV configuration module, and can also realize the simulation environment based on the configurations provided by the artificial intelligence module.

[0011] The task scheduling module reads the logistics flow information from the logistics picking unit, and uses the simulation module to drive the simulation robot in the simulation environment. Based on the simulation AGV parameters, the robot operates between the simulation rack and the simulation transfer vehicle to calculate the logistics picking efficiency. The artificial intelligence module includes a large model and an MCP service. The MCP service reads the logistics picking efficiency in the simulation environment, uses the large model to generate simulation configuration information for experiments, and sends it to the simulation module to generate a new simulation environment for operation.

[0012] Furthermore, the logistics picking module includes a material order module and a light-up picking module; the material order module generates real logistics flow information from the order information; the logistics flow information is sent to the digital twin platform as parameters of the simulation environment, and also sent to the light-up picking module to complete real production;

[0013] The light-up picking module includes a light-up material rack, a picking robot, and a transfer cart. Driven by the logistics flow information, the light-up material rack that needs to be picked lights up, and at the same time, the picking robot transfers materials between the light-up material rack and the transfer cart.

[0014] The present invention also provides a configuration method for the above-mentioned embodied robot picking system based on a digital twin platform, characterized by comprising the steps of:

[0015] Step S100: Collect parameters from the light-up picking module in the logistics picking unit, input them into the digital twin platform, establish a simulation environment, and form a basic configuration;

[0016] Step S200: Using the material order module in the logistics picking unit, the logistics flow information is sent to the simulation module in the digital twin platform; the simulation module obtains the logistics picking efficiency under the current configuration based on the basic configuration.

[0017] Step S300: Based on the basic configuration, develop an original orthogonal array containing various factors and factor levels; conduct simulation experiments and use range analysis or variance analysis to determine key factors;

[0018] Step S400: Extract the key factors from the original orthogonal array, use the artificial intelligence module in the digital twin platform to regenerate an orthogonal array containing only the key factors, and adjust the factor level of the key factors;

[0019] Step S500: Perform the simulation experiment again and use range analysis or variance analysis to find the optimal parameters corresponding to the key factors.

[0020] Step S600: Obtain the material picking efficiency of the logistics based on the key factors and the optimal parameter combination under the simulation environment, and evaluate whether the key factors and the optimal parameter combination are feasible;

[0021] Step S700: Feedback the key factors and the optimal parameter combination to the light-up picking module in the logistics picking unit for on-site verification;

[0022] Step S800: Collect the logistics picking efficiency in the real environment and compare it with the logistics picking efficiency in the simulation environment. If the comparison result does not reach the threshold, fine-tune the parameter configuration based on the result of the real environment and proceed to step S100; otherwise, proceed to step S900.

[0023] Step S900: Based on the key factors and the optimal parameter combination, a standardized parameter template is formed.

[0024] Further, step S100 includes the following steps:

[0025] Step S101: Collect the station configuration information of the light-up material rack, the robot configuration information of the picking robot, and the material configuration information in the light-up picking module;

[0026] Step S102: Establish the simulation environment corresponding to the light-up picking module in the logistics picking unit in the simulation module block of the digital twin platform;

[0027] Step S103: Input the workstation configuration information, the robot configuration information, and the material configuration information into the simulation module to establish the basic configuration for the simulation environment.

[0028] Further, step S200 includes the step,

[0029] Step S201: The material order module sends order information to the simulation module;

[0030] Step S202: The simulation module simulates the operation of the light-up picking module in the logistics picking unit in the simulation environment according to the basic configuration.

[0031] Step S203: The simulation module collects the logistics picking efficiency of the light-up picking module, including the completion time and cycle time data of each station.

[0032] Further, in steps S300 and S500, the range analysis includes the following steps:

[0033] Step S301: Under the same factor, classify according to the factor level and calculate the average efficiency of the corresponding logistics picking efficiency;

[0034] Step S302: Calculate the range of each of the factors, that is, under the specified factor, calculate the difference between the maximum and minimum values ​​among all the average efficiency values;

[0035] Step S303: Sort the factors in descending order of their corresponding ranges, and select one or more of the top-ranked factors as the key factors; then find the optimal parameters corresponding to the key factors.

