Warehouse input and output method based on large model, electronic device and sample library

Abstract

The invention relates to the technical field of sample libraries, in particular to a large-model-based warehouse-in and warehouse-out method, an electronic device and a sample library, and aims to solve the problems of low development efficiency, poorer flexibility and lack of dynamic adjustment capability when warehouse-in and warehouse-out of the sample library are controlled through a pre-programming task scheduling process. In order to achieve the purpose, the warehouse-in and warehouse-out method comprises the steps that a plurality of pre-established process steps and an equipment knowledge base are obtained, the process steps are independent action units of components in the warehouse-in and warehouse-out process, and the equipment knowledge base is used for providing combination rules and constraint conditions of the process steps; docking with the large model to enable the large model to learn process steps and an equipment knowledge base; according to the warehouse-in and warehouse-out task, obtaining a question word oriented to a large model; inputting the questioning words into the large model, and generating a task scheduling process used for completing warehouse-in and warehouse-out tasks; and controlling the operation of the components according to the task scheduling process to complete warehouse-in and warehouse-out. The task scheduling process is dynamically generated according to the large model, and the dynamic adjustment capability of the sample library is improved.

Description

Technical Field

[0001] This invention relates to the field of sample library technology, specifically providing a method for adding and removing samples based on a large model, an electronic device, and a sample library. Background Technology

[0002] With the rapid development of the biopharmaceutical industry, the demand for biological sample storage is increasing. Biological samples such as tissues and cells in clinical and laboratory settings usually need to be stored in sample banks to achieve long-term preservation and efficient management.

[0003] In related technologies, the inbound and outbound operations of sample libraries mainly adopt a control method based on a preset task scheduling process. Specifically, the system needs to generate a task scheduling process in advance based on the task order, and then strictly control the operation of each component according to the task scheduling process to perform sample retrieval and placement operations.

[0004] However, each specific inbound and outbound process, such as the separate storage of two sample boxes to the first and second positions, or the picking and unloading of individual tubes, requires a separate task scheduling process, resulting in a large development workload and low development efficiency. When the equipment structure changes, such as adding identification sensors or adjusting the robotic arm configuration, all related task scheduling processes need to be rewritten and debugged, leading to insufficient flexibility. Furthermore, once a task begins execution, it must be completed according to the preset task scheduling process, making it impossible to respond to unexpected needs during execution, such as urgently pausing the current task or inserting a new priority task, resulting in a lack of dynamic adjustment capability for the sample library. Controlling the inbound and outbound processes of the sample library through preset task scheduling processes suffers from low development efficiency, poor flexibility, and a lack of dynamic adjustment capability.

[0005] Accordingly, a new technical solution is needed in this field to solve the above problems. Summary of the Invention

[0006] The present invention aims to solve the above-mentioned technical problems, namely, the problems of low development efficiency, poor flexibility and lack of dynamic adjustment capability in controlling the entry and exit of the sample library through pre-programmed task scheduling process.

[0007] Firstly, this invention provides a large-model-based inbound / outbound method for storage devices, which includes multiple components. The inbound / outbound method includes: acquiring multiple pre-established process steps and a device knowledge base, where each process step is an independent action unit of the component during the inbound / outbound process, and the device knowledge base provides combination rules and constraints for the process steps; interfacing with a large model to enable the large model to learn the process steps and the device knowledge base; acquiring query terms for the large model based on the inbound / outbound task; inputting the query terms into the large model to generate a task scheduling flow for completing the inbound / outbound task, wherein the task scheduling flow consists of multiple process steps arranged in execution order; and controlling the operation of the components according to the task scheduling flow to complete the inbound / outbound process.

[0008] In some embodiments, the process steps are established by the following method: the inbound and outbound process of the storage device is divided into indivisible independent action units, each of which is executed independently by a single component; the execution components, action types, action parameters, and action times of multiple independent action units are determined as multiple process steps; wherein, the action type is used to distinguish different mechanical behaviors, and the action parameters are used to quantify the specific variables in the execution process of the mechanical behavior.

[0009] In some embodiments, the motion parameters include at least the starting position, target position, and path trajectory of the independent motion unit.

