Artificial intelligence-based user-customized modular building design and BIM linkage system

Abstract

The present invention relates to an artificial intelligence-based user-customized modular building design and BIM linkage apparatus comprising: a memory including at least one instruction; and at least one processor electrically connected to the memory and configured to perform the at least one instruction, wherein the at least one processor may: receive a user input related to generation of housing space data, wherein the user input includes natural language; apply space requirement analysis criteria to the user input to calculate a housing requirement index; determine, by means of a specification mapping model, hard specifications and soft specifications on the basis of the housing requirement index; if the housing requirement index cannot be calculated or if there exists an element, among the hard specifications, which cannot be calculated by the housing requirement index alone, analyze insufficient information to determine necessary additional requirement information; request, from a user, information necessary to obtain the additional requirement information; reflect an additional input from the user to sequentially update the housing requirement index, the hard specifications, and the soft specifications; and calculate, by means of a space generation model, housing space data on the basis of the updated hard specifications and soft specifications.

Description

AI-based customized modular building design and BIM integration system

[0001] The present invention relates to artificial intelligence-based modular building housing design and BIM linkage technology.

[0002] Recently, attempts to automate design using artificial intelligence (AI) have been made in the architectural design field. Conventional AI-based architectural design solutions have primarily been used to generate ideas during the planning or building scale estimation stages. They remain limited to the concept design stage, failing to connect to the detailed design or costing stages. This has led to delays in the creation of drawings for actual construction and in the cost estimation process. For example, while image-based generative AI technology can automatically generate architectural concept design images, this approach makes it difficult to obtain detailed design drawings that reflect essential requirements such as structural safety and building codes. Consequently, modifying AI-generated concept designs to make them constructible requires professional designers to reinterpret the initial concept and create a new BIM model, resulting in significant time and cost.

[0003] Furthermore, conventional technology has made it difficult to immediately calculate quantities or estimate construction costs for AI-generated designs, delaying cost impact analysis when design changes occur. While modular construction facilitates design changes by assembling standardized blocks, existing systems fail to interpret AI output results on a module-by-module basis or link them to BIM libraries, preventing automated estimates or detailed drawing generation.

[0004] In summary, the problems with prior technology are as follows. First, AI-based architectural design has been limited to conceptual design and has not been linked to detailed design and manufacturing information. Second, image-based AI designs have failed to meet structural and regulatory requirements, requiring additional manual work. Third, immediate quantity and cost estimation for design proposals has been difficult. To overcome these limitations, a new system is needed that integrates BIM databases and modular block-level design with AI, enabling integrated management of information from the design stage through construction.

[0005] One embodiment of the present invention relates to an artificial intelligence-based customized modular building design and BIM linkage system capable of collecting information suitable for creating a building and automatically creating a building based on the collected information.

[0006] One embodiment of the present invention relates to an artificial intelligence-based customized modular building design and BIM linkage system that can generate customized housing space data for users by taking into account whether the building complies with building regulations.

[0007] One embodiment of the present invention relates to an artificial intelligence-based customized modular building design and BIM linkage system that can generate housing space data based on BIM data and generate an expected estimate when constructing a building.

[0008] One embodiment of the present invention relates to an artificial intelligence-based customized modular building design and BIM linkage system that can automatically generate spatial data for various sized housing types, and can be applied to the design of low-rise (3-4 floors) or mid-rise modular buildings.

[0009] According to one embodiment of the present invention, a device for designing a modular building based on artificial intelligence and linking with a BIM system is provided, comprising: an electronic device comprising: a memory including at least one instruction; and at least one processor electrically connected to the memory and configured to perform the at least one instruction; wherein the at least one processor receives a user input related to generating housing space data, the user input being configured in a natural language, applies a space demand analysis criterion to the user input to calculate a housing demand index, determines hard specifications and soft specifications through a specification mapping model based on the housing demand index, and, when the housing demand index cannot be calculated or there is an element among the hard specifications that cannot be calculated based on the housing demand index alone, analyzes insufficient information to determine necessary additional demand information, requests the user for information necessary to obtain the additional demand information, sequentially updates the housing demand index, hard specifications, and soft specifications by reflecting the additional input of the user, and calculates housing space data through a space generation model based on the updated hard specifications and soft specifications.

[0010] The above housing needs indicators may include openness, storage space ratio, privacy needs, energy efficiency, and budget weighting.

[0011] The at least one processor can calculate the housing demand index in the range of 0 to 10 points and map hard specifications and soft specifications according to the value of the housing demand index.

[0012] The at least one processor may request additional input from the user to derive one or more of the housing need indicators if the user input alone cannot yield the indicators.

[0013] The at least one processor may request additional input from the user to derive the hard specifications if the hard specifications cannot be determined solely from the user input and the housing demand indicators.

[0014] The above at least one processor can refer to building standards including legal information among hard specifications, and adjust the space layout by applying building coverage ratio, floor area ratio, floor limit, and sunlight right criteria to the housing space data according to the building standards.