[0036] Further, in steps S300 and S500, the analysis of variance includes the following steps:

[0037] Step S501: Calculate the total sum of squares using all the aforementioned logistics picking efficiencies;

[0038] Step S502: For each of the factors, calculate the corresponding sum of squares of the factors and the sum of squares of the errors;

[0039] Step S503: Calculate the factor degrees of freedom and error degrees of freedom for each of the factors;

[0040] Step S504: Calculate the factor mean square and the error mean square from the sum of squares of the factors and the sum of squares of the errors;

[0041] Step S505: Calculate the significance test index F for each factor using the factor mean square and the error mean square.

[0042] Step S506: Based on the factor degrees of freedom and the error degrees of freedom, look up the F-distribution table; if the significance test index corresponding to the factor is greater than the first significance level, then mark the factor as the key factor; if the significance test index corresponding to the factor is greater than the second significance level, then mark the factor as the critical factor; then find the optimal parameter corresponding to the key factor or the critical factor.

[0043] Step S507: Calculate the variance contribution of each of the key factors and the critical factors, in conjunction with the total sum of squares.

[0044] Furthermore, in steps S300 and S500, the simulation experiment is repeated 5 to 10 times, and the average value is taken as the result.

[0045] Furthermore, in the digital twin platform, the simulation experiments, which are repeated multiple times, are carried out simultaneously using multiple threads.

[0046] In view of the above technical features, this invention constructs a virtual picking scenario and a virtual avatar robot on a digital twin platform, and simultaneously deploys a lighting picking system. By adjusting the actions and operating parameters of the virtual avatar robot, the picking actions in the virtual picking scenario are optimized, thereby enabling the avatar robot to achieve the target cycle time in future real-world use. Compared with existing technologies, this invention has the following significant advantages:

[0047] 1. In this invention, the basic data comes from the real environment, and simulation can be carried out quickly in the existing twin environment and real picking system.

[0048] 2. This invention utilizes a digital twin platform to achieve real-time simulation of multiple processes, improving the efficiency of finding key factors and allowing for simultaneous real-time viewing of the simulation process.

[0049] 3. This invention introduces an artificial intelligence module, which uses a dedicated AI large model in conjunction with MCP services to reduce the difficulty and operational process when adjusting the solution.

[0050] 4. The collection and analysis of experimental data in this invention can drive the iterative optimization of the solution.

[0051] 5. In this invention, the final parameter scheme can be directly sent to the picking system and scheduling system to achieve logical updates.

[0052] 6. In this invention, the layout of the twin screen during the parameter optimization process can be provided to the real site for comparison and updating, guiding the on-site layout. Attached Figure Description

[0053] Figure 1 This is a system block diagram of a preferred embodiment of the embodied robot picking system based on a digital twin platform of the present invention;

[0054] Figure 2 This is a flowchart of a preferred embodiment of the configuration method of the embodied robot picking system based on the digital twin platform of the present invention;

[0055] Figure 3 yes Figure 2 Flowchart of the method for analysis of the mean range;

[0056] Figure 4 yes Figure 2 A flowchart illustrating the method of analysis of variance.

[0057] In the diagram: 1-Vehicle order unit, 2-Digital twin platform, 3-Logistics picking unit;

[0058] 21-Material Configuration Module, 22-Robot Configuration Module, 23-AGV Configuration Module, 24-Task Scheduling Module, 25-Simulation Module, 26-Artificial Intelligence Module; 261-Large Model, 262-MCP Service Module;

[0059] 31-Material order module; 32-Light-up picking module;

[0060] 321-Light-up material rack, 322-Material picking robot, 323-Transfer trolley. Detailed Implementation

[0061] The present invention will be further described below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0062] Please see Figure 1 This invention discloses an embodied robot picking system based on a digital twin platform. As shown in the figure, a preferred embodiment includes a vehicle order unit 1, a digital twin platform 2, and a logistics picking unit 3.

[0063] Vehicle order unit 1 provides vehicle order information and connects to logistics picking unit 3. Logistics picking module 3 includes material order module 31 and illuminated picking module 32. Material order module 31 reads the order information, determines the material information needed to produce the vehicle based on the order information, and then generates logistics flow information. This logistics flow information has two uses: in a real environment, it is sent to illuminated picking module 32, which completes the logistics sorting work to prepare for vehicle production. Illuminated picking module 32 includes illuminated racks 321, picking robots 322, and transfer carts 323. Driven by the logistics flow information, the illuminated racks 321 that require picking light up, while the picking robots 322 transfer materials between the illuminated racks 321 and the transfer carts 323. Simultaneously, material order module 31 also sends the logistics flow information to digital twin platform 2, thereby driving simulated sorting work in the simulation environment of digital twin platform 2. On the other hand, the material order module 31 also stores logistics configuration information, such as rack dimensions, placement location, robot location, number of robots, etc. This logistics configuration information is also sent from the material order module 31 to the data twin platform 2 for generating the simulation environment.