[0010] In some embodiments, the inbound and outbound database includes multiple task types, and the device knowledge base includes process templates, action restrictions, and component information for the task types. The component information includes component function descriptions and the location of the component.

[0011] In some embodiments, the process template includes a plurality of process steps arranged in execution order; and / or, the action constraint is used to limit the constraints when the process steps are executed; and / or, the task scheduling process includes the step order of the process steps, the execution component, and the preceding process step of each process step.

[0012] In some embodiments, the query terms include at least the target object, the initial and final positions of the target object, the task type, and the output requirements.

[0013] In some embodiments, the step of "inputting the question word into a large model to generate a task scheduling process for completing the inbound / outbound task" further includes: inputting the question word into multiple dialog boxes of the large model to generate multiple candidate processes; determining whether the candidate processes can complete the inbound / outbound task based on the large model; if the determination result is negative, regenerating candidate processes until a candidate process capable of completing the inbound / outbound task is obtained; determining the task scheduling process based on the candidate processes capable of completing the inbound / outbound task; and if the determination result is positive, determining the task scheduling process based on the candidate processes capable of completing the inbound / outbound task.

[0014] In some embodiments, the step of "determining a task scheduling process based on candidate processes capable of completing the inbound / outbound task" further includes: determining the total duration of each candidate process capable of completing the inbound / outbound task based on the large model; determining the candidate process with the shortest total duration as the iterative process; optimizing the duration of the iterative process based on the large model to generate an iterative process with a shorter total duration, until the total duration of the regenerated iterative process is the same as the total duration of the iterative process input into the large model; and using the iterative process finally obtained through duration optimization as the task scheduling process.

[0015] In a second aspect, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor is capable of calling and running the computer program to perform the inbound / outbound method described in any of the preceding claims.

[0016] Secondly, the present invention provides a sample library, the sample library including a cache room, a storage room, a plurality of the aforementioned components, and a controller, the controller being configured to execute the inbound / outbound method described above; wherein, the components include an inner door, an outer door, a cache room robot arm, an external track, a barcode reader camera, an internal track, and a storage room robot arm; the inner door is located at the connection between the cache room and the storage room, the cache room is connected to or isolated from the external space through the outer door, the cache room robot arm, the external track, and the barcode reader camera are located in the cache room, and the internal track and the storage room robot arm are located in the storage room.

[0017] By employing the above technical solution, the inbound / outbound method of this invention enables a large model to learn pre-established process steps and equipment knowledge bases. Based on the inbound / outbound tasks, it obtains query terms for the large model, and inputting these query terms into the large model generates a task scheduling process. This invention breaks down inbound / outbound operations into reusable process steps, and utilizes the large model to combine these process steps to generate a task scheduling process in real time, eliminating the need for pre-setting the task scheduling process and improving development efficiency. When the equipment structure changes, only the equipment knowledge base and process steps need to be updated. The large model learns the new equipment knowledge base and process steps to generate a task scheduling process, eliminating the need to debug each pre-set task scheduling process as in related technologies, thus improving the flexibility of inbound / outbound control. Furthermore, during the execution of inbound / outbound tasks, the task can be paused, and then query terms can be regenerated based on new inbound / outbound requirements, allowing the large model to regenerate the task scheduling process. This facilitates responding to unexpected needs, such as urgently pausing the current task or inserting a new priority task, improving the dynamic adjustment capability of the sample library and solving the problem of process rigidity. Attached Figure Description

[0018] The preferred embodiments of the present invention are described below with reference to the accompanying drawings, in which:

[0019] Figure 1 This is a flowchart of the main steps of the database entry and exit method based on a large model in this invention;

[0020] Figure 2 yes Figure 1 Detailed flowchart of step S104;

[0021] Figure 3 yes Figure 2 A detailed flowchart of one of the steps. Detailed Implementation

[0022] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the invention and are not intended to limit the scope of protection of the invention. Those skilled in the art can make adjustments as needed to adapt to specific applications.

