Optimal logistics packaging information creation and quotation automation system using generative ai

Abstract

An apparatus for automatically generating logistics packaging information and quotation information based on product specification information according to the present disclosure for achieving the aforementioned technical problem comprises: a memory storing at least one instruction; and a processor that performs an operation according to the instruction. The processor receives product specification, weight, and quantity information, generates packaging information including the dimensions of a packaging box, cushioning material structure, wood cutting specifications, and required amount of packaging materials using a generative artificial intelligence model, calculates a standard unit price based on the generated packaging information to generate a quotation, generates a container loading simulation image based on the packaging information, and outputs the generated packaging information, quotation, and loading image through a user interface.

Description

Technology Field

[0001] The present disclosure relates to technology for logistics packaging design and quotation automation, and more specifically, to an optimal logistics packaging information generation and quotation automation system that utilizes Generative Artificial Intelligence (Generative AI) based on information such as product specifications, weight, and quantity to automatically generate packaging box dimensions, cushioning material placement, material requirements, and wood cutting specifications without product drawings, and automatically calculates example container loading images and quotations. Background Technology

[0002] Unless otherwise indicated in this specification, the contents described in this section are not prior art for the claims of this application, and are not to be recognized as prior art simply because they are included in this section.

[0003] Packaging design and cost estimation in the logistics industry are primarily manual tasks that require significant time and manpower due to difficulties in securing product drawings, repetitive design work, and inconsistencies in performance among designers. Particularly for large export items such as industrial equipment, customized packaging designs are required that consider product size, weight, quantity, and transportation environments; however, under existing methods, product drawings or specification information necessary for design are often not directly transmitted to external packaging specialists. This is attributed to internal corporate security policies and confidentiality obligations, forcing packaging companies to perform indirect designs based on limited information, which can lead to reduced packaging quality, rework, and material waste. Furthermore, customers (shippers) frequently request visual information regarding container loading and packaging appearance in addition to packaging unit costs; however, existing design tools are developed for internal and external design purposes and have limitations in providing visual materials for customer interaction.

[0004] Furthermore, as detailed work items such as packaging material quantities, wood cutting specifications, and cushioning material placement are managed manually based on packaging design results, repetitive and inefficient work is required when calculating packaging costs and preparing quotations. This results in problems such as delays in the packaging process, work errors, and losses in logistics costs.

[0005] Recently, as the ability to generate complex information based on input text using Generative AI has become possible, automation technology is spreading across various fields, ranging from simple document creation to image generation and design assistance. In particular, Large Language Model (LLM)-based Generative AI can be utilized to automatically generate structural information, material requirements, and container loading examples necessary for packaging design by receiving unstructured or semi-structured information such as product names, specifications, weights, and quantities.

[0006] However, most logistics packaging design automation systems to date are based on manual input and drawing-based design methods, and there are currently no services that integrate design, estimation, and visual data by directly linking generative AI. Accordingly, the development of technologies for automatic packaging information generation and estimation automation based on generative AI can serve as a significant technological turning point that contributes to the digital transformation of the logistics industry, operational efficiency, and packaging quality improvement. Prior art literature

[0007] 1. Korean Patent Registration No. 10-2830771 (July 2, 2025) 2. Korean Patent Publication No. 10-2014-0116994 (October 7, 2014) The problem to be solved

[0008] The purpose of the embodiments disclosed in this disclosure is to automatically generate packaging box dimensions, cushioning material structure, material requirements, and example container loading images using only input values ​​such as product specifications, weight, and quantity, without product drawings, by utilizing generative artificial intelligence, and to automatically generate a quotation by calculating a standard unit price based on the automatically generated packaging information.

[0009] In addition, the embodiments disclosed in this disclosure aim to provide a logistics packaging information generation and quotation automation system that realizes design efficiency, business standardization, and logistics quality improvement by providing visualized packaging / loading images for customer response and marketing.

[0010] Meanwhile, the technical problems of the present invention are not limited to those mentioned above, and other unmentioned technical problems will be clearly understood by a person skilled in the art from the description below. means of solving the problem

[0011] An apparatus for automatically generating logistics packaging information and quotation information based on product specification information according to the present disclosure for achieving the aforementioned technical problem comprises: a memory storing at least one instruction; and a processor that performs an operation according to the instruction. The processor receives product specification, weight, and quantity information, generates packaging information including the dimensions of a packaging box, cushioning material structure, wood cutting specifications, and required amount of packaging materials using a generative artificial intelligence model, calculates a standard unit price based on the generated packaging information to generate a quotation, generates a container loading simulation image based on the packaging information, and outputs the generated packaging information, quotation, and loading image through a user interface.

[0012] In addition, the above-mentioned generative artificial intelligence model is based on a Large Language Model (LLM) and can generate packaging information based on natural language or text-based information inputted by the product.

[0013] In addition, the processor can generate packaging information by receiving unstructured data such as product name, dimensions, weight, and order quantity as input, even without a product drawing.

[0014] In addition, the processor can generate an estimate in a standard format including the type and quantity of packaging materials used, labor costs, and design costs based on the generated packaging information.

[0015] In addition, the processor can generate a three-dimensional image or a rendering image of the generated packaging box arranged in a container stacking form and provide it to the user.

[0016] In addition, the above user interface is configured as a web-based subscription (SaaS) structure, and can enable multiple small and medium-sized export manufacturers or 3PL companies to view and download the generated design results online.

[0017] In addition, the processor may include a self-learning loop that periodically evaluates a generative AI model based on feedback data from a user and improves packaging design accuracy and estimate prediction precision through learning. Effects of the invention

[0018] According to the aforementioned means for solving the problem of the present disclosure, packaging box dimensions, cushioning material structure, material requirements, etc. can be automatically designed using only information such as product specifications, weight, and quantity, without providing product drawings (CAD, etc.) externally, thereby enabling security-friendly and drawing-based design.

[0019] Furthermore, according to the aforementioned means for solving the problem of the present disclosure, since generative AI automatically generates packaging information and automatically generates an estimate through standard unit price calculation based thereon, the speed and accuracy of design and estimation work are improved, deviations among designers are reduced, and standardization is enabled.

[0020] In addition, according to the aforementioned means for solving the problem of the present disclosure, packaging and container loading images generated through generative AI are provided to cargo customers in real time, thereby improving reliability and satisfaction and enabling the use of customer-tailored visual materials for marketing.