[0015] The above electronic device further includes a modular block library, wherein the modular block library includes information on specifications, uses, materials, unit prices, and edge conditions of each block, and the space generation model can generate housing space data by calculating a block combination for space configuration based on the modular block library.

[0016] The above space generation model can convert the generated housing space data into BIM data in the Industry Foundation Classes (IFC) format.

[0017] The at least one processor may provide a construction cost estimate based on the BIM data.

[0018] A pricing model may further be included that calculates the construction cost in real time by aggregating the quantity per block based on the above BIM data and applying unit price information.

[0019] According to one embodiment of the present invention, a method for designing a modular building customized for an artificial intelligence-based user and linking with a BIM is provided, the method comprising: receiving, by a memory and at least one processor electrically connected to the memory, a user input related to generating housing space data, the user input being composed in a natural language; applying a space demand analysis criterion to the user input to derive a housing demand index; determining hard specifications and soft specifications through a specification mapping model based on the housing demand index; determining necessary additional demand information by analyzing insufficient information when the housing demand index cannot be derived or there is an element among the hard specifications that cannot be derived solely from the housing demand index; requesting the user for information necessary to obtain the additional demand information; sequentially updating the housing demand index, the hard specifications, and the soft specifications by reflecting the additional input from the user; and generating housing space data through a space generation model based on the updated hard specifications and soft specifications.

[0020] The above housing needs indicators may include openness, storage space ratio, privacy needs, energy efficiency, and budget weighting.

[0021] The artificial intelligence-based customized modular building design and BIM linkage method can calculate the housing demand index in the range of 0 to 10 points and map hard specifications and soft specifications according to the value of the housing demand index.

[0022] The AI-based customized modular building design and BIM linkage method may request additional input from the user to calculate the indicators if one or more of the housing demand indicators cannot be calculated using the user input alone.

[0023] The disclosed technology may have the following effects. However, this does not mean that a particular embodiment must include all or only the following effects, and thus the scope of the disclosed technology should not be construed as being limited thereby.

[0024] The present invention relates to an artificial intelligence-based customized modular building design and BIM linkage system according to one embodiment of the present invention, which can generate customized housing space data for the user by taking into account whether the building complies with regulations.

[0025] The present invention relates to an artificial intelligence-based customized modular building design and BIM linkage system according to one embodiment of the present invention, which can generate customized housing space data for the user by taking into account whether the building complies with regulations.

[0026] The present invention relates to an artificial intelligence-based customized modular building design and BIM linkage system according to one embodiment of the present invention, which can generate housing space data based on BIM data to generate an expected estimate when constructing a building.

[0027] Figure 1 is a diagram showing the configuration and operation flow of an artificial intelligence-based concept design automation system according to conventional technology.

[0028] FIG. 2 is a block diagram schematically showing the configuration and operation flow of an artificial intelligence-based customized modular building design and BIM linkage system according to one embodiment of the present invention.

[0029] FIG. 3 is a drawing of an artificial intelligence-based customized modular building design and BIM linkage system according to one embodiment of the present invention.

[0030] FIG. 4 is a drawing illustrating the physical configuration of an artificial intelligence-based customized modular building design and BIM linkage device according to one embodiment of the present invention.

[0031] FIG. 5 is a drawing for explaining the operation of an artificial intelligence-based customized modular building design and BIM linkage device according to one embodiment of the present invention.

[0032] Figure 6 is a drawing for explaining the configuration of a model according to one embodiment of the present invention.

[0033] FIG. 7 is a drawing illustrating the operation sequence of an artificial intelligence-based customized modular building design and BIM linkage device according to one embodiment of the present invention.

[0034] FIG. 8 is a diagram illustrating a process of conducting a user survey according to one embodiment of the present invention.

[0035] FIG. 9 is a drawing for explaining a process of creating a space according to one embodiment of the present invention.

[0036] The description of the present invention is merely an example for structural and functional explanation, and therefore, the scope of the present invention should not be construed as being limited by the embodiments described in the text. That is, since the embodiments can be modified in various ways and can take various forms, the scope of the present invention should be understood to include equivalents that can realize the technical idea. In addition, the purposes or effects presented in the present invention do not mean that a specific embodiment must include all of them or only such effects, and therefore, the scope of the present invention should not be construed as being limited thereby.

[0037] Meanwhile, the meaning of the terms described in this application should be understood as follows.

[0038] Terms such as "first" and "second" are intended to distinguish one component from another, and the scope of the rights should not be limited by these terms. For example, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component.

[0039] When a component is said to be "connected" to another component, it should be understood that while it may be directly connected to that other component, there may also be other components intervening. Conversely, when a component is said to be "directly connected" to another component, it should be understood that there are no other intervening components. Similarly, other expressions describing relationships between components, such as "between" and "directly between," or "adjacent to" and "directly adjacent to," should be interpreted similarly.