[0023] It should be noted that, in the description of this invention, a "module" can include hardware, software, or a combination of both. A module may include hardware circuitry, various suitable sensors, communication ports, memory, and may also include software components, such as program code, or a combination of software and hardware. The term "A and / or B" represents all possible combinations of A and B, such as only A, only B, or A and B.

[0024] This application provides a sample library, which includes a buffer room, a storage room, and multiple components. The components include an inner door, an outer door, a buffer room robotic arm, an external track, a barcode reader camera, an internal track, and a storage room robotic arm. The inner door is located at the connection between the buffer room and the storage room. The buffer room is connected to or isolated from the external space through the outer door. The buffer room robotic arm, external track, and barcode reader camera are located in the buffer room, while the internal track and storage room robotic arm are located in the storage room. The storage room is used to store biological samples, and the buffer room acts as a buffer, preventing cold air leakage from the storage room during barcode reading and transfer operations.

[0025] This application provides a large-model-based inbound / outbound method for storage devices, which include multiple components.

[0026] Combination Figure 1 As shown, the inbound and outbound methods provided in this application include:

[0027] S101, Obtain multiple pre-established process steps and equipment knowledge bases. A process step is an independent action unit of a component during the entry and exit process. The equipment knowledge base is used to provide combination rules and constraints for process steps.

[0028] Optionally, the process steps are established as follows: the inbound and outbound process of the storage device is broken down into indivisible independent action units, each of which is executed independently by a single component. The executing component, action type, action parameters, and action time of each of the multiple independent action units are determined, forming multiple process steps. The action type is used to distinguish different mechanical behaviors, such as picking up a box, opening, closing, extending the track, retracting the track, moving the robotic arm, and reading codes. The action parameters are used to quantify the specific variables during the execution of the mechanical behavior.

[0029] This configuration breaks down the complex inbound and outbound processes into modular, independent action units, allowing the larger model to freely combine these units to generate task scheduling flows. When the storage device structure is adjusted, only the affected independent action units need to be updated, without reconstructing the entire process. Determining the execution components, action types, action parameters, and action times of multiple independent action units clarifies the specific execution method of each process step, laying the foundation for subsequent task scheduling.

[0030] Optionally, the motion parameters include at least the starting position, target position, and path trajectory of the independent motion unit. By defining the starting position, target position, and path trajectory of the independent motion unit, the component can precisely execute the independent motion unit.

[0031] The following uses a sample library as an example to illustrate the process steps:

[0032] Taking the internal doors of the sample library as an example, in the entry and exit operations of the sample library, the internal doors consist of two independent action units: opening and closing. The execution component of the opening action unit is the internal door, the action type is opening, the action parameter moves from the closed position to the open position, and the action time is 4 seconds. The execution component of the closing action unit is the internal door, the action type is closing, the action parameter moves from the open position to the closed position, and the action time is 4 seconds.

[0033] Taking the robotic arm in the sample library's storage area as an example, in the sample library's inbound and outbound operations, the storage area robotic arm includes four independent action units: retrieving boxes from the internal track, placing boxes onto the internal track, retrieving boxes from a designated location on the shelf, and placing boxes onto a designated location on the shelf. The execution component for retrieving boxes from the internal track is the storage area robotic arm, the action type is box retrieval, the action parameter is moving from the current position to the position where the box is located on the internal track, and the action time is 10 seconds.

[0034] All process steps of the sample library are shown in Table 1:

[0035] Table 1. Flowchart of the Sample Library

[0036] Optionally, the inbound and outbound processes include various task types, such as single-box inbound, single-box outbound, pipe-picking inbound, and pipe-picking outbound. The equipment knowledge base includes process templates, action constraints, and component information for each task type. The process template includes multiple process steps arranged in execution order. For example, the single-box inbound process template includes multiple process steps arranged in execution sequence, which, when executed sequentially, complete the single-box inbound function. Action constraints are used to define the constraints during process step execution. For example, the external track can only extend and retract after the outer door is opened, and the outer door must be closed immediately after the external track is retracted into the buffer compartment. Component information includes component function descriptions and component locations. By including process templates, action constraints, and component information in the equipment knowledge base, the large model is provided with the combination relationships of process steps, the constraint relationships of actions, and the location and function descriptions of components, facilitating the large model's reasoning and precise combination of process steps.