[0021] Furthermore, according to the aforementioned means for solving the problem of the present disclosure, packaging design automation reduces the design time by more than half, minimizes waste of materials and labor costs due to incorrect packaging or rework, and lowers transportation costs by maximizing the utilization of container space. This enables the company to enhance its logistics competitiveness and increase sales.

[0022] In addition, according to the aforementioned means for solving the problem of the present disclosure, 3PL companies and small and medium-sized export manufacturing companies can easily access and utilize it through a web-based lightweight SaaS platform structure, thereby accelerating the transition from a traditional manual-centered logistics design environment to a digital-based AI system.

[0023] Furthermore, according to the aforementioned means for solving the problem of the present disclosure, as an example of applying generative AI technology to the B2B industrial design domain, it can contribute to the design automation of various industries and the expansion of AI-based services in the future.

[0024] The effects obtainable from the exemplary embodiments of the present disclosure are not limited to those mentioned above, and other unmentioned effects can be clearly derived and understood by those skilled in the art to which the exemplary embodiments of the present disclosure belong from the description below. That is, unintended effects resulting from the implementation of the exemplary embodiments of the present disclosure can also be derived by those skilled in the art from the exemplary embodiments of the present disclosure. Brief explanation of the drawing

[0025] Figure 1 is a diagram showing an optimal logistics packaging information generation and quotation automation system using generative AI. FIG. 2 is a diagram illustrating a logistics packaging information generation and quotation automation platform utilizing generative AI according to an embodiment. FIG. 3 is a block diagram of an optimal logistics packaging information generation and quotation automation device using generative AI according to an embodiment. FIG. 4 is a diagram illustrating the linkage structure of a logistics packaging service metadata collection and generative artificial intelligence system according to an embodiment. Figure 5 is a diagram illustrating the process of generating optimal logistics packaging information and automating quotations using generative AI according to an embodiment. Specific details for implementing the invention

[0026] Hereinafter, various embodiments of the present disclosure are described in conjunction with the accompanying drawings. As various embodiments of the present disclosure may be subject to various modifications and may have various forms, specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the various embodiments of the present disclosure to specific forms, and it should be understood that they include all modifications and / or equivalents and substitutions that fall within the spirit and scope of the various embodiments of the present disclosure. In relation to the description of the drawings, similar reference numerals have been used for similar components.

[0027] In various embodiments of the present disclosure, terms such as “comprising” or “having” are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0028] In various embodiments of the present disclosure, expressions such as “or” include any and all combinations of the words listed together. For example, “A or B” may include A, may include B, or may include both A and B.

[0029] Expressions such as "first," "second," "first," or "second" used in various embodiments of the present disclosure may modify various components of the various embodiments, but do not limit such components. For example, such expressions do not limit the order and / or importance of such components and may be used to distinguish one component from another.

[0030] When it is mentioned that a component is "connected" or "joined" to another component, it should be understood that the component may be directly connected or joined to the other component, but that a new component may also exist between the component and the other component.

[0031] In the embodiments of the present disclosure, terms such as "module," "unit," "part," etc. are used to refer to a component that performs at least one function or operation, and such component may be implemented in hardware or software, or in a combination of hardware and software. Additionally, a plurality of "modules," "units," "parts," etc. may be integrated into at least one module or chip and implemented as at least one processor, except where each needs to be implemented in specific individual hardware.

[0032] Terms such as those defined in commonly used dictionaries should be interpreted as having meanings consistent with their meanings in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in the various embodiments of the present disclosure.

[0033] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the attached drawings.

[0034] Figure 1 is a diagram showing an optimal logistics packaging information generation and quotation automation system using generative AI.

[0035] Referring to FIG. 1, an optimal logistics packaging information generation and quotation automation system utilizing generative AI according to an embodiment may be configured to include a packaging information generation and quotation automation device (100) and a user terminal (200). The packaging information generation and quotation automation device (100) generates packaging design information and quotation data through a generative artificial intelligence model based on input data such as product specifications, weight, quantity, and transportation conditions received from the user terminal (200). In the embodiment, the packaging information generation and quotation automation device (100) receives product specification information from the user terminal (200) and transmits it to an internal data processing module. Subsequently, it includes a generative AI engine that receives product specifications and automatically generates the size of the packaging box, the internal cushioning material structure, the amount of packaging material required, wood cutting specifications, etc. Additionally, based on the generated packaging information, it generates a loading simulation image inside the container in a 2D or 3D form. Subsequently, the packaging information generation and quotation automation device (100) calculates packaging material unit costs, labor costs, transportation costs, etc., based on the packaging design results and automatically generates a standardized quotation. Subsequently, the design information, visual materials, and quotation generated in the embodiment are transmitted or stored in a user terminal (200) or a cloud-based storage. Additionally, a self-learning loop is performed to periodically improve the performance of the generative AI model based on user feedback, work history, and error data.

[0036] Meanwhile, the user terminal (200) can be implemented in the form of a web browser or an application and provides an interface that allows the user to input product information and view automatically generated design information, quotations, and visualization images in real time. In addition, the terminal can contribute to improving design accuracy and user satisfaction by transmitting user feedback, modification requests, etc., to the system.

[0037] The system for automating the generation of optimal logistics packaging information and quotations using generative AI according to the embodiment is differentiated from conventional technology in that it enables automatic design and quotation using only text-based product information without drawing files, automatically generates visual materials usable for customer service, and can be provided as a SaaS type, making it useful for small and medium-sized logistics and manufacturing companies.

[0038] FIG. 2 is a diagram illustrating a logistics packaging information generation and quotation automation platform utilizing generative AI according to an embodiment.

[0039] Referring to FIG. 2, the logistics packaging information automatic generation system according to an embodiment of the present invention is broadly composed of a data collection / linking step, a generative AI-based packaging information generation and quotation automation step, and a field application demonstration step. First, in the data collection / linking step, various metadata required for logistics packaging automation is collected. Specifically,

[0040] In the embodiment, logistics metadata includes information such as port authority and container information, and maritime export and import transportation costs from the Korea Customs Service, while timber metadata includes timber material information collected from port authorities and sawmills.

[0041] Furthermore, logistics packaging item information and quotation form data are collected based on data held by the AI ​​Hub and demonstration institutions. Based on this multi-source data, the generative AI-based packaging information generation platform performs the same sequential processing. First, in the foundational knowledge generation stage, various input data are preprocessed to be processed and refined into a form suitable for AI learning. Then, in the shipment information input stage, structured information such as actual product measurements, design drawings, and projected shipment quantities is entered into the system. Subsequently, in the prompt development stage, automatic generation of packaging information is performed through generative models (e.g., based on OpenAI, Gemini, or Deepseek) by optimizing question-response and search functions for the input information.