[0040] Singular expressions should be understood to include plural expressions unless the context clearly indicates otherwise, and terms such as "comprises" or "have" should be understood to specify the presence of a feature, number, step, operation, component, part or combination thereof, but not to exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

[0041] For each step, the identifiers (e.g., a, b, c, etc.) are used for convenience of explanation and do not describe the order of the steps. The steps may occur in a different order than stated unless the context clearly dictates a specific order. That is, the steps may occur in the same order as stated, may be performed substantially simultaneously, or may be performed in the opposite order.

[0042] The present invention can be implemented as computer-readable code on a computer-readable recording medium, and the computer-readable recording medium includes all types of recording devices that store data that can be read by a computer system. Examples of the computer-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc., and also includes those implemented in the form of a carrier wave (e.g., transmission via the Internet). Furthermore, the computer-readable recording medium can be distributed across network-connected computer systems, so that the computer-readable code can be stored and executed in a distributed manner.

[0043] Unless otherwise defined, all terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this invention pertains. Terms defined in commonly used dictionaries should be interpreted to be consistent with their meaning within the context of the relevant technology, and should not be interpreted as having an idealized or overly formal meaning unless explicitly defined herein.

[0044]

[0045] Figure 1 is a diagram illustrating the configuration and operational flow of a conventional AI-based concept design automation system. As illustrated in Figure 1, conventional systems use an AI model to automatically generate a concept-level design (S2) based on user input (S1), and the resulting design is then visually output (S3). This approach primarily interprets user requests based on natural language or keywords, and simply presents conceptual ideas regarding architectural style or spatial layout as images or 3D shapes.

[0046] Furthermore, since the technical configuration in Figure 1 does not reflect information necessary for actual construction, such as building codes, structural stability, indoor and outdoor facility conditions, and constructability, the automatically generated design results can only be used as a reference during the planning stage. Furthermore, the output design is not linked to BIM data or quantity estimation information, limiting the ability to continuously automate the process from estimation to construction planning.

[0047] Therefore, although the conventional technology according to Fig. 1 can be utilized for the initial conception of design ideas, it has the problem of being difficult to utilize as basic information for actual detailed design or construction execution.

[0048]

[0049] FIG. 2 is a block diagram schematically illustrating the configuration and operation flow of an artificial intelligence-based customized modular building design and BIM linkage system according to one embodiment of the present invention. As illustrated in FIG. 2, the system according to the present invention receives a user input (100), and the input can be analyzed through a natural language interpretation module (200) using a large-scale language model (LLM). In this process, a housing demand index (210) based on the user's housing preference or requirement is derived, and this index can be converted into hard and soft specifications by linking it to a block DB (300) through a specification mapping module (400).

[0050] The specification mapping results are transmitted to the space creation module (500), which automatically creates space layout data that meets structural conditions, regulatory restrictions, and user needs, and the created space data can be converted into a model (510) in BIM (Industry Foundation Classes; IFC) format. The space creation results are also simultaneously transmitted to the quantity aggregation module (600), which performs cost estimation based on attribute information such as materials, equipment, and finishing materials included in each space block, thereby enabling quotation generation (610).

[0051] The present invention can be implemented in various modified and expanded forms in addition to the configuration illustrated in FIG. 2, and specific embodiments and processing procedures for each component are described in more detail below.

[0052]

[0053] FIG. 3 is a diagram illustrating a system of the present invention according to one embodiment of the present invention. Referring to FIG. 3, the system (1) may include a user terminal (10), an electronic device (20), and a database (30).

[0054] The user terminal (10) may be implemented as a smartphone or wearable device capable of verifying data generated by the electronic device (20) and metadata analyzed therefrom, but is not necessarily limited thereto and may also be implemented as various devices such as a tablet PC. The user terminal (10) may be connected to the electronic device (20) via a network, and a plurality of user terminals (10) may be connected to the electronic device (20) simultaneously.

[0055] The electronic device (20) may include an artificial intelligence-based customized modular building design and BIM linkage device. The electronic device (20) may be provided by being included in a computer-readable recording medium by tangibly implementing a program of commands for implementing the same. In other words, the electronic device (20) may be implemented in the form of program commands that can be executed through various computer means and may be recorded in a computer-readable recording medium. In addition, the electronic device (20) may be configured as a computer program that sequentially or non-sequentially performs operations of receiving user input, calculating a housing demand index, determining hard and soft specifications, analyzing insufficient information to determine additional demand information, reflecting the user's additional input to sequentially update the housing demand index, hard and soft specifications, and calculating housing space data through a space creation model. The computer program may be stored in a computer-readable recording medium.