[0037] The following uses a sample database as an example to illustrate the component information in the device knowledge base:

[0038] Table 2. Component Information Table of Equipment Knowledge Base

[0039] The following uses a sample database as an example to illustrate the action restrictions of the device knowledge base:

[0040] (1) The external track can only be extended and retracted after the outer door is opened, and the external track must be closed immediately after it is retracted into the buffer room.

[0041] (2) The internal rail can only be extended and retracted after the internal door is opened. The internal rail must be closed immediately after it is retracted into the buffer room.

[0042] (3) The box must be at the photo and code reading position for the code reading camera to perform the photo and code reading action. After the box is put into storage and the tube is picked, the box must be sent to the photo and code reading position for photo and code reading operation.

[0043] Optionally, the task types of the equipment knowledge base include single box inbound, single box outbound, pipe inbound, and pipe outbound, etc.

[0044] The following uses a sample database as an example to illustrate the workflow template for a device knowledge base:

[0045] (1) Single-box receiving is the process of moving a box from outside the warehouse, through the buffer area, to the storage shelf. The template flow for single-box receiving is as follows:

[0046] The outer door opens, the outer track extends, and the operator places the box onto the outer track. The outer track retracts the box, the outer door closes, and the buffer area robotic arm moves the box from the outer track to the photo and barcode reader position. The barcode reader takes a photo and reads the barcode at the photo and barcode reader position. After the photo and barcode reading is completed, the buffer area robotic arm moves the box from the photo and barcode reader position to the inner track. The inner door opens, the inner track extends the box into the storage area, and the storage area robotic arm moves the box from the inner track to the storage area shelf. The inner track retracts back into the buffer area, the inner door closes, and the warehousing process is complete.

[0047] (2) Pipe picking and storage involves placing the pipe in the source box, moving the source box from outside the warehouse to the pipe picking position 1 in the buffer area, moving the target box from the storage shelf to the pipe picking position 2 in the buffer area, and then using the buffer area robotic arm to retrieve the pipe from the source box at pipe picking position 1 and transfer it to the target box in pipe picking position 2. The template flow for pipe picking and storage is as follows:

[0048] Part 1: The outer door opens, the outer track extends, the manual places the box onto the outer track, the outer track retracts the box, the outer door closes, the buffer area robot moves the box from the outer track to the photo and code reading position, the code reading camera takes a photo and reads the code at the photo and code reading position, after the photo and code reading is completed, the buffer area robot moves the box from the photo and code reading position to the picking tube position 1.

[0049] Part Two: The inner door opens, the inner track extends to the storage area, the storage area robot moves the target box from the storage area shelf to the inner track, the inner track retracts the box, the inner door closes, and the buffer area robot moves the box from the inner track to the picking tube position 2.

[0050] Part 3: The robotic arm in the buffer room takes a tube from the source box at position 1 and moves it to the target box at position 2.

[0051] Part 4: The robotic arm in the buffer area moves the target box from the picking position 2 to the photo and code reading area. The code reading camera reads the code and takes a picture to identify whether the tube to be picked has been picked up. After the identification is completed, the robotic arm in the buffer area moves the target box from the photo and code reading position to the internal track. The internal door opens, and the internal track carries the target box out to the storage area. The robotic arm in the storage area moves the box from the internal track to the storage area shelf. The internal track retracts to the buffer area, and the internal door closes.

[0052] Part 5: The robotic arm in the buffer area moves the source box from the picking position 1 to the photo and code reading area. The code reading camera reads the code and takes a picture to identify whether the tube to be picked has been picked. After the identification is completed, the robotic arm in the buffer area moves the box from the photo and code reading position to the outer track. The outer door opens, and the box on the inner track extends out of the warehouse. After the box is taken away by the manual, the outer track is retracted, the outer door is closed, and the picking and storage of the tube is completed.

[0053] Single-box outbound is the process of moving a box from the storage area shelf, through the buffer area, and out of the warehouse. The process is the reverse of single-box inbound, and will not be listed here.