[0042] Accordingly, outputs such as material requirement calculation, automatic quotation generation, loading example images, 3D packaging information, and container loading simulation images are automatically generated. Furthermore, based on these calculation results, services such as AI-automated quotation generation, provision of packaging design optimization screens, and provision of 3D image-based customer communication materials are ultimately provided to the user.

[0043] Meanwhile, the system of the present invention is directly applied at the site of a logistics company, such as Lloyd's Logistics, to verify its effectiveness. Based on the results of the demonstration, scenario-based improvements, system enhancements, and the incorporation of user feedback are carried out. After the demonstration, the system is integrated into the interfaces of system users (logistics companies) and shippers (customers) and applied to actual logistics operations. This flow enables the application of generative AI-based logistics design automation technology to actual industrial sites, thereby generating various effects such as reduced design time, improved quotation accuracy, and the automatic generation of visual materials for customer response.

[0044] FIG. 3 is a block diagram of an optimal logistics packaging information generation and quotation automation device using generative AI according to an embodiment.

[0045] In the embodiment, the device for generating optimal logistics packaging information and estimating using generative AI may be configured as a server. In the embodiment, the server is a computing system that provides services to other computers or devices in a computer network or stores and manages data. The server (200) accepts requests from other computers or devices called clients and provides responses or data to those requests. The configuration of the server (200) shown in FIG. 2 is merely a simplified example.

[0046] The communication module (110) can be configured regardless of the mode of communication, such as wired or wireless, and can be configured with various communication networks, such as a Personal Area Network (PAN) or a Wide Area Network (WAN). Additionally, the communication module (110) can operate based on the known World Wide Web (WWW) and may utilize wireless transmission technologies used for short-range communication, such as Infrared Data Association (IrDA) or Bluetooth. For example, the communication module (110) may be responsible for transmitting and receiving data necessary to perform a technique according to one embodiment of the present disclosure.

[0047] Memory (120) may refer to any type of storage medium. For example, memory (120) may include at least one type of storage medium among flash memory type, hard disk type, multimedia card micro type, card type memory (e.g., SD or XD memory, etc.), RAM (Random Access Memory), SRAM (Static Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, magnetic disk, and optical disk. Such memory (120) may also constitute the database shown in FIG. 1.

[0048] The memory (120) can store at least one instruction that can be executed by the processor (130). Additionally, the memory (120) can store any form of information generated or determined by the processor (130) and any form of information received by the server (200). Additionally, the memory (120) stores various types of modules, instruction sets, or models.

[0049] The processor (130) can perform technical features according to embodiments of the present disclosure to be described below by executing at least one instruction stored in memory (120). In one embodiment, the processor (130) may be composed of at least one core and may include a processor for data analysis and / or processing, such as a central processing unit (CPU) of a computer device, a general purpose graphics processing unit (GPGPU), or a tensor processing unit (TPU).

[0050] This processor (130) can train a neural network or model designed in a machine learning or deep learning manner. To this end, the processor (130) can perform calculations for training the neural network, such as processing input data for training, extracting features from input data, calculating errors, and updating the weights of the neural network using backpropagation. Additionally, the processor (130) can perform inference for a specific purpose using a model implemented in an artificial neural network manner.

[0051] In an embodiment, the processor (130) receives product specifications, weight, and quantity information and generates packaging information including the dimensions of the packaging box, cushioning material structure, wood cutting specifications, and required amount of packaging materials using a generative artificial intelligence model. To this end, the processor (130) receives product-related information input from a user terminal (200). The product-related information may include the width, length, and height dimensions of the product, the weight of the product, the shipment quantity, handling precautions (such as potential for damage, loading direction, etc.), and the shipping method (air, sea, mixed). This input information may be received via a web-based UI or file upload method, and is refined into structured JSON, XML, or CSV formats and interpreted by an input processing module.

[0052] Subsequently, the processor (130) automatically configures a prompt for packaging design based on the input product information. In the embodiment, the prompt consists of specification questions that the generative AI will refer to, for example, “Suggest the optimal packaging box size when exporting a quantity of N products with a weight of W kg, width X mm, length Y mm, and height Z mm,” “Present a table of the cutting specifications and material requirements for wooden packaging under the above conditions,” and “Present the arrangement of cushioning materials for product protection during loading.” These prompts can be contextually enhanced by linking with existing learned prior knowledge and domain-based ontologies.

[0053] Subsequently, the above prompt is passed to a generative artificial intelligence model (e.g., GPT-based LLM, Gemini, DeepSeek, etc.), and the model outputs packaging design information including packaging box dimensions, internal cushioning structure, wood cutting specifications, and packaging material requirements. In the embodiment, the packaging box dimensions include the minimum / optimal outer dimensions for product loading, and the internal cushioning structure may include shock-protecting materials such as foam, jigs, and foaming agents, and their arrangement methods. The wood cutting specifications may include the cutting dimensions by standard size of the wood used, the number of cuts, and a cutting list. The packaging material requirements may include the required quantities of each material, such as wood, cushioning material, packaging tape, nails, and straps. In this process, the model ensures the realism and industrial applicability of the design results by referring to pre-trained packaging structure examples, material specification databases, and industry standards (e.g., IPPC, ISPM 15).

[0054] Subsequently, the output from the generative AI is post-processed by the processor (130) and converted into a structured data format. In the embodiment, packaging box information may be generated in a 3D model format such as STL or OBJ or a drawing format such as DXF or SVG, and material requirements and cutting specifications are stored in a specification format based on CSV or PDF, and each result is stored in a database within the system or transmitted to a user terminal.

[0055] According to one embodiment, the generated packaging information can be fed back along with evaluation scores or modifications after user review and accumulated as training data to improve the performance of the AI ​​model. This enables the realization of a self-adaptive structure in which accuracy and suitability are improved upon repeated use. In the embodiment, the processor (130) utilizes generative artificial intelligence based on input product information to automatically calculate key components required for packaging design without manual work, thereby realizing effects such as a drastic reduction in design time, improved packaging quality, and securing a basis for automated quotation.

[0056] Additionally, the processor (130) calculates a standard unit price based on the generated packaging information to generate an estimate, and generates a container loading simulation image based on the packaging information. To this end, the processor (130) extracts items necessary for calculating the estimate from the previously generated packaging information. For example, items such as the type and quantity of packaging materials used (e.g., wood, cushioning material, fixing jig, strap, etc.), the dimensions and quantity of packaging boxes, the quantity of products and the number of packaging operations, estimated labor costs (number of workers × hours), shipping methods, and transportation conditions can be extracted. This information is extracted as standardized fields from the packaging design results calculated by the AI ​​and is organized into tables for each item necessary for calculating the estimate.