[0056] The database (30) may correspond to a storage device that stores various information generated through an operational process of receiving user input, calculating a housing demand index, determining hard and soft specifications, analyzing insufficient information to determine additional demand information, sequentially updating the housing demand index, hard and soft specifications by reflecting the user's additional input, and calculating housing space data through a space generation model. In addition, the database (30) not only stores block and material information, but also includes a structure capable of storing version management tags and update history for each data item. This allows the change history to be recorded whenever a design is modified or material information is changed, and data consistency and traceability can be secured in a collaborative environment among multiple users. For example, by maintaining a version tag and a modification history log, such as "v1.0: concrete / v1.2: insulation added" for a specific wall module, the ability to compare or rollback past states during subsequent quotation or design review stages can be provided.

[0057]

[0058] Fig. 4 is a drawing illustrating the physical configuration of an electronic device (20) according to one embodiment. Referring to Fig. 4, the electronic device (20) may be implemented to include a processor (21), a memory (23), a user input / output unit (25), and a network input / output unit (27).

[0059] The processor (21) may include at least one processor implemented to provide at least some different functions. The processor (21) may control the overall operation of the electronic device (20) and may be electrically connected to the memory (23), the user input / output unit (25), and the network input / output unit (27) to control data flow therebetween. The processor (21) may be implemented as a CPU (Central Processing Unit) of the electronic device (20). According to one embodiment, the processor (21) may include a main processor (e.g., a central processing unit or an application processor) or an auxiliary processor (e.g., a graphics processing unit, a neural processing unit (NPU), an image signal processor, a sensor hub processor, or a communication processor) that can operate independently of or together with the main processor. For example, when the electronic device (20) includes a main processor and an auxiliary processor, the auxiliary processor may be configured to use lower power than the main processor or to be specialized for a given function. The auxiliary processor may be implemented separately from the main processor or as a part thereof. The auxiliary processor may control at least a portion of functions or states associated with at least one of the components of the electronic device (20), for example, on behalf of the main processor while the main processor is in an inactive (e.g., sleep) state, or together with the main processor while the main processor is in an active (e.g., application execution) state. In one embodiment, the auxiliary processor (e.g., image signal processor or communication processor) may be implemented as part of another functionally related component. In one embodiment, the auxiliary processor (e.g., neural network processing unit) may include a hardware structure specialized for processing artificial intelligence models. The artificial intelligence models may be generated through machine learning.This learning can be performed, for example, in the electronic device (20) itself where the artificial intelligence model is executed, or can be performed through a separate server. The learning algorithm can include, for example, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but is not limited to the examples described above. The artificial intelligence model can include a plurality of artificial neural network layers. The artificial neural network can be one of a deep neural network (DNN), a convolutional neural network (CNN), a recurrent neural network (RNN), a restricted Boltzmann machine (RBM), a deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), deep Q-networks, or a combination of two or more of the above, but is not limited to the examples described above. In addition to the hardware structure, the artificial intelligence model can additionally or alternatively include a software structure. Meanwhile, the operation of the electronic device (20) described below can be understood as the operation of the processor (21).

[0060] The memory (23) may include an auxiliary memory device implemented with a non-volatile memory such as an SSD (Solid State Drive) or an HDD (Hard Disk Drive) and used to store all data required for the electronic device (20), and may include a main memory device implemented with a volatile memory such as a RAM (Random Access Memory). In addition, the memory (23) may include a plurality of instructions that direct the operations of the processor (21) to implement the functions provided by the service. At this time, the processor (21) may include a software server that executes the functions provided by the service based on the plurality of instructions stored in the memory (23).

[0061] The user input / output unit (25) may include an environment for receiving user input and an environment for outputting specific information to the user. For example, the user input / output unit (25) may include an input device including an adapter such as a touchpad, a touch screen, a virtual keyboard, or a pointing device, and an output device including an adapter such as a monitor or a touch screen. In one embodiment, the user input / output unit (25) may correspond to a computing device accessed via remote access, in which case the electronic device (20) may function as a server.

[0062] The network input / output unit (27) includes an environment for connecting to an external device or system via a network, and may include an adapter for communication such as a LAN (Local Area Network), MAN (Metropolitan Area Network), WAN (Wide Area Network), and VAN (Value Added Network).

[0063]

[0064] The operations described below can be performed via the processor (21).

[0065] In addition, the operation of the device or artificial intelligence-based customized modular building design and BIM linkage device can be understood as an operation of the processor (21), and conversely, the operation of the processor (21) can be understood as an operation of the artificial intelligence-based customized modular building design and BIM linkage device.

[0066]

[0067] FIG. 5 is a diagram illustrating the operation of an AI-based customized modular building design and BIM linkage device according to one embodiment of the present invention. Referring to FIG. 5 , the AI-based customized modular building design and BIM linkage device may include an agent (40), a model (50), and an LLM model (60). Specifically, the AI-based customized modular building design and BIM linkage device may receive user input (10-1) and perform internal processing procedures through the agent (40).

[0068] User input (10-1) is in the form of sentences or paragraphs in natural language and may include the user's desired housing design conditions or preferences. The agent (40) preprocesses this user input and may be linked to an LLM model (60) to interpret the meaning of the input. The LLM model (60) is a large-scale language model that can be used to understand the context of the input sentence and identify correlations with factors such as openness, privacy, and budget for conversion into housing demand indicators.