[0054] Pipe picking out of the warehouse is the process of transferring the pipes from the source boxes in the storage area to the target boxes outside the warehouse. The process is the reverse of the pipe picking in of the warehouse, and will not be listed here.

[0055] S102, interfaces with the large model, enabling the large model to learn the process steps and the device knowledge base.

[0056] Optionally, large models include ChatGPT, DeepSeek, Wenxin Yiyan, Kimi, or Doubao, etc. The large model can learn from the process steps and device knowledge base by inputting them into its dialog box.

[0057] S103, based on the inbound and outbound tasks, obtain the query terms for the large model.

[0058] Optionally, the query terms should at least include the target object, the initial and final positions of the target object, the task type, and the output requirements. This setting allows the large model to clearly define the task requirements and guides it in generating the necessary task scheduling process.

[0059] The following describes the query terms with reference to specific examples:

[0060] Suppose the inbound / outbound task is: to place two sample boxes into the first shelf of the first row, first column, and first level, and the second shelf of the first row, first column, and second level, respectively. Then the question word is:

[0061] There is currently an automated sample storage unit that performs sample box inbound and outbound operations. The process steps and equipment knowledge base of the automated sample storage unit are as shown in the previous input dialog box. For the inbound and outbound tasks, a task scheduling process consisting of multiple process steps arranged in the execution order needs to be generated. The task is to put two sample boxes into the first shelf of the first row, first column, and first layer of the storage unit, and into the second shelf of the first row, first column, and first layer. Based on this task, generate the process steps with the minimum overall time consumption and that meet the process requirements.

[0062] S104. Input the query terms into the large model to generate a task scheduling process for completing the inbound and outbound tasks. The task scheduling process consists of multiple process steps arranged in the order of execution.

[0063] Input the question words into the large model's dialog box. The large model can learn the combination rules and constraints of the process steps based on the device knowledge base, and combine the process steps according to the question words to generate a task scheduling process for completing the inbound and outbound tasks.

[0064] In some embodiments, the task scheduling process includes the step sequence, execution components, and the preceding step of each step. The preceding step refers to the previous step that must be completed before the current step begins execution. By including the step sequence, execution components, and the preceding step of each step in the task scheduling process, the execution order of the task scheduling process can be clearly defined, ensuring that each component executes in an orderly manner, avoiding interference or errors, and facilitating the accurate implementation of inbound and outbound tasks.

[0065] In the task scheduling process, all process steps of each component constitute a task queue. After the task scheduling process is generated, the program establishes a thread corresponding to each component to execute the process steps to be run in the task queue of this component. The execution time is after the execution of the preceding process step is completed.

[0066] The task scheduling process is explained below using a sample library as an example:

[0067] Table 3 Task Scheduling Flowchart

[0068] S105 controls the operation of components according to the task scheduling process to complete the inbound and outbound processes.

[0069] By employing the above technical solution, the inbound / outbound method of this invention enables a large model to learn pre-established process steps and equipment knowledge bases. Based on the inbound / outbound tasks, it obtains query terms for the large model, and inputting these query terms into the large model generates a task scheduling process. This invention breaks down inbound / outbound operations into reusable process steps, and utilizes the large model to combine these process steps to generate a task scheduling process in real time, eliminating the need for pre-setting the task scheduling process and improving development efficiency. When the equipment structure changes, only the equipment knowledge base and process steps need to be updated. The large model learns the new equipment knowledge base and process steps to generate a task scheduling process, eliminating the need to debug each pre-set task scheduling process as in related technologies, thus improving the flexibility of inbound / outbound control. Furthermore, during the execution of inbound / outbound tasks, the task can be paused, and then query terms can be regenerated based on new inbound / outbound requirements, allowing the large model to regenerate the task scheduling process. This facilitates responding to unexpected needs, such as urgently pausing the current task or inserting a new priority task, improving the dynamic adjustment capability of the sample library and solving the problem of process rigidity.