[0057] Subsequently, the processor (130) calculates the cost for each item by referring to an internal unit price table or an externally linked material / labor unit price DB. For example, material costs can be calculated as material quantity × material unit price, and cutting costs can be calculated as number of cuts × cutting unit price. Packaging labor costs can be calculated as estimated work time × labor cost unit price, and other auxiliary materials and consumable costs can be calculated as item-specific quantity × unit price.

[0058] The processor (130) calculates the total packaging cost estimate by summing the above items and automatically generates a standard estimate form containing detailed items. The estimate is generated in an output format such as HTML, PDF, or XLS, and is automatically transmitted to the shipper or packaging company or stored in the system.

[0059] Subsequently, the processor (130) performs a three-dimensional loading simulation based on the generated packaging box information, shipment quantity, and container specification information. To this end, a container type such as 20ft, 40ft, or Open-top is selected according to the export destination and cargo volume, and a loading arrangement plan within the container is established based on the dimensions and quantity of the packaging boxes. In the embodiment, a loading arrangement plan can be established by considering the loading efficiency, loading stability, and center of gravity. Subsequently, the final arrangement result is generated as a visualized image through a 3D rendering engine or a CAD rendering module. In the embodiment, top, side, and front view images are generated, and key dimensions and the quantity of the arrangement can be displayed.

[0060] The above loading image can be utilized in various ways, such as marketing materials for customer response, loading plan documents for delivery to carriers, and materials for worker instructions. Additionally, the processor (130) can integrate the generated quotation information and container loading image to form a single quotation report. This report may include items such as a packaging overview and target product, material requirements and unit prices, total estimated amount, container loading simulation images, estimated delivery date, and work period. The report is sent to the customer or stored as a quotation history to be reused later or utilized for learning feedback. In the embodiment, the processor (130) can utilize packaging information automatically generated through generative AI to realize automation of quotation calculation, enhanced visual-based customer response, and ensuring connectivity between packaging design and quotation. This overcomes the limitations of existing manual-based quotation calculation methods and enables an organic connection between design, quotation, and logistics execution.

[0061] Additionally, the processor (130) outputs the generated packaging information, quotation, and loading images through a user interface. In an embodiment, the processor (130) can visually output the packaging information, quotation, and container loading simulation images calculated based on generative artificial intelligence through a user interface (UI) implemented on a user terminal (200) and support user interaction. To this end, the processor (130) converts each data item (packaging information, quotation, loading image) generated internally into a format suitable for the user interface. For example, packaging information can be converted from a JSON / XML format containing box dimensions, material specifications, internal structure, etc., into an HTML or SVG structure for UI output. Additionally, the quotation can be converted from tabular data containing item-specific amounts, material quantities, total costs, etc., into a PDF, XLS, or web-based quotation template. Additionally, the loading image can be converted from a 2D / 3D simulation result of packaging boxes loaded in a container into a WebGL, PNG, JPG, or SVG rendered image. All of this data is configured to be renderable on a web browser or dedicated application of the user terminal.

[0062] The processor (130) outputs information such as a packaging information visualization area, a quotation display area, and a loading simulation image area on the screen of the user terminal (200) based on the converted data structure. The packaging information visualization area displays the packaging box exterior, internal structure, cushioning material arrangement, wooden frame structure, etc., in the form of a 3D model or a plan view image.

[0063] Users can intuitively check the design structure through functions such as box rotation, zooming in / out, and cross-section viewing.

[0064] The quotation display area outputs detailed breakdowns such as the quoted amount, material unit prices, quantity-based calculations, labor costs, and transportation costs in a table format, and includes PDF or Excel download buttons.

[0065] Each item expands to detailed calculation history or material requirements information when clicked.

[0066] The loading simulation image area provides multiple viewpoints, such as front, side, and top views, for the generated container loading images; clicking displays a high-resolution image in a pop-up window. Additionally, simulation result analysis information, including loading efficiency (%), available space, and points of interest, is also displayed.

[0067] In addition, according to one embodiment, the user interface may provide functions for user feedback input, modification requests, and comment writing in addition to simple output functions. For example, input fields for changes to packaging information, suggestions for adjustments to quotation items, and evaluation selection results for loading images (e.g., “Loading stability high / average / unstable”) may be received. This feedback is then transmitted back to the processor (130) and may be used as training data for future model improvement or reflected as a regenerated result.

[0068] Additionally, the processor (130) can provide the generated result to the user through one or more of the following paths. For example, it can be printed directly from a web screen on a user terminal or downloaded in the form of a PDF, image, Excel, etc. Additionally, it can be automatically transmitted via email or message API, and the result can be transmitted through integration with an external system (ERP, logistics management system, etc.). Furthermore, the user can use the saving and version management functions for the generated output, and the result is automatically saved to a history database for reuse.

[0069] In the embodiment, the generative artificial intelligence model is based on a Large Language Model (LLM) and generates packaging information based on input natural language or text-based information of the product. In the embodiment, the generative artificial intelligence model is based on a Large Language Model (hereinafter “LLM”) and performs the function of automatically generating packaging information by receiving input of natural language or unstructured text-based information of the product. To this end, the user or the system inputs the following unstructured or natural language information. In the embodiment, the information may include product name and model name (e.g., “XYZ-500 Compressor”), product specifications (e.g., “Width 1200mm, Length 800mm, Height 650mm”), product weight (e.g., “Total weight is approximately 400kg”), product quantity (e.g., “5 units scheduled for export”), transport conditions (e.g., “Sea export, use of 40ft container”), and special notes (e.g., “Caution: Damage, cannot be stacked vertically”).

[0070] The processor (130) analyzes the input sentence and automatically generates a prompt in a format suitable for input to an LLM model. In the embodiment, the preprocessed input sentence is converted into a prompt configured so that the generative AI can understand the packaging design context. For example, it can be converted into a prompt such as Q1. Product name: XYZ-500 compressor, Q2. Dimensions: 1200mm × 800mm × 650mm, Q3. Weight: 400kg, Q4. Quantity: 5, Q5. Export method: Sea, 40ft container, Q6. Caution: Cannot be stacked vertically, cushioning material required. Specifically, it can be converted into a prompt such as, “Provide the optimal wooden packaging box dimensions, internal cushioning structure, material requirements, and cutting specifications for the export of the above product.”