[0064] The digital twin platform 2 is the core module of this system. It not only simulates sorting operations in a real-world environment, providing fundamental parameters for system optimization, but also efficiently assesses material picking efficiency during optimization, guiding the optimization process. The digital twin platform 2 includes a material configuration module 21, a robot configuration module 22, an AGV configuration module 23, a task scheduling module 24, a simulation module 25, and an artificial intelligence module 26. The material configuration module 21 stores parameters of the simulated material rack, the robot configuration module 22 stores parameters of the simulated robot, and the AGV configuration module 23 stores parameters of the simulated AGV in the simulation environment. These modules obtain their real-world configurations from the logistics configuration information sent by the material picking unit 3 and then store them in their respective modules. On one hand, the simulation module 25 establishes a simulation environment based on the various configuration information stored in the material configuration module 21, robot configuration module 22, and AGV configuration module 23. Because these configurations originate from the real environment, the generated simulation environment and the real environment have a one-to-one twin relationship. In the twin simulation environment, simulation module 25 reads logistics flow information and, similar to the real environment, drives the simulated robot in the simulation environment to sort materials on the simulated rack. Simultaneously, simulation module 25 can also calculate the logistics picking efficiency of the sorting operation in real time. On the other hand, simulation module 25 can also generate a simulation environment based on configuration parameters from artificial intelligence module 26. This source is used to optimize system configuration. When artificial intelligence module 26 adjusts the configuration of the simulation environment according to the algorithm, simulation module can perform simulations based on these adjusted configurations to obtain new logistics picking efficiency, thereby evaluating the effectiveness of the adjusted parameters. Task scheduling module 24 reads logistics flow information from logistics picking unit 3 and sends it to simulation module 25, thereby driving the simulated robot in the simulation environment to operate between the simulated rack and the simulated transfer cart according to the simulated AGV parameters, thus calculating the logistics picking efficiency. The simulation environment generated based on the logistics configuration information sent by logistics picking unit 3 can be used to predict the real environment or compared with the real environment to evaluate the simulation effect. The simulation environment, in conjunction with artificial intelligence module 26, is used to evaluate scenarios that do not exist in the real environment and to assess the effectiveness of configuration adjustments. The artificial intelligence module 26 consists of a large model 261 and an MCP service module 262. The MCP service module 262 is used for data conversion. It reads the logistics picking efficiency output by the simulation module 25, injects it into the large model 261, and then the large model 261 generates simulation configuration information for the experiment. This information is then sent to the simulation module 25 to generate a new simulation environment to evaluate the logistics picking efficiency corresponding to the simulation configuration information.

[0065] Please see Figure 2 The present invention also discloses a configuration method for the embodied robot picking system based on the above-mentioned digital twin platform. A preferred embodiment includes the following steps:

[0066] Step S1: Use real-world scene parameters to establish a simulation environment and form a basic configuration.

[0067] Collect parameters from the light-up picking module in the logistics picking unit, input them into the digital twin platform, and form a basic configuration;

[0068] Step S11: Collect configuration information from the lighting picking module.

[0069] Collect station configuration information for the illuminated material racks in the illuminated picking module, including the rack, material frame, placement method, and location. Collect robot configuration information for the picking robot, including each robot's business process, actions within the process, and related motion parameters. Collect material configuration information, such as AGV locations, routes, rules, and material frame information.

[0070] Step S12: Establish the simulation environment.

[0071] In the digital twin platform, a simulation module is used to establish a simulation environment corresponding to the light-up picking module. Using this simulation environment, the module records the task load rate of the simulated robot, records experimental results, and generates multi-dimensional cycle time analysis charts and key factors.

[0072] Step S13: Set up basic configurations for the simulation environment.

[0073] Workstation configuration information, robot configuration information, and material configuration information are entered into the simulation module to ensure that the configuration in the simulation environment corresponds to the real environment, forming a basic configuration. For example, information such as rack information, material box information, coordinate information, size information (used to generate object placement rules, such as aisle width, material box height and position on the rack, etc.), route information, AGV information, robot motion information, configuring motion speed parameters, adding robots, associating animation information, configuring movement speed-related parameters, etc.