[0070] In some embodiments, combined with Figure 2 As shown, the step of "inputting the query terms into the large model to generate a task scheduling process for completing the inbound and outbound tasks" further includes:

[0071] S1041, Input the question terms into multiple conversation boxes of the large model to generate multiple candidate flows. For example, input the question terms into two conversation boxes of the large model to generate two candidate flows.

[0072] S1042, determine whether the candidate process can complete the inbound and outbound tasks based on the large model.

[0073] Step S1042 can be implemented through the following sub-steps: inputting the candidate process into the large model; generating judgment question words for the large model, which are used to guide the large model to judge whether the candidate process can complete the inbound and outbound tasks.

[0074] Assuming candidate process A, the inbound / outbound task is: to place two sample boxes into the first shelf of the first row, first column, and first level, and into the second shelf of the first row, first column, and second level, respectively. The query is: There is an automated sample warehouse performing inbound and outbound operations for sample boxes. The automated sample warehouse's process steps and equipment knowledge base are as shown in the previously entered dialog box. Given candidate process A, determine whether candidate process A can successfully place two sample boxes into the first shelf of the first row, first column, and first level, and into the second shelf of the first row, first column, and second level, respectively.

[0075] S1043, if the judgment result is negative, regenerate candidate processes until a candidate process capable of completing the inbound / outbound task is obtained. Based on the candidate processes capable of completing the inbound / outbound task, determine the task scheduling process.

[0076] S1044, if the judgment result is yes, determine the task scheduling process based on the candidate processes that can complete the inbound and outbound tasks.

[0077] By generating multiple candidate processes and judging whether each candidate process can complete the inbound and outbound tasks based on a large model, the candidate processes can be filtered through competition among multiple candidate processes and the large model can be used to judge the candidate processes. Candidate processes that cannot complete the inbound and outbound tasks can be removed, and the optimal task scheduling process can be generated.

[0078] In some embodiments, combined with Figure 3 As shown, the step of "determining the task scheduling process based on candidate processes capable of completing inbound and outbound tasks" further includes:

[0079] S301, Based on the large model, determine the total duration of each candidate process that can complete the inbound / outbound task. In this step, the total duration of the candidate process is obtained by adding up the action times of the process steps contained in the candidate process.

[0080] S302, determine the candidate process with the shortest total duration as the iterative process.

[0081] In this step, the total duration of multiple candidate processes is compared, and the candidate process with the shortest total duration is selected as the iterative process.

[0082] S303: Optimize the iteration process based on the large model to generate an iteration process with a shorter total duration, until the total duration of the regenerated iteration process is the same as the total duration of the iteration process input to the large model.

[0083] S304 uses the iterative process obtained from time optimization as the task scheduling process.

[0084] For example, candidate processes R1 and R2 can complete the inbound / outbound task, but their total durations are different. The shorter total duration of R1 is chosen as the iterative process, and this iterative process is input into the large model's dialog box. This informs the large model that R1 has a shorter total duration, allowing it to re-infer a task scheduling process that can complete the inbound / outbound task in a shorter time. If the total duration of the regenerated iterative process is the same as the total duration of the iterative process input into the large model, then this task scheduling process is considered optimal and is adopted as the final task scheduling process.

[0085] This setting allows for the selection of the task scheduling process with the shortest execution time. It not only ensures that the generated task scheduling process can complete basic data entry and exit functions, but also locks in the optimal solution among multiple feasible solutions.

[0086] This application provides a computer-readable storage medium storing a computer program, wherein the computer program executes the above-described inbound / outbound method when it runs.

[0087] The aforementioned computer-readable storage medium may be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.

[0088] This application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor is able to call and run the computer program to perform the above-described inbound and outbound methods.

[0089] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0090] Memory, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as the program instructions / modules corresponding to the methods in the embodiments of this application. The processor executes the program instructions / modules stored in the memory to perform functional applications and data processing, that is, to implement the inbound and outbound methods in the above embodiments.

[0091] The memory may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function. The data storage area may store data created based on the use of the terminal device. Furthermore, the memory may include high-speed random access memory and may also include non-volatile memory.

[0092] The technical solutions of this application embodiment can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes one or more instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method of this application embodiment. The aforementioned storage medium can be a non-transitory storage medium, including: USB flash drive, portable hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, and other media capable of storing program code; it can also be a transient storage medium.