[0071] This prompt is designed to guide accurate packaging design by leveraging the linguistic and domain knowledge acquired by LLMs. Additionally, it can enhance design consistency and industrial applicability by incorporating contextual information, such as product categories and industry standards.

[0072] The above prompt is input into a generative AI engine based on large-scale language models such as OpenAI GPT, Gemini, Cohere, DeepSeek, and Mistral. The LLM can output packaging information by referring to pre-trained technical documents, design knowledge, specification dictionaries, etc. The packaging information may include the external dimensions of the packaging box (per product or for batch packaging), wood specifications and cutting standards (e.g., 70x20x1500mm, 14EA), cushioning material application method and thickness (e.g., 50T expanded PE foam, applied to sides and bottom), material requirement calculations (e.g., total wood 2.3m³, strap 12m, etc.), and estimated weight and loading type per packaging unit. This output is derived in a natural language format, but is structured and visualized in a subsequent module to be used as a 3D packaging design image or a material specification sheet.

[0073] In addition, in the embodiment, packaging information generated from the LLM is standardized into a format including a material list in CSV / JSON format, a specification that can be converted into a PDF / HTML table for quotation, and dimension / position information for linking with a 3D modeling module. In the embodiment, the standardized information is output to a user interface through a processor (130) or transmitted to a subsequent module (quotation automation, loading simulation, storage, etc.).

[0074] As such, the generative artificial intelligence model of the present invention directly interprets natural language or unstructured data based on LLM and automatically calculates structural information and material specifications required for pavement design, thereby realizing an automated method for generating pavement information that integrates design, estimation, and visual data without product drawings. Unlike existing drawing-based design systems, this is security-friendly and enables the implementation of a field-oriented, flexible design process.

[0075] Additionally, the processor (130) generates packaging information by receiving unstructured data, such as product name, dimensions, weight, and order quantity, as input without product drawings. In an embodiment, the processor (130) automatically generates packaging information through a generative artificial intelligence model based on unstructured input data, such as product name, external dimensions, weight, and order quantity, without product drawing information. To this end, unstructured data in the following forms is received from a user terminal (200) or a linked system. This data may include unstructured text, natural language descriptions, handwritten input, etc. Such input can be received through various channels, such as web forms, Excel uploads, and email bodies, and is distinguished from existing design systems in that it is not in a structured form.

[0076] Subsequently, the processor (130) refines the unstructured data and separates it into semantic units through a parsing algorithm or preprocessing module based on natural language processing (NLP). For example, “approx. 1300mm × 900mm × 800mm” → width: 1300mm, length: 900mm, height: 800mm, “weight approx. 450kg” → weight: 450kg, “10 units scheduled for export” → quantity: 10, etc. The information is converted into a structured key-value pair form and is then inserted into a prompt template to be delivered to a generative artificial intelligence model.

[0077] Subsequently, the processor (130) generates a design request prompt composed of refined information. Specifically, it may include a prompt such as, "Provide the optimal wooden packaging specifications (box dimensions, type and quantity of materials, cushioning structure, cutting specifications) for exporting 10 units of the above product." The prompt is input into a generative artificial intelligence model (e.g., GPT-based LLM, Gemini, etc.) to infer and design the packaging specifications of the product without a drawing.

[0078] In addition, the generative AI model automatically generates packaging information based on input prompts. In the embodiment, the packaging information may include the external dimensions of the packaging box, material requirements, the structure of the internal cushioning material, and wood cutting specifications. The external dimensions of the packaging box are the optimal box specifications based on product size and quantity, and the material requirements are the required quantities for each material, such as wood (by specification), straps, cushioning material, and tape. The structure of the internal cushioning material is the structure and application location of materials such as foam, jigs, and fixing frames, and the wood cutting specifications are a list of cutting dimensions and quantities. This information is output in natural language form and is structured into CSV, PDF, 3D models, etc., by a subsequent module.

[0079] Afterward, the processor (130) standardizes the generated packaging information and outputs it through a user interface or transmits it to an automated quotation and loading simulation module. At this time, the packaging information can be visualized as a 3D image of the box's exterior and internal structure, a material list and quantity table, cutting specifications and work instructions, etc.

[0080] Through this, the embodiment automatically performs packaging design using only unstructured text information such as product name, size, weight, and quantity, without utilizing detailed product drawings. This enables packaging design automation even in manufacturing environments where drawing transmission is difficult, security-restricted environments, and small and medium-sized logistics company environments. Furthermore, since the generative AI-based reasoning method reflects packaging examples learned from various industries, it is effective in systematizing experience-based design capabilities and ensuring design consistency and quality.

[0081] Additionally, the processor (130) generates an estimate in a standard format including the type and quantity of packaging materials used, labor costs, and design costs based on the generated packaging information.

[0082] To this end, the processor (130) extracts quotation items based on packaging information. Specifically, material information (names and specifications of material items such as wood, plywood, foam, cushioning material, jig, strap, tape, etc.), material quantity (quantity used or number of cuts for each material), packaging unit and quantity (packaging unit per product, total packaging quantity), work time (estimated work time (personnel × time)), design difficulty or complexity index (design deviation and correction coefficient based on automatic design), etc.

[0083] Subsequently, the processor (130) references standard unit price information required for estimation calculation from an internal DB or an external linked API. Among the items, material costs can be calculated by multiplying the unit price of each material by the quantity. Labor costs are calculated by multiplying the estimated packaging time by the standard unit price of labor costs. Design costs are calculated based on a fixed unit price or weighting, taking into account the complexity of the automatically generated design or the nature of the drawing replacement. Additionally, auxiliary materials and other costs are calculated through the calculation of consumable material items such as straps, packaging materials, nails, and tape. All of these cost items are calculated including information such as subtotals, totals, and whether VAT is included.

[0084] Subsequently, the processor (130) automatically generates a standard quotation format based on the above calculation results. The generated quotation is provided to the user in the form of a web screen output (HTML-based quotation UI), automatic saving or download in formats such as PDF, Excel, or CSV, or via email or transmission linked to an ERP / logistics management system. In addition, the quotation history is stored in an internal system, allowing for reuse or comparative analysis when quoting the same product repeatedly.

[0085] In this way, the processor (130) of the present invention can realize full automation of the logistics design and quotation process by using packaging information calculated based on generative AI without product drawings to automatically calculate practical quotation elements such as materials, manpower, and design, and by automatically generating a standard quotation that can be used for customer service or transaction documents. This offers superior speed, accuracy, and consistency compared to manual quotation and provides high usability in various environments, such as small and medium-sized export companies and logistics agencies.