[0074] Step S2: Obtain the logistics picking efficiency based on the basic configuration.

[0075] The material order module in the logistics picking unit sends order information to the simulation module in the digital twin platform. Based on the basic configuration, the simulation module calculates the logistics picking efficiency under the current configuration. The basic configuration metrics include the number of robots, the number of racks handled by the robots, the action execution cycle, the number of racks and boxes, the AGV scheduling frequency, and the number of AGV-carried boxes.

[0076] Step S21: The material order module sends logistics flow information to the simulation module.

[0077] Step S22: The simulation module simulates the material picking process.

[0078] The simulation module simulates the operation of the light-up picking module in the simulation environment.

[0079] Step S23: Collect logistics picking efficiency.

[0080] The simulation module collects the material picking efficiency of the light-up picking module based on the simulation results. The material picking efficiency includes the completion time and cycle time data for each workstation.

[0081] Step S3: Using orthogonal arrays, determine the key factors using range analysis or variance analysis.

[0082] After obtaining the material picking efficiency based on the basic configuration, we identify potential optimization factors and their corresponding levels, and then create an orthogonal array containing these factors and levels. Within the orthogonal array, we adjust individual variables and observe the cycle time trend to identify key factors affecting efficiency.

[0083] In this embodiment, there are six possible factors, as shown in the table below. Each factor has three levels, corresponding to the core formula. An orthogonal array.

[0084] Factor code Factor Name Level 1 (Low Level) Level 2 (Intermediate) Level 3 (High Level) Value retrieval logic A Number of robots (on each side) 3 units 5 units 7 units Covering small to medium-sized picking scenarios and avoiding extreme values B Number of racks handled by the robot 8 sets / unit 12 sets / unit 16 sets / unit Matching robots by number to ensure a balanced workload. C Robot action execution cycle 8s / time 12s / time 16s / time Includes movement + grab and release time D Number of material racks and material boxes 20 sets of material racks (80 material boxes) 30 sets of material racks (120 material boxes) 40 sets of material racks (160 material boxes) Number of material boxes = Number of material racks × 4 E AGV scheduling frequency 1 time / 3 minutes 1 time / 2 minutes 1 time / 1min Scheduling frequency = AGV round trip count / time; higher frequency means higher efficiency. F Number of AGV trailer boxes 1 2 3 Load capacity, avoid overload

[0085] Based on the factor level combinations listed in the orthogonal array, multiple simulation experiments (e.g., 5 to 10) are conducted, and the average value of the final logistics picking efficiency is taken as the result to reduce random errors. The simulation experiments are driven by logistics flow information. In the digital twin platform, multiple simulation experiments are carried out simultaneously using multi-threading. After obtaining the results for each item in the orthogonal array, range analysis or variance analysis methods are used to determine which factors are key factors.

[0086] Step S4: Adjust the factor level corresponding to the key factor.

[0087] After identifying the key factors, the large model and service modules within the artificial intelligence module are used to regenerate orthogonal arrays specifically for these key factors, and the number of factor levels is adjusted to further search for the optimal parameters corresponding to these key factors. Generally, the number of factor levels for each key factor is set to 3 to 4. The adjustment of the orthogonal arrays is completed using the artificial intelligence module within the digital twin platform.

[0088] Step S5: Using orthogonal arrays, determine the optimal parameters using range analysis or variance analysis.

[0089] Based on the combination of factor levels listed in the orthogonal array, multiple simulation experiments (e.g., 5 to 10) are conducted, and the average value of the final logistics picking efficiency is taken as the result to reduce random errors. The simulation experiments are driven by logistics flow information. In the digital twin platform, multiple simulation experiments are carried out simultaneously using multi-threading.

[0090] After obtaining the results for each item in the orthogonal array, the optimal combination of parameters is determined using range analysis or variance analysis.

[0091] Step S6: Simulate and evaluate key factors and optimal parameter combinations.

[0092] In the simulation environment, the key factors are set to the optimal parameters, and then the simulation experiment is driven again by the logistics flow information to evaluate whether the logistics picking efficiency under this configuration is indeed the best. At the same time, it is evaluated whether it is feasible in the real environment, such as whether the number of robots and the size and location of the shelves are suitable for the production line site.

[0093] Step S7: Verify key factors and optimal parameter combinations in a real-world environment.

