A system and method for generating industrial robot process packages based on a large language model

Abstract

The present invention relates to the field of process robotics, specifically a system and method for generating process packages for industrial robots using a large language model. The system comprises a process package template library; a data editor; a large language model interface process package testing and verification program module; a process package library; a data visualization analysis interface; and an industrial robot integration interface. The method comprises: step (1) data acquisition; step (2) data preprocessing; step (3) model training; step (4) process package generation; and step (5) process package adjustment. The present invention improves process package development efficiency and quality, facilitates process package maintenance and upgrades, and adapts to complex process flows.

Description

Technical Field

[0001] The present invention relates to the technical field of process robots, and in particular to a system and method for generating a process package for an industrial robot using a large language model. Background Art

[0002] Industrial robots are widely used on production lines across various industries, serving diverse application scenarios such as spraying, grinding, loading and unloading, palletizing, and welding. They are also widely used in various industries, including automotive and photovoltaics. Different industries and scenarios require unique robot process programs. Traditional industrial robot production lines require manual process package programming, which is not only time-consuming and labor-intensive, but also limited to specific scenarios and lacks scalability. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention proposes a system and method for generating industrial robot process packages based on a large language model.

[0004] The technical problem to be solved by the present invention is achieved by adopting the following technical solutions:

[0005] A large language model industrial robot process package generation system, including:

[0006] The process package template library is used to provide various process package templates and is stored in the cloud;

[0007] The data editor allows users to design and edit the data collected by the industrial robot in a visual way, and can perform logical definition and parameter setting of abnormal data;

[0008] Large language model interface, with a built-in large language model, allows users to interact with the large language model through the interface and apply the results generated by the large language model to the development of process packages;

[0009] The process package testing and verification program module is used to test and verify the process package in simulated environments or real scenarios to ensure its feasibility and reliability;

[0010] The process package library is used to save the developed process packages in the process package library, manage and share them, and facilitate other users to use and modify them;

[0011] The data visualization analysis interface provides data analysis and visualization functions, using a large language model to analyze and display data during process package generation and runtime, as well as post-generation statistics;

[0012] Industrial robot integration interface, used to provide an interface for integration with industrial robots.

[0013] Preferably, the process package template library is applied to the process package generation stage, and the process package template library includes a palletizing process package template and a spraying process package template.

[0014] Preferably, the data editor is applied in the data preprocessing stage.

[0015] Preferably, the large language model interface and the industrial robot integration interface are applied to the execution stage after the process package is generated.

[0016] Preferably, process package testing and verification are applied in the verification stage after the process package is generated.

[0017] Preferably, the process package library is used in the storage stage of process package generation.

[0018] Preferably, the data visualization analysis interface is used in the display stage of process package generation, operation and other information.

[0019] A method for generating a process package for an industrial robot using a large language model, using the above-mentioned system for generating a process package for an industrial robot using a large language model, includes the following steps:

[0020] Step (1) Data collection: including but not limited to sensor output signal data, CAD model files of the equipment to be processed, robot model parameters, actual operating status of the robot, work tasks, factory scenes, current and voltage, and gripper workpiece data;

[0021] Step (2) Data preprocessing: Clean, normalize, and process the data to ensure the accuracy and reliability of the trained model. Abnormal data can be manually labeled and corrected.

[0022] Step 3: Model training: Use the open source project Stanford Alpaca based on the large language model for private deployment. It can dynamically access multiple different models and use the pre-processed text data and feedback data after the process package is executed to train the large language model.

[0023] Step (4) Process package generation: According to different industrial robot task requirements, select the corresponding process package template, robot configuration, code language, and operation data and generate the corresponding process package through the trained large language model;

[0024] Step (5) Process package adjustment: Determine whether the generated process package is optimized. If optimization is required, optimize it through model training in step (3) and fine-tuning of process package parameters.

[0025] Preferably, the model training in step (3) adopts a method combining FineTuning and LoRA.

[0026] The beneficial effects of the present invention are:

[0027] Compared with the prior art, the present invention has the following advantages:

[0028] 1. Improve the development efficiency of process packages: Traditional process package development requires a lot of manpower and time. However, the use of large language models can quickly generate process packages that meet the requirements, improving development efficiency.

[0029] 2. Improve the quality of process packages: The large language model has powerful language understanding and generation capabilities. It can more accurately understand the requirements of process packages and generate process packages that meet the requirements, thereby improving the quality of process packages.