[0086] Additionally, the processor (130) generates a three-dimensional image or a rendering image of the generated packaging box arranged in a container loading form and provides it to the user. In an embodiment, the processor (130) can simulate the arrangement of the box in a container loading form based on the generated packaging box information, and generate a three-dimensional image or a rendering image that visualizes this and provide it to the user.

[0087] To this end, the processor (130) receives packaging box related data from a generative artificial intelligence model or design module. The packaging box related data may include the external dimensions of the packaging box (length × width × height), the quantity of packaging boxes, weight information (optional), packaging direction restriction information (e.g., cannot be stacked upright), and the stacking unit (whether a pallet is included, etc.). In the embodiment, the data is input in JSON, CSV, or structured list format and is used as reference data for stacking optimization.

[0088] Subsequently, the processor (130) retrieves container type information selected by the user or automatically recommended by the system. For example, the container type information may include a 20ft or 40ft ISO ocean container, an open-top container, a flat-rack container (optional), an effective loading space within the container (units: mm, m³), ​​a maximum weight, an allowable loading height, etc. This information is referenced from a logistics DB or a pre-registered container specification database and is used to define loading conditions.

[0089] Subsequently, the processor (130) performs a simulation to optimize the placement of packaging boxes in a three-dimensional space, taking into account the dimensions and quantity of the packaging boxes and space constraints within the container. In the embodiment, the placement simulation algorithm may consider conditions such as maximizing the filling rate (ratio of the actual box volume to the loading space), considering the center of gravity and weight distribution, directional restriction conditions such as preventing items that are fragile from being placed on top and heavy items on the bottom, requiring items to be loaded horizontally, the direction of container door opening, and the loading order. In the embodiment, the placement result is calculated as coordinate information (x, y, z) that positions the packaging box model in a virtual 3D space.

[0090] In the embodiment, coordinate information can be calculated through mathematical formula 1.

[0091] Mathematical formula 1

[0092]

[0093] In Equation 1, xi, yi, and zi are the placement coordinates (unit: mm) of the i-th box within the container, and x_start, y_start, and z_start are the origin positions relative to the container. hi, wi, di, and hi are the width, depth, and height of the i-th packing box, and ny and nz are the number of boxes that can be placed in a single line along the y-axis and z-axis directions within the container. ] is a floor function.

[0094] In the embodiment, the processor (130) can convert the box arrangement into a row-column-layer structure using Equation 1, and assign three-dimensional coordinates in a manner that aligns along the row in the x-axis direction, accumulates along the column in the y-axis direction, and stacks layers in the z-axis direction. Additionally, the processor (130) can set conditions such that for fragile items, the ziy_start value is placed above a certain level for priority placement, and for items heavier than a certain weight, zi is placed first starting from the bottom position close to the reference coordinate. Furthermore, in the embodiment, if there is a direction restriction, the axis change of wi and hi can also be applied by determining whether the box is rotated.

[0095] Additionally, the processor (130) receives the dimensions and quantity of the packaging boxes, as well as the actual weight value of each packaging box in real time from an IoT weight sensor or measuring device, and places the packaging boxes in three-dimensional space based on the weight value so that the gravity center inside the container aligns with the centerline of the container.

[0096] Specifically, the processor (130) constructs a multidimensional input vector including the dimensions and weight values ​​of each box, and when simulating container loading, applies an objective function-based placement algorithm that includes not only a simple fill rate (max fill rate) but also a horizontal alignment condition of the center of gravity vector (CG vector). At this time, the position coordinates (xi, yi, zi) of the box are optimized to satisfy the placement conditions. In the embodiment, the placement conditions may include at least one of the following: even distribution of weight load based on the container floor, placement of heavy boxes at the bottom and light boxes at the top, placement of damage-caution items at the top or impact-minimum area, and determination of the loading order in the order of the container door opening direction.

[0097] Additionally, the processor (130) is configured to calculate the center of gravity coordinates after the entire loading is completed, and to confirm the corresponding layout as an output result only if the vector difference with the center inside the container is within the allowable error range.

[0098] In this way, the present embodiment goes beyond simple space optimization or box coordinate allocation and implements an intelligent 3D loading design that reflects physical load safety and real-time weight data, thereby enabling the achievement of advanced loading quality required in logistics sites, such as preventing the risk of container tipping, reducing the damage rate, and ensuring loading and unloading safety.

[0099] Afterward, based on the above placement coordinates, the processor (130) generates a visualization image in one or more ways among a 3D model image, a front / side / top viewpoint image, a cross-section view, or a transparency rendering image.

[0100] 3D model images visualize the actual loading shape using 3D rendering engines such as WebGL, Three.js, and Babylon.js, while front / side / top view images generate 2D images in PNG, JPG, or SVG formats, or isometric images. The cross-section view or transparency rendering function generates visual materials for verifying the loading status inside the container. Visual elements such as box quantity, box number, position number, filling rate (%), and clearance space may be displayed together in the rendered images.

[0101] Subsequently, the processor (130) can provide the generated loading image as a real-time output on a web UI (including 3D model rotation and zooming) or as a downloadable PDF or PNG file. Additionally, user feedback or requests for modification regarding the loading image can be recorded and reflected in regeneration or model improvement. The processor (130) according to the embodiment can optimize transportation and improve design accuracy by utilizing packaging information generated based on generative AI to perform container loading simulations and automatically generating and providing a 3D image or rendering image that visualizes the results. Additionally, customer trust can be improved.

[0102] In the embodiment, the user interface is configured as a web-based subscription (SaaS) structure, allowing multiple small and medium-sized export manufacturers or 3PL companies to view and download the generated design results online. The user interface according to the embodiment is configured as a web-based subscription (SaaS: Software as a Service) structure and is designed to allow multiple users to access it simultaneously via the Internet.

[0103] In particular, users such as small and medium-sized export manufacturers or 3PL (Third Party Logistics) companies can access the system through a web browser without separate installation and view generated packaging design results, quotations, loading simulation images, etc. online in real time or download them as files.

[0104] Furthermore, the user interface visually integrates packaging information, material specifications, quoted prices, and container loading images automatically generated by a generative AI model, and enhances the practical applicability and ease of sharing of design results by supporting various output formats such as HTML5, WebGL, PDF, and CSV. Additionally, users can log into the system based on their accounts to view the generation history of each design result, compare with previous versions, save templates, and input feedback, thereby minimizing repetitiveness and inefficiency in design and estimation tasks.