[0094] The key factors and optimal parameter combinations are fed back to the light-up picking module in the logistics picking unit, that is, the corresponding factors of the material racks and picking robots in the light-up picking module are set to optimal parameters. Then, the logistics flow information is used again to conduct on-site verification in a real environment.

[0095] Step S8: Compare and verify the results.

[0096] The material picking efficiency in the real environment is collected and compared with that in the simulation environment. If the comparison result does not reach the threshold, the parameter configuration is fine-tuned based on the result of the real environment, and the process proceeds to step S1 to find the optimal parameters again. Otherwise, the process proceeds to step S9.

[0097] Step S9: Form a standardized parameter template.

[0098] Key factors and optimal parameter combinations are recorded as standardized parameter templates to support rapid deployment and debugging of similar production lines in the future.

[0099] Please see Figure 3 In steps S3 and S5, based on each item of the orthogonal array, i.e., under different factors and factor levels, a simulation experiment is conducted using a digital twin platform to obtain the material picking efficiency corresponding to each item in the orthogonal array. Then, range analysis is used to determine the key factors or optimal parameters. Specifically, it includes the following steps:

[0100] Step S31: Calculate the average efficiency.

[0101] The method for calculating the average efficiency is as follows: under the same factor, classify according to its optional factor level, and then calculate the average value of the material picking efficiency for each factor level after classification.

[0102] Step S32: Calculate the range of each factor.

[0103] The range of a factor is calculated as follows: For a given factor, there is an efficiency mean at different factor levels. The difference between the maximum and minimum efficiency means among these multiple efficiency means is the range for that factor. This process is repeated for all factors to obtain the range for each factor.

[0104] Step S33: Filter key factors and optimal parameters.

[0105] Sort all factors in descending order of their range. A larger range indicates that the factor introduces greater volatility and is therefore more likely to be a key factor. Depending on the specific production line conditions, select one or more of the top-ranked factors as the key factors identified in this optimization.

[0106] For a given factor, there are multiple optional factor levels, each corresponding to a logistics picking efficiency. This method aims for the highest possible logistics picking efficiency; therefore, the highest efficiency is selected, and its corresponding parameter is retrieved from the orthogonal array to find the optimal parameter for that factor.

[0107] Please see Figure 4 In steps S3 and S5, based on each item of the orthogonal array, i.e., under different factors and factor levels, a simulation experiment is conducted using a digital twin platform to obtain the material picking efficiency corresponding to each item in the orthogonal array. Then, analysis of variance is used to determine the key factors or optimal parameters. Analysis of variance is more accurate than analysis of range, but the steps are slightly more cumbersome. Specifically, it includes the following steps:

[0108] Step S51: Calculate the total sum of squares of logistics picking efficiency.

[0109] Calculate the total sum of squares using the material picking efficiency corresponding to each entry in the orthogonal array. The total sum of squares represents the total variation of all experimental results, as follows:

[0110]

[0111] in, Let i be the average time interval of the i-th group of experiments. Let n be the total average beat rate, and n be the total number of trials. In this example, n = 27.

[0112] Step S52: Calculate the sum of squares of factors and the sum of squares of errors.

[0113] Similarly, for each factor, the sum of squares of the factors and the sum of squares of the errors are calculated separately.

[0114] First, divide the data by factors, then group it by factor level, and calculate the sum of squares for each factor using logistics picking efficiency. Then, divide the data by factors again and calculate the sum of squares of error for that factor using logistics picking efficiency.

[0115] Step S53: Calculate the factor degrees of freedom and error degrees of freedom.

[0116] Each factor and error contains degrees of freedom, representing the number of independent variations.

[0117] Total degrees of freedom: Where n is the total number of experiments, in this example n = 27.

[0118] Degrees of freedom for a single factor: Where m is the number of levels for each factor. In this example, factors A through F each have 3 levels, so m = 3.

[0119] Error degrees of freedom: Where q is the number of error columns, in this example q = 1.

[0120] Step S54: Calculate the factor mean square and the error mean square.

[0121] By combining the factors' degrees of freedom and the error's degrees of freedom for each factor, the factor mean square and the error mean square for each factor are obtained from the factor sum of squares and the error sum of squares.

[0122] Mean square = sum of squares / degrees of freedom, eliminating the influence of degrees of freedom on the sum of squares:

[0123] Factor mean square: One for each factor, that is, one for each of factors A to F.