[0030] 3. Facilitate maintenance and upgrades of process packages: Combining process packages with large language models makes process package development more standardized and modular, facilitating subsequent maintenance and upgrades.

[0031] 4. Adapt to complex process flows: In complex process flows, the development of process packages is difficult. However, the use of large language models can better cope with complex process flows, generate process packages that meet the requirements, and improve the stability and efficiency of the process flows. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0033] Figure 1 It is a structural block diagram of the system of the present invention;

[0034] Figure 2 Flow chart of the method of the present invention;

[0035] Figure 3 This is a flow chart of the model training of the present invention;

[0036] Figure 4 Generate a flow chart for the process package of the present invention. DETAILED DESCRIPTION

[0037] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below with reference to the accompanying drawings and embodiments.

[0038] like Figure 1 As shown, a large language model industrial robot process package generation system includes a process package template library, a data editor, a large language model interface, a process package testing and verification program module, a process package library, a data visualization analysis interface, and an industrial robot integration interface.

[0039] Specifically, the process package template library is used during the process package generation phase. It provides various process package templates, such as palletizing process package templates and spraying process package templates. Users can select and modify templates based on their needs. The process package template library is stored in the cloud and is accessed when generating process packages from a large model.

[0040] The data editor is used in the data preprocessing stage and is used to allow users to design and edit the data collected by the industrial robot in a visual way, and can perform logical definition and parameter setting of abnormal data.

[0041] The Large Language Model Interface is used during the execution phase after a process package is generated. This interface, which has a built-in large language model, allows users to interact with the large language model through the interface and apply the results generated by the large language model to process package development. Both the process package and the execution program can use the large language model interface to fine-tune the process package in real time.

[0042] The process package testing and verification is applied in the verification stage after the process package is generated. The process package testing and verification program module is used to test and verify the process package in a simulated environment or real scene to ensure its feasibility and reliability.

[0043] The process package library is used in the storage stage of process package generation. The process package library is used to store developed process packages in the process package library, manage and share them, and facilitate other users to use and modify them;

[0044] The data visualization analysis interface is used to display information during process package generation and operation. It provides data analysis and visualization capabilities, using a large language model to analyze and display data during process package generation and operation, as well as post-generation statistics, to help users better understand and optimize process packages.

[0045] The industrial robot integration interface is applied to the execution stage after the process package is generated. The industrial robot integration interface is used to provide an interface for integration with the industrial robot, so that users can directly apply the developed process package to actual production.

[0046] like Figure 2 As shown, a method for generating a process package for an industrial robot using a large language model, using the above-mentioned system for generating a process package for an industrial robot using a large language model, includes the following steps:

[0047] Step (1) Data collection.

[0048] The large language model can process raw data such as sensor data, CAD files, and manufacturing process records, and can also process text data such as robot operating instructions. Existing robot operation data is annotated and key information is extracted from the robot operation data, enabling the large language model to better understand the robot operation data. The data collected in this stage includes but is not limited to sensor output signal data, CAD model files of the equipment to be processed, robot model parameters, the robot's actual operating status, work tasks, factory scenes, current and voltage, and gripper workpiece data. This collection stage uses the industrial robot's built-in programs and device interfaces to collect internal and external data, without the need for additional specialized data acquisition equipment.

[0049] Step (2) Data preprocessing.

[0050] The purpose of data preprocessing is to clean, normalize, and process data to ensure the accuracy and reliability of the trained model. During data processing, abnormal data can be manually labeled and corrected. This step utilizes the preprocessing module, which provides basic data cleaning functionality, including prompts for incomplete data completion, standardization of data formats to JSON, automatic deletion of duplicate data, data relevance checking, and error data alerts. The preprocessing module utilizes a rule table and rule engine to implement configurable data relevance and error logic.

[0051] Step (3) Model training.

[0052] This platform uses open-source projects based on large language models, such as Stanford Alpaca, for private deployment. It uses preprocessed text data and feedback from process package execution to train large language models. It can integrate multiple large language models and provides a unified model calling interface. By collecting feedback, it continuously fine-tunes the large language model to achieve a more accurate and efficient model.

[0053] The model training adopts a method that combines FineTuning and LoRA. The present invention can try to use the FineTune method to integrate the data set into the large language model, or use the LoRA method to perform secondary training on the model and add LoRA weights to the model.

[0054] like Figure 3As shown in the figure, data collection is performed on the robot and equipment through the robot's built-in interface program. Data preprocessing processes and analyzes relevant data, helping to collect and organize the data required for the process package, reduce text noise, and extract key information. The process package feedback module generates training data based on the process package execution results, which can be fed back to the large language model.