[0069] The agent (40) can transmit the results derived from the LLM model (60) to the model (50). The model (50) may include, for example, a specification mapping model (51), a space creation model (53), a pricing model (55), etc., and may play a role in determining hard and soft specifications or generating specific space design data based on input housing demand indicators.

[0070] Additionally, the agent (40) continuously manages interactions with the user, and if input is insufficient or unclear, it can utilize the LLM model (60) to generate appropriate questions and then deliver them back to the user, thereby forming an iterative feedback loop. This is described in detail in Fig. 7.

[0071]

[0072] FIG. 6 is a diagram illustrating the configuration of a model according to one embodiment of the present invention. Referring to FIG. 6, the model (50) may include a specification mapping model (51), a space generation model (53), a pricing model (55), and a regulation review model (57). These models may be implemented using rule-based algorithms, machine learning-based prediction models, deep learning-based artificial intelligence models, large-scale language models (LLMs), etc., but are not limited to any one of them.

[0073] The specification mapping model (51) determines hard and soft specifications based on housing demand indicators derived from users. For example, factors such as floor area, number of floors, window layout, and material preferences can be determined based on quantifiable indicators such as openness and budget weighting.

[0074] The spatial generation model (53) is a module that automatically generates actual spatial configuration data based on established hard and soft specifications. The output is converted into IFC-based BIM data for subsequent use.

[0075] The pricing model (55) can estimate construction costs based on housing space data or BIM data output from the space creation model (53). It calculates estimates based on factors such as total area, materials used, number of floors, and construction difficulty, and can utilize regional unit price lists or real-time raw material price information. This model can also be used to assess the fit between a user's budget and the actual design.

[0076] Additionally, it can automatically aggregate the quantity of materials per block of a designed space based on BIM data and calculate construction costs in real time by applying a unit price database (DB).

[0077] The legal review model (57) can review local legal standards, such as building codes, local ordinances, and urban planning standards, to determine whether the resulting space creation meets them. For example, it reviews building-to-land ratios, floor area ratios, sunlight restrictions, and road diagonal restrictions to determine if violations exist, and, if necessary, provides feedback on the design plan. This model can be linked to an external legal database to update standards in real time.

[0078]

[0079] In addition, the overall flow of the invention is described with reference to Figure 7 below, but specific embodiments may be described with reference to other drawings.

[0080] FIG. 7 is a diagram illustrating the operational sequence of an AI-based customized modular building design and BIM linkage device according to one embodiment of the present invention. Referring to FIG. 7, the AI-based customized modular building design and BIM linkage device can receive relevant user input for generating housing space data (S100). Specifically, the user input is data in the form of sentences or paragraphs expressing the user's requirements for the house he or she wishes to design in natural language, such as "I need a two-story villa where my family can relax on weekends" or "It needs to have three bedrooms, a living room, and a kitchen connected, and the budget should be reasonable." Such user input can be received through a user interface connected to the AI-based customized modular building design and BIM linkage device, such as a chatbot, a voice input interface, or a text-based input window. Such input is internally converted into a structured form through a natural language processing (NLP) module and can then be utilized in the step of applying space requirement analysis criteria and calculating housing requirement indicators (S200). Since user input can include not only explicit requirements but also metaphorical or informal expressions, the device can interpret the user's utterances based on a large language model (LLM), extract semantic expressions, and convert them into a data structure suitable for quantitative analysis. That is, the user input can be embedded in a form that the processor (21) can understand. For example, if the user inputs, "I want a house with good lighting," the device can recognize that sentence as the basis for calculating indicators related to "openness" or "need for lighting" and store it in a form that can be converted into a quantified score at a later stage. Such user input can be acquired in the form of a survey, which will be described in detail below.

[0081]

[0082] Next, the AI-based customized modular building design and BIM integration device can apply space demand analysis criteria to user input to calculate a housing demand index (S200). The housing demand index is an indicator that quantitatively expresses the user's housing preferences and may include items such as openness, storage space ratio, privacy needs, energy efficiency, and budget weighting. Each index can be calculated within a range of 0 to 10 points. Specifically, the values ​​of each index can indicate the following: A higher openness score may indicate expanded open areas such as windows, terraces, and hallways. A higher storage space ratio may indicate increased built-in closets, storage, and modular areas. A higher privacy demand may indicate partitions between rooms and separation of movement routes. A higher energy efficiency score may indicate a preference for passive design (insulation, windows, etc.). A higher budget weighting may indicate a preference for inexpensive finishes and blocks.