[0093] Although the steps in the above embodiments are described in the above order, those skilled in the art will understand that in order to achieve the effect of this embodiment, different steps do not need to be executed in such order. They can be executed simultaneously (in parallel) or in reverse order. These simple changes are all within the protection scope of this application.

[0094] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A large-model-based inbound / outbound method applied to storage devices, the storage devices comprising multiple components, characterized in that... The inbound and outbound methods include: Acquire multiple pre-established process steps and equipment knowledge bases. The process steps are independent action units of the components during the entry and exit process. The equipment knowledge base is used to provide combination rules and constraints for the process steps. Interact with the large model to enable the large model to learn the process steps and the equipment knowledge base; Based on the inbound and outbound tasks, obtain the query terms for the large model; The query terms are input into the large model to generate a task scheduling process for completing the inbound and outbound tasks. The task scheduling process consists of multiple process steps arranged in the order of execution. The components are controlled to operate according to the task scheduling process to complete the inbound and outbound operations.

2. The inbound / outbound method according to claim 1, characterized in that, The process steps are established using the following method: The inbound and outbound process of the storage device is broken down into indivisible independent action units, each of which is executed independently by a single component; The execution components, action types, action parameters, and action times of multiple independent action units are determined as multiple process steps; wherein, the action type is used to distinguish different mechanical behaviors, and the action parameters are used to quantify the specific variables in the execution process of mechanical behaviors.

3. The inbound / outbound method according to claim 2, characterized in that, The motion parameters include at least the starting position, target position, and path trajectory of the independent motion unit.

4. The inbound / outbound method according to claim 1, characterized in that, The inbound and outbound database includes various task types, and the equipment knowledge base includes process templates, action restrictions, and component information for each task type. The component information includes component function descriptions and the location of the component.

5. The warehousing and outbound method according to claim 4, characterized in that, The process template includes multiple process steps arranged in execution order; and / or, The action restriction is used to limit the constraints when the process step is executed; and / or, The task scheduling process includes the step sequence of the process steps, the execution unit, and the preceding process step of each process step.

6. The inbound / outbound method according to claim 4, characterized in that, The question words shall include at least the target object, the initial and final positions of the target object, the task type, and the output requirements.

7. The inbound / outbound method according to any one of claims 1 to 5, characterized in that, The step of "inputting the query terms into the large model to generate a task scheduling process for completing the inbound and outbound tasks" further includes: The question words are input into multiple dialogue boxes of the large model to generate multiple candidate processes; Based on the large model, determine whether the candidate process can complete the inbound / outbound task; If the result is negative, regenerate the candidate process until a candidate process capable of completing the inbound / outbound task is obtained; determine the task scheduling process based on the candidate process capable of completing the inbound / outbound task. If the judgment result is yes, the task scheduling process is determined based on the candidate processes that can complete the inbound and outbound tasks.

8. The inbound / outbound method according to claim 7, characterized in that, The step of "determining the task scheduling process based on candidate processes capable of completing the inbound and outbound tasks" further includes: Based on the large model, determine the total time for each candidate process that can complete the inbound / outbound task; The candidate process with the shortest total duration is determined as the iterative process; The iteration process is optimized based on the large model to generate a shorter iteration process, until the total duration of the regenerated iteration process is the same as the total duration of the iteration process input into the large model. The iterative process obtained after time optimization is used as the task scheduling process.

9. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is capable of calling and running the computer program to perform the inbound / outbound method as described in any one of claims 1 to 8.

10. A sample library, characterized in that, The sample library includes a cache room, a storage room, multiple components, and a controller, the controller being configured to perform the inbound / outbound method as described in any one of claims 1 to 8; The components include an inner section, an outer section, a buffer room robot arm, an external track, a barcode reader camera, an internal track, and a storage room robot arm; The inner door is located at the connection between the cache room and the storage room. The cache room is connected to or isolated from the external space through the outer door. The cache room robot arm, the external track, and the barcode reader camera are located in the cache room. The inner track and the storage room robot arm are located in the storage room.

Images