[0105] The user interface according to the embodiment maximizes accessibility and scalability of packaging information generation results through a web-based SaaS platform, thereby enabling logistics and manufacturing companies of various sizes to receive packaging design services of the same quality without constraints of time and place.

[0106] Additionally, the processor (130) includes a self-learning loop that periodically evaluates a generative AI model based on feedback data from a user and improves the accuracy of the packaging design and the precision of the estimate prediction through learning. In an embodiment, the processor (130) includes a self-learning loop that periodically evaluates the performance of a generative artificial intelligence model based on feedback data received from a user terminal (200) and improves the accuracy of the packaging design and the precision of the estimate prediction through relearning.

[0107] Specifically, feedback data entered through the user interface includes qualitative and quantitative information regarding satisfaction with packaging design results or quotation results provided by the generative AI, the presence of errors, and requests for modifications; for example, it may include comments or evaluation scores such as “insufficient cushioning material placement,” “oversized box design,” or “quotation higher than actual cost.”

[0108] The processor (130) classifies such feedback data by design result item, performs statistical analysis on error occurrence patterns and repetition frequencies, and then quantifies the output results of the generative artificial intelligence model into evaluation indicators (e.g., accuracy, match rate, user satisfaction, etc.) based on this. These evaluation results are automatically reflected at the time of model performance check according to a preset period or the number of accumulated feedback cases, and if precision below a certain standard is confirmed, the process of re-tuning the model parameters by updating the AI ​​training data or adding a correction dataset for error patterns is performed.

[0109] This self-learning loop enables the continuous improvement of factors such as the design suitability of the AI ​​model, the reliability of estimate predictions, and container loading efficiency, and allows for the provision of results with increasingly high realism and industrial applicability as user feedback accumulates for the same input values.

[0110] As a result, the processor (130) of the present invention can realize a self-optimizing AI-based logistics design system that learns on its own and enhances accuracy through repeated use and interaction by the user, which provides significant technical effects in terms of maintaining long-term performance and the possibility of evolution compared to existing fixed rule-based design systems.

[0111] Additionally, the processor (130) may include a function to dynamically correct or regenerate existing packaging information based on data such as images, dimensions, and weight collected in real time from IoT sensor modules or vision sensors installed at the packaging work site. For example, after a user inputs product specifications to generate packaging information, when the actual product is received, the IoT sensor automatically recognizes the actual size, external shape, and pallet loading status of the product, and the processor (130) compares the above information with the existing generated information and automatically recalculates the design drawings and material requirements if an error exists. Through this, packaging errors caused by differences between the input values ​​and the actual product specifications are prevented, and sensor-based automatic correction is enabled for irregular / immeasurable products.

[0112] In the example, packaging information can be corrected through mathematical formula 2.

[0113] Mathematical formula 2

[0114]

[0115] In Equation 2, P^ is corrected packaging information reflecting real-time sensor data, and P0 is initial packaging information initially generated by a generative AI model. The initial packaging information may include box dimensions, material quantity, loading direction, etc. D_sensordiz_start is the origin position relative to the container. hi, wi are measurement data collected in real-time from IoT sensors or vision sensors. The measurement data may include actual dimensional errors, weight, positional discrepancies, visual distortion, etc. ΔP is a sensor-based error correction term. f(·) is a dynamic correction function that calculates a correction value based on the initial packaging information P0P_0P0 and the sensor input value D_sensor. In the embodiment, the dynamic correction function may include a linear correction function, a weighted average-based correction function, a non-linear function, a normalization function, a rule-based condition correction function, a machine learning-based function, etc.

[0116] In the embodiment, the pavement design value P0, which is calculated in advance based on generative AI through mathematical formula 2, is compared with actual measured field data, and the structure can be corrected by ΔP or redesigned based on the error with the actual value. This can be utilized as a calculation model in a self-adaptive pavement design system based on a real-time feedback loop, and supports effects such as error prevention, minimization of material waste, and improvement of design reliability.

[0117] FIG. 4 is a diagram illustrating the linkage structure of a logistics packaging service metadata collection and generative artificial intelligence system according to an embodiment.

[0118] Referring to FIG. 4, the artificial intelligence system according to the embodiment may be composed of a metadata collection unit, a data management and linkage system, and a generative AI-based packaging information generation system. In the embodiment, the metadata collection unit is configured to secure base data for system construction and consists of detailed data. The detailed data may include logistics metadata, timber metadata, and data held by the demonstration agency. Logistics metadata is collected in XML format, such as container information provided by the Port Authority and maritime import / export transportation cost information from the Korea Customs Service. Timber metadata is secured in CSV format, consisting of timber and sawing-related data collected from the Port Authority and sawmills. Data held by the demonstration agency includes historical data such as timber pallet production history, quotation forms, and specifications.

[0119] The collected input data consists of various attribute values, such as container information, maritime transport costs, material information, timber pallet specifications, and quotation standards, and undergoes data normalization and standardization processing. During this process, the data is converted into a standard format for each item and stored as "structured data." Subsequently, the structured data is integrated with various external or internal information through a data linkage system.

[0120] In the embodiments, the linkage system includes an external data linkage API, a public data receiving module, an internal file linkage module, etc., and these systems can be linked with structured data in real-time or asynchronously. The structured data linked in this way is input into the generative AI system and utilized for various generative tasks such as packaging design automation, cost estimation, and the generation of customer response documents. Furthermore, the generative AI system is largely composed of a user UI service, a chatbot solution, a management service, and a history analysis service. The user UI service enables users to make design requests or perform Q&A through a chatbot interface, and the chatbot solution can perform message processing, memory services, conversation modules, and knowledge reference functions. Additionally, the management service includes conversation history, statistical analysis, and user permission management. Moreover, the generative AI system is linked with external systems and existing legacy systems, allowing past design history, cost estimation databases, pallet production data, etc., to be utilized as training data or as a result repository.

[0121] As described above, the generative artificial intelligence system according to the embodiment enables design automation and decision support based on generative AI, based on the collection and standardization of industrial data specialized for logistics packaging tasks, thereby automating existing manual-centered design and estimation tasks and facilitating a transition to an AI-based data-centric logistics design ecosystem.

[0122] Below, we will look at FIG. 5. The method for generating optimal logistics packaging information and automating quotations using generative AI illustrated in FIG. 5 can be performed by an optimal logistics packaging information generation and quotation automation device (100) using generative AI including a processor (130).