[0083] In one embodiment, if a user inputs phrases like "I wish I had a spacious living room" or "I wish I had lots of windows," the AI-based customized modular building design and BIM integration device may recognize these phrases as expressions related to spatial openness and assign relatively high scores to the openness index. Conversely, inputs like "I wish the rooms were well-divided so that children can run around and play" may be utilized to increase the index scores for items related to privacy needs. These housing need indices are inferred based on natural language processing (NLP) results and spatial need analysis criteria. These spatial need analysis criteria may include predefined keyword dictionaries, context-based classification algorithms, or machine learning models (e.g., classifiers, regression models, LLM-based inference models). For example, input sentences may be mapped to a multidimensional feature space based on a dataset that has learned how closely certain words or phrases are associated with certain indices, and scores may be calculated based on this. Furthermore, the calculation of the housing needs index goes beyond simply quantifying the user's subjective sensibilities. It can also serve as a reference value in the hard and soft specification determination (S300) stage. Therefore, the accuracy of quantification at this stage can directly impact the quality of subsequent space creation. If the user's input is insufficient or ambiguous, making it difficult to reliably calculate a specific index, the device can withhold the score for that index or request additional input (S400) to obtain supplementary information. Thus, the S200 stage can be considered a key process for deriving the housing needs index, which serves as the foundational data for housing design.

[0084]

[0085] Next, the AI-based customized modular building design and BIM linkage device can determine hard and soft specifications through a specification mapping model based on housing demand indicators (S300). Here, 'hard specifications' refer to structurally essential physical constraints or requirements in housing design, and may include, for example, building use, building scale (total area, number of floors, etc.), budget range, land information (location, area, topographical characteristics, etc.), and space requirements (number of rooms, number of bathrooms, number of parking spaces, etc.). On the other hand, 'soft specifications' refer to relatively qualitative requirements related to user preferences or design directions, and may include, for example, the proportion of openness / privacy, storage space preferences, kitchen / living room connection methods, lighting / view requirements, and preferences for exterior materials and finishing materials. Such hard and soft specifications are expressed as examples, but are not limited thereto.

[0086] Here, the specification mapping model (51) receives as input each housing requirement index (e.g., 8 points for openness, 6 points for privacy, etc.) calculated in step S200, and determines appropriate hard and soft specifications according to pre-learned mapping rules or algorithms. For example, if the 'openness' score is high, the floor area of ​​the hard specification can be increased to expand the connected area of ​​the living room and external space (terrace, balcony, etc.), and the window arrangement density and open structure can be selected as the soft specification. If the 'budget weight' is evaluated high, a mid-low budget range can be set as the hard specification, and relatively low-priced interior and exterior materials can be selected as the soft specification.

[0087] In one embodiment, the specification mapping model (51) may be implemented using a rule-based mapping table, a regression model, a decision tree-based classifier, or a deep learning-based artificial neural network (e.g., MLP, GNN, etc.). In particular, by supplementally utilizing an LLM-based inferencer, non-standard user needs can be reflected in the specification based on inference in addition to explicit rules.

[0088] Additionally, some hard specifications (e.g., land location, building use, etc.) are reflected as-is if explicitly included in the user's input, while the remaining specifications are determined based on their correlation with housing demand indicators and inter-constraints (e.g., legal restrictions, floor area limits, etc.). Soft specifications prioritize design elements by reflecting the user's emotional needs, and can then be implemented as specific architectural designs in the space generation model (S700).

[0089] Thus, the S300 stage serves as a process for generating key input values ​​for automated design by concretizing the user's subjective needs into physical / design elements. Since this process directly impacts the quality of residential space design and user satisfaction, the accuracy and flexibility of specification mapping can be key factors in system performance.

[0090]

[0091] Next, if the housing demand index cannot be calculated or if there are elements among the hard specifications that cannot be calculated based on the housing demand index alone, the AI-based customized modular building design and BIM linkage device can analyze the missing information to determine the necessary additional demand information (S400). For example, if the user inputs only vague expressions such as "I want a house where I can feel a lot of nature" and there is no direct mention of the budget, the device will have difficulty reliably calculating the "budget weight" index. In this case, the AI-based customized modular building design and BIM linkage device can detect the impossibility of calculating the relevant index and automatically derive the necessary additional demand information (e.g., budget range, construction quality preference, etc.) based on the user's response history and missing items. In other words, if some of the housing demand indexes cannot be calculated based on the user's input alone, the processor determines this and requests additional input from the user, enabling the relevant index to be calculated.

[0092] Even if housing demand indicators are calculated correctly, there may be cases where specific hard specifications (e.g., number of parking spaces, site location, etc.) cannot be sufficiently identified with these indicators alone. In such cases, the device can comprehensively analyze the user's input and previously calculated specifications to identify items requiring further verification and determine methods (e.g., in the form of questions) for obtaining additional information. This analysis can go beyond simply identifying missing items and can be performed using machine learning-based decision confidence analysis, uncertainty estimation models, or LLM-based context-completion inference algorithms. For example, if the reliability of the estimated results for a specific indicator in the user's natural language input falls below a certain threshold, the item can be judged as incomplete data and classified as additional demand information. For example, if the user inputs, "I would like a spacious kitchen," and the "openness" indicator is calculated, but hard specifications such as "total floor area" or "number of parking spaces" remain unidentifiable, the device can then ask questions such as, "What is your budget?" or "How many family members do you have?" You can gather information to determine hard specifications by asking the user supplementary questions such as:

[0093] More specifically, the LLM model (60) can include a feedback loop structure that, in addition to simply interpreting input to derive a housing needs index, determines the incompleteness or ambiguity of the input and dynamically reconfigures prompts accordingly, or generates and presents appropriate additional questions to the user. Furthermore, it can be linked to external workflows (e.g., quotation systems, design proposal modules, construction review engines, etc.) in a triggered manner based on the user's input or response results. This structure can also be implemented using a flow-based LLM orchestration structure similar to LangChain or n8n. Through this, the user interface can be expanded beyond simple question-and-answer to a multi-step workflow-based design automation system.

[0094] In one embodiment, the device may configure a specific list of questions or context-based question scenarios to request additional information from the user in the next step (S500), including an additional request information determination model or LLM model. These questions may be presented to the user through a chatbot-based interface or a GUI-based survey screen, and the presentation method may vary depending on system settings or the user interface.

[0095] Step S400 defines the input conditions for Step S500 and plays a key role in ensuring data integrity throughout the entire design flow and improving the quality of user responses. Effective operation of this step ensures uninterrupted operation of the entire design process, resulting in highly reliable automated design results.

[0096]

[0097] Next, the AI-based customized modular building design and BIM integration device can request the user for additional information to obtain the required information (S500). The request for additional information can be presented in the form of context-based questions to facilitate user understanding and elicit responses. For example, if budget information is determined to be insufficient at step S400, the device can pose questions such as, "Do you have a budget limit?" or "Is cost reduction or quality more important to you?" The questions can be automatically generated using predefined survey questions, condition-based branching questions, or an LLM-based chatbot question generator.

[0098] In one embodiment, an AI-based customized modular building design and BIM integration device presents questions in natural language through a chatbot-based user interface. It also receives user responses in natural language and interprets their meaning, thereby supplementing insufficient indicators or specifications. For example, a response like "I'm thinking about around 200 million won" can be used to calculate a budget weighting indicator and a budget range for hard specifications.

[0099] Furthermore, requests for additional information can go beyond simple data acquisition and serve as a feedback loop, further clarifying user needs and leading to personalized design outcomes. For example, if the question, "How important do you consider privacy?" yields the answer, "I want my children to have separate rooms," the device can reflect this in quantitative updates to its privacy needs index, while also influencing soft specifications like room layout and partition configuration.

[0100] The timing and frequency of requests for additional information can be automatically adjusted by the system. For example, questions can be minimized when the response is highly reliable, while supplementary questions can be presented sequentially when there is potential for ambiguity or contradiction. Such adaptive questioning strategies can contribute to enhancing the completeness of the design without disrupting the user experience.

[0101] Step S500 is a key step that encourages active user participation, enhancing the degree of customization of the automated design system. It directly links to the subsequent step (S600) of updating indicators and specifications. This enables the design automation system to move beyond simple recommendations to perform advanced, interactive design coordination functions.

[0102]

[0103] Next, the AI-based customized modular building design and BIM integration device can sequentially update housing demand indicators, hard specifications, and soft specifications by reflecting additional user input (S600). First, the user's additional input is received in natural language, then preprocessed and interpreted into semantic units. The content is then analyzed based on which housing demand indicator item the input is associated with. For example, if the user additionally responds, "I wish my electricity bill was lower," the device recognizes this as an input corresponding to the "energy efficiency" item and can adjust the existing calculated indicator value (e.g., 6 points) upward or improve its reliability.

[0104] In this way, additional inputs are first used to recalculate the housing demand index, and the updated index is then sequentially reflected in the re-determination of hard and soft specifications. For hard specifications, items such as budget range, total floor area, number of floors, and number of parking spaces can be subject to update, while soft specifications can be applied to items such as openness, level of privacy, and material preferences.

[0105] Additionally, the housing demand index, hard specifications, and soft specifications can be updated sequentially. That is, the housing demand index can be updated first, followed by the hard specifications, and then the soft specifications. For example, the housing demand index can be updated, followed by the hard specifications and soft specifications.

[0106] In one embodiment, the update process is implemented as an iterative feedback loop, allowing a single additional input to simultaneously impact multiple metrics. For example, the input "I need to reduce costs but have more rooms" could affect both budget weighting and space efficiency, resulting in hard specifications being adjusted to prioritize smaller module combinations.

[0107] When updating specifications, the device can calculate the difference and impact from existing results to determine whether to update and adjust the intensity of the update. For example, if the difference from the existing metric score is minimal or a conflict with the existing specification occurs, the update can be implemented conservatively or by requesting user confirmation.

[0108] These S600 steps play a key role in ensuring the completeness of the final space design by complementing missing or inaccurate requirements through user interaction. Furthermore, this step can be accurately reflected in subsequent space creation models, ultimately leading to more precise and user-friendly design results.