[0124] Mean square error: .

[0125] Step S55: Calculate the significance test index of the factor.

[0126] The significance index F for each factor is calculated using the factor mean square and the error mean square. The significance index F = factor mean square / error mean square, used to determine whether the influence of a factor is significant.

[0127] .

[0128] Step S56: Determine the key factors and optimal parameters.

[0129] Based on the factors' degrees of freedom and the error's degrees of freedom, consult the F-distribution table. The larger the significance index for a factor, the more significant it is, and the more likely it is to become a key factor. Pre-set two thresholds: a first significance level and a second significance level.

[0130] If the significance level of a factor is greater than the first significance level, then the factor is marked as a critical factor. If the significance level of a factor is greater than the second significance level, then the factor is marked as a highly critical factor.

[0131] Consult the F-distribution table, based on the molecular degrees of freedom df j = 2, and the denominator's degrees of freedom df e = 2 as input, to obtain the critical value:

[0132] • This corresponds to a 10% significance level.

[0133] • This corresponds to a 5% significance level.

[0134] • This corresponds to a 1% significance level.

[0135] In this embodiment, the first significance level is set at 5%, and the second significance level is set at 1%. That is, when the significance test index F of a factor is greater than 19, it is classified as a critical factor, and when the significance test index F of a factor is greater than 99, it is classified as a critical factor.

[0136] After identifying key or critical factors, find the one with the highest logistics picking efficiency among the multiple optional factor levels. Then, look up the corresponding parameter in the orthogonal array to find the optimal parameter for that factor.

[0137] Step S57: Calculate the variance contribution.

[0138] Based on each key factor and critical factor, and combined with the total sum of squares, calculate its variance contribution.

[0139] The formula for variance contribution is: .

[0140] in, This indicates the proportion of factor j's contribution to the total variation; the larger the value, the more critical the impact.

[0141] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural 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 patent protection scope of the present invention.

Claims

1. A robotic picking system based on a digital twin platform, characterized in that, It includes vehicle order units, a digital twin platform, and logistics picking units; The vehicle order unit is connected to the logistics picking unit and provides vehicle order information to the logistics picking unit; The logistics picking unit generates logistics flow information from the order information to guide the logistics sorting work in the real environment; at the same time, the logistics picking unit also sends the logistics configuration information and logistics flow information to the digital twin platform. The data twin platform generates a simulation environment based on the logistics configuration information, and then, driven by the logistics flow information, completes the simulation of the real environment.

2. The embodied robot picking system based on a digital twin platform according to claim 1, characterized in that, The digital twin platform includes a material configuration module, a robot configuration module, an AGV configuration module, a task scheduling module, a simulation module, and an artificial intelligence module. The material configuration module is used to set the parameters of the simulation rack, the robot configuration module is used to set the parameters of the simulation robot, and the AGV configuration module is used to configure the parameters of the simulation AGV in the simulation environment. They are all configured based on the logistics configuration information. The simulation module realizes the simulation environment based on the configurations provided by the material configuration module, the robot configuration module, and the AGV configuration module, and can also realize the simulation environment based on the configurations provided by the artificial intelligence module. The task scheduling module reads the logistics flow information from the logistics picking unit, and uses the simulation module to drive the simulation robot in the simulation environment. Based on the simulation AGV parameters, the robot operates between the simulation rack and the simulation transfer vehicle to calculate the logistics picking efficiency. The artificial intelligence module includes a large model and an MCP service. The MCP service reads the logistics picking efficiency in the simulation environment, uses the large model to generate simulation configuration information for experiments, and sends it to the simulation module to generate a new simulation environment for operation.

3. The embodied robot picking system based on a digital twin platform according to claim 1, characterized in that, The logistics picking module includes a material order module and a light-up picking module; the material order module generates real logistics flow information from the order information; the logistics flow information is sent to the digital twin platform as parameters of the simulation environment, and also sent to the light-up picking module to complete real production; The light-up picking module includes a light-up material rack, a picking robot, and a transfer cart; driven by the logistics flow information, the light-up material rack that needs to be picked lights up, and at the same time, the picking robot transfers materials between the light-up material rack and the transfer cart.