[0055] Step (4) Process package generation.

[0056] In industrial robot applications, different types of process packages are developed using different code languages ​​and key parameters to meet specific task requirements and robot models. Therefore, when generating process packages using a large language model, some basic configuration information is required. This configuration information includes the process package type, robot model, code language, and key parameters. First, select the appropriate process package type based on the specific industrial robot task requirements and configure it. Configure the process package based on the specific robot model to ensure compatibility with the robot. Select the appropriate code language to ensure the generated process package can be correctly executed by the robot.

[0057] like Figure 4 As shown in the figure, the process package template library provides process package templates for different scenarios, such as spraying, grinding, welding, and other specific templates. The large language model will generate specific process packages based on the templates. Robot configuration refers to the robot's configuration information, including software and hardware configuration information; code language is the programming language used to control the robot's operation; and operation data refers to the robot's status and data characteristics during operation.

[0058] Step (5) Process package adjustment.

[0059] Determine whether the generated process package is optimized. If optimization is required, optimize it through model training in step (3) and fine-tuning of process package parameters. This will improve the efficiency and accuracy of the robot's task execution, thereby improving production efficiency and quality.

[0060] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and description merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A system for generating process packages for industrial robots using a large language model, characterized by: include: The process package template library is used to provide various process package templates and is stored in the cloud; The data editor allows users to design and edit the data collected by the industrial robot in a visual way, and can perform logical definition and parameter setting of abnormal data; Large language model interface, with a built-in large language model, allows users to interact with the large language model through the interface and apply the results generated by the large language model to the development of process packages; The process package testing and verification program module is used to test and verify the process package in simulated environments or real scenarios to ensure its feasibility and reliability; The process package library is used to save the developed process packages in the process package library, manage and share them, and facilitate other users to use and modify them; The data visualization analysis interface provides data analysis and visualization functions, using a large language model to analyze and display data during process package generation and runtime, as well as post-generation statistics; Industrial robot integration interface, used to provide an interface for integration with industrial robots; Applying the results generated by the large language model to the development of the process package includes the following steps: (a) Select the process package type based on the industrial robot task requirements; (b) Configure the robot model and code language to ensure compatibility; (c) Call the scene template in the process package template library and generate the specified process package through the large language model.

2. The industrial robot process package generation system based on a large language model according to claim 1, characterized in that: The process package template library is used in the process package generation stage. The process package template library includes palletizing process package templates and spraying process package templates.

3. The industrial robot process package generation system based on a large language model according to claim 1 is characterized in that: The data editor is used in the data preprocessing stage.

4. The industrial robot process package generation system for a large language model according to claim 1 is characterized in that: The large language model interface and the industrial robot integration interface are applied to the execution stage after the process package is generated.

5. The industrial robot process package generation system based on a large language model according to claim 1 is characterized in that: Process package testing and verification are applied in the verification stage after the process package is generated.

6. The industrial robot process package generation system based on a large language model according to claim 1 is characterized in that: The technology pack library is used in the storage stage of technology pack generation.

7. The industrial robot process package generation system based on a large language model according to claim 1 is characterized in that: The data visualization analysis interface is used in the process package generation and operation information display stage.

8. A method for generating a process package for an industrial robot using a large language model, characterized by: An industrial robot process package generation system using a large language model according to any one of claims 1 to 7 comprises the following steps: Step (1) Data collection: including but not limited to sensor output signal data, CAD model files of the equipment to be processed, robot model parameters, actual robot operating status, work tasks, factory scenes, current and voltage, and gripper workpiece data; Step (2) Data preprocessing: supports labeling of abnormal data through manual intervention, and cleansing and normalizing the labeled data; Step 3: Model training: Private deployment is performed using the open-source Stanford Alpaca project based on large language models. This allows for dynamic access to multiple models and uses pre-processed text data and feedback data from process package execution to train the large language model. Step 4: Process package generation: Based on different industrial robot task requirements, select the corresponding process package template, robot configuration, code language, and operating data, and generate the corresponding process package using the trained large language model; Step (5) Process package adjustment: Determine whether the generated process package is optimized. If optimization is required, optimize it through model training in step (3) and fine-tuning of process package parameters.

9. The method for generating a process package for an industrial robot using a large language model according to claim 8, wherein: In step (3), the model training adopts a method combining FineTuning and LoRA.

Images