[0123] Meanwhile, FIG. 5 is merely illustrative, and the concept of the present invention is not to be interpreted as being limited to that shown in FIG. 5. For example, each step may be configured in a different order than that shown in FIG. 5, at least one of the steps shown in FIG. 5 may not be performed, or one or more steps not shown in FIG. 5 may be additionally performed.

[0124] Below, the method for generating optimal logistics packaging information and automating quotations using generative AI will be explained in turn. Since the operation (function) of the method according to the embodiment is essentially the same as the function of the system, descriptions that overlap with FIGS. 1 to 4 will be omitted.

[0125] Figure 5 is a diagram illustrating the process of generating optimal logistics packaging information and automating quotations using generative AI according to an embodiment.

[0126] Referring to FIG. 5, in step S110, product specifications, weight, and quantity information are received, and packaging information including the dimensions of the packaging box, cushioning material structure, wood cutting specifications, and required amount of packaging materials is generated using a generative artificial intelligence model. In step S120, a standard unit price is calculated based on the generated packaging information to generate an estimate. In step S130, a container loading simulation image is generated based on the packaging information. In step S140, the generated packaging information, estimate, and loading image are output through a user interface.

[0127] Meanwhile, the methods according to the various embodiments of the present invention described above can be implemented in the form of an application or software program that can be installed on an existing electronic device.

[0128] In addition, the whole or part of the method may be composed of multiple software function modules and implemented on an operating system (OS). Alternatively, each step may be composed of a single software function module, or each step may be combined to form a single software function module and implemented on an operating system. Therefore, even if all of the embodiments of the present disclosure are not implemented as a single software function module, if multiple software function modules implement each step of the present disclosure and multiple software function modules are implemented on a single operating system, it can be understood that the method of the present disclosure has been implemented.

[0129] In addition, the methods according to the various embodiments of the present invention described above can be implemented solely through software upgrades or hardware upgrades of existing electronic devices. Furthermore, the various embodiments of the present invention described above can also be performed through an embedded server equipped in an electronic device or an external server of the electronic device.

[0130] Meanwhile, according to one embodiment of the present invention, the various embodiments described above may be implemented as software comprising instructions stored on a computer-readable recording medium using software, hardware, or a combination thereof. In some cases, the embodiments described herein may be implemented as the processor itself. According to the software implementation, embodiments such as the procedures and functions described herein may be implemented as separate software modules. Each of the software modules may perform one or more functions and operations described herein.

[0131] Meanwhile, a computer or a similar device may include a device according to the disclosed embodiments, which is capable of calling instructions stored from a storage medium and operating according to the called instructions. When said instructions are executed by a processor, the processor may perform a function corresponding to said instructions directly or by using other components under the control of said processor. The instructions may include code generated or executed by a compiler or an interpreter.

[0132] A computer-readable recording medium may be provided in the form of a non-transitory computer-readable recording medium. Here, "non-transitory" simply means that the storage medium does not contain a signal and is tangible, without distinguishing whether data is stored semi-permanently or temporarily on the storage medium. In this context, a non-transitory computer-readable medium refers to a medium that stores data semi-permanently and is readable by a device, rather than a medium that stores data for a short moment, such as registers, caches, or memory. Specific examples of non-transitory computer-readable media may include CDs, DVDs, hard disks, Blu-ray discs, USBs, memory cards, and ROMs.

[0133] As described above, exemplary embodiments have been disclosed in the drawings and specification. Although specific terms have been used to describe the embodiments in this specification, they are used only for the purpose of explaining the technical concept of this disclosure and are not intended to limit the meaning or the scope of this disclosure as defined in the claims. Therefore, those skilled in the art will understand that various modifications and equivalent alternative embodiments are possible therefrom. Accordingly, the true technical scope of protection of this disclosure should be determined by the technical concept of the appended claims.

[0135] This patent application was filed as part of the following project.

[0137]

Claims

Claim 1 A memory storing at least one command for automatically generating logistics packaging information and quotation information based on product specification information; and includes a processor that performs operations according to the above command, wherein the processor receives product specifications, weight, and quantity information, generates packaging information including packaging box dimensions, cushioning material structure, wood cutting specifications, and packaging material requirements using a generative artificial intelligence model, calculates a standard unit price based on the generated packaging information to generate an estimate, generates a container loading simulation image based on the packaging information, and outputs the generated packaging information, estimate, and loading image through a user interface, wherein the generative artificial intelligence model is based on a Large Language Model (LLM) and generates packaging information based on input natural language or text-based information of the product, wherein the processor generates packaging information by receiving unstructured data such as product name, dimensions, weight, and order quantity as input without product drawings, wherein the processor performs a simulation of placing packaging boxes in a 3D space considering the dimensions and quantity of packaging boxes and space constraints within the container, and the placement result is calculated as coordinate information (x, y, z) for positioning a packaging box model in a virtual 3D space, and wherein the processor [performs] the box placement Converted into a row-column-layer structure, 3D coordinates are assigned in a format where rows are aligned along the x-axis, columns are accumulated along the y-axis, and layers are stacked along the z-axis; the processor sets conditions such that for fragile items, the zi value is prioritized to be positioned above a certain level, and for items heavier than a certain weight, the zi value is prioritized to be placed starting from the bottom position closest to the reference coordinate; the actual weight value of each packaging box is received in real time, and based on the weight value, the packaging box is placed in 3D space so that the gravity center inside the container aligns with the container centerline.A generative AI-based logistics packaging information generation and quotation automation device that dynamically corrects generated packaging information based on real-time collected image, dimension, and weight data, compares packaging design values ​​calculated in advance based on generative AI with actual measured field data, and corrects the packaging design values ​​by the difference between the design values ​​and the field data. Claim 2 delete Claim 3 delete Claim 4 In claim 1, the processor generates an estimate in a standard format including the type and quantity of packaging materials used, labor costs, and design costs based on the generated packaging information, a generative artificial intelligence-based logistics packaging information generation and quotation automation device. Claim 5 In claim 1, the processor generates a three-dimensional image or a rendering image of the generated packaging box arranged in a container loading form and provides it to the user, a generative artificial intelligence-based logistics packaging information generation and quotation automation device. Claim 6 In claim 1, the user interface is configured as a web-based subscription (SaaS) structure, and the generative AI-based logistics packaging information generation and quotation automation device enables multiple small and medium-sized export manufacturers or 3PL companies to check and download generated design results online. Claim 7 A device for automating the generation of logistics packaging information and quotation based on generative artificial intelligence according to claim 1, wherein the processor includes a self-learning loop that periodically evaluates a generative AI model based on feedback data from a user and improves packaging design accuracy and quotation prediction precision through learning.

Images