4. A configuration method for an embodied robot picking system based on a digital twin platform as described in claim 1, characterized in that, Includes steps, Step S100: Collect parameters from the light-up picking module in the logistics picking unit, input them into the digital twin platform, establish a simulation environment, and form a basic configuration; Step S200: Using the material order module in the logistics picking unit, the logistics flow information is sent to the simulation module in the digital twin platform; the simulation module obtains the logistics picking efficiency under the current configuration based on the basic configuration. Step S300: Based on the basic configuration, develop an original orthogonal array containing various factors and factor levels; conduct simulation experiments and use range analysis or variance analysis to determine key factors; Step S400: Extract the key factors from the original orthogonal array, use the artificial intelligence module in the digital twin platform to regenerate an orthogonal array containing only the key factors, and adjust the factor level of the key factors; Step S500: Perform the simulation experiment again and use range analysis or variance analysis to find the optimal parameters corresponding to the key factors. Step S600: Obtain the material picking efficiency of the logistics based on the key factors and the optimal parameter combination under the simulation environment, and evaluate whether the key factors and the optimal parameter combination are feasible; Step S700: Feedback the key factors and the optimal parameter combination to the light-up picking module in the logistics picking unit for on-site verification; Step S800: Collect the logistics picking efficiency in the real environment and compare it with the logistics picking efficiency in the simulation environment. When the comparison result does not reach the threshold, fine-tune the parameter configuration based on the result of the real environment and proceed to step S100. Otherwise proceed to step S900; Step S900: Based on the key factors and the optimal parameter combination, a standardized parameter template is formed.

5. The configuration method of the embodied robot picking system based on a digital twin platform according to claim 4, characterized in that, Step S100 includes the following steps: Step S101: Collect the station configuration information of the light-up material rack, the robot configuration information of the picking robot, and the material configuration information in the light-up picking module; Step S102: Establish the simulation environment corresponding to the light-up picking module in the logistics picking unit in the simulation module block of the digital twin platform; Step S103: Input the workstation configuration information, the robot configuration information, and the material configuration information into the simulation module to establish the basic configuration for the simulation environment.

6. The configuration method of the embodied robot picking system based on a digital twin platform according to claim 4, characterized in that, Step S200 includes the following steps: Step S201: The material order module sends order information to the simulation module; Step S202: The simulation module simulates the operation of the light-up picking module in the logistics picking unit in the simulation environment according to the basic configuration. Step S203: The simulation module collects the logistics picking efficiency of the light-up picking module, including the completion time and cycle time data of each station.

7. The configuration method of the embodied robot picking system based on a digital twin platform according to claim 4, characterized in that, In steps S300 and S500, the range analysis includes the following steps: Step S301: Under the same factor, classify according to the factor level and calculate the average efficiency of the corresponding logistics picking efficiency; Step S302: Calculate the range of each of the factors, that is, under the specified factor, calculate the difference between the maximum and minimum values ​​among all the average efficiency values; Step S303: Sort the factors in descending order of their corresponding ranges, and select one or more of the top-ranked factors as the key factors; then find the optimal parameters corresponding to the key factors.

8. The configuration method of the embodied robot picking system based on a digital twin platform according to claim 4, characterized in that, In steps S300 and S500, the analysis of variance includes the following steps: Step S501: Calculate the total sum of squares using all the aforementioned logistics picking efficiencies; Step S502: For each of the factors, calculate the corresponding sum of squares of the factors and the sum of squares of the errors; Step S503: Calculate the factor degrees of freedom and error degrees of freedom for each of the factors; Step S504: Calculate the factor mean square and the error mean square from the sum of squares of the factors and the sum of squares of the errors; Step S505: Calculate the significance test index F for each factor using the factor mean square and the error mean square. Step S506: Based on the factor degrees of freedom and the error degrees of freedom, look up the F-distribution table; if the significance test index corresponding to the factor is greater than the first significance level, then mark the factor as the key factor; If the significance test index corresponding to the factor is greater than the second significance level, then the factor is marked as a critical factor; then, based on the critical factor or the critical factor, the corresponding optimal parameter is found. Step S507: Calculate the variance contribution of each of the key factors and the critical factors, in conjunction with the total sum of squares.

9. The configuration method of the embodied robot picking system based on a digital twin platform according to claim 4, characterized in that, In steps S300 and S500, the simulation experiment is repeated 5 to 10 times, and the average value is taken as the result.

10. The configuration method of the embodied robot picking system based on a digital twin platform according to claim 9, characterized in that, In the digital twin platform, the simulation experiments, which are repeated multiple times, are carried out simultaneously using multiple threads.

Images