[0109]

[0110] Next, the AI-based customized modular building design and BIM linkage device can generate housing space data through a space generation model (53) based on the updated hard and soft specifications (S700). The space generation model (53) is a core module that automatically designs the spatial configuration, floor plan layout, elevation shape, and inter-space connection structure of the house to comply with the above specifications, and corresponds to the step of deriving actual design results. Here, the space generation model (53) can perform space layout design by referencing a modular block library. The modular block library includes information on the specifications, purpose, material, unit price, and edge conditions of each block, and the space generation model (53) can generate the optimal block combination of the housing space configuration based on this information. Through this, an actual design plan can be automatically generated in a direction that satisfies space efficiency, ease of construction, and compliance with regulations.

[0111] Here, edge information is the core metadata that defines the connectability of each modular block. It specifies whether the left, right, top, and bottom edges of the block are 'open (true),' 'blocked (false),' or 'conditionally connectable (both),' respectively. For example, if the right edge of a specific block is 'true,' the left edge of the adjacent block must also be 'true' to enable a two-way connection. In this way, edge conditions serve as a criterion for judging the geometric and functional compatibility between blocks, and the space generation model determines the optimal layout by filtering combinable blocks or calculating a combination score based on this information. In particular, edge conditions play a key role in practical and constructible space design, as they are linked not only to simple physical contact but also to various design elements such as lighting, ventilation, and circulation.

[0112] The space generation model (53) can be implemented, for example, as a combination algorithm of architectural module units, a graph-based space layout engine, or a deep learning-based space generation network. At this time, the derivation of initial candidate blocks or the determination of placement priorities can be performed based on AI-based score evaluation or graph-like structure analysis, but the final space layout process can be implemented in a rule-based manner according to the edge conditions and combination rules of the modular blocks. In one embodiment, the space generation model (53) is configured to be linked with a modular architectural block library so that the total area, number of floors, and architectural use specified in the hard specifications can be reflected in the design layout, while also reflecting emotional factors such as openness, lighting, and privacy reflected in the soft specifications.

[0113] For example, if the hard specifications are 'total area of ​​120㎡, 3 bedrooms, 2 bathrooms, 2-story structure' and the soft specifications are 'openness score of 8 points, privacy score of 6 points', the space creation model (53) can derive a space configuration that arranges the living room into a structure that includes large windows and a terrace connected to the outside, and secures an appropriate sense of separation by adjusting the distance or wall configuration between bedrooms.

[0114] The space generation model (53) can also consider functional connectivity and traffic flow efficiency between spaces. It can also be implemented by generating multiple design proposals and then selecting the optimal one through multi-criteria evaluation or presenting multiple candidate proposals to the user. In some embodiments, pre-filtering can be performed using reinforcement learning-based reward functions or rule-based rules to ensure that spatial layouts do not conflict with building standards, sunlight rights, building-to-land ratios, and other requirements.

[0115] In addition, the output of the space creation model (53) can be provided in the form of structured data, and for example, can be converted into BIM data in the IFC (Industry Foundation Classes) format and linked to subsequent processing (S800, etc.). This BIM data automatically includes the area of ​​each space, wall configuration, window location, module ID, etc., and thereby supports architects or construction workers to perform actual design review and cost analysis. Specifically, after the output of the space creation model (53), it can be linked to the block unit price DB through the pricing model to calculate the quantity and estimate in real time.

[0116] In an embodiment, the space generation model (53) of the present invention can be applied to the design of various types of modular buildings, such as single-family homes, modular office or retail buildings of 3-4 stories, and low- to mid-rise apartments of 10 stories or less. Depending on each of these types, the space configuration algorithm can automatically perform design by reflecting conditions such as floor-count restrictions, stacking methods between structural modules, and securing vertical movement lines.

[0117] Furthermore, the space creation results are linked to BIM attribute data defined at the block level. Each block has a structured attribute table containing information such as material type (e.g., concrete, wood, insulation, etc.), quantity information (e.g., wall area, number of windows), and unit cost information (e.g., construction cost per pyeong, module unit price). This allows the system to provide quantified cost estimates simultaneously with space design, which can be immediately utilized for subsequent quotation calculations and construction planning.

[0118] Meanwhile, the output of the space creation model (53) can be converted into BIM data in IFC format and then exported in a structure suitable for a commercial BIM software environment, and for example, can be linked with the plug-in API of Autodesk Revit to automatically create and export an IFC file. The IFC file thus created can be automatically transferred to a commercial estimation system for real-time material quantity, construction unit price, budget aggregation, etc. through linkage with an ERP system or an external estimation analysis system. According to this embodiment, a series of flows from space design to BIM creation, ERP linkage, and estimation generation are automated based on API, thereby maximizing the efficiency of design-construction linkage.

[0119] The S700 stage is the central process for generating practical design proposals throughout the system. It automatically generates spatial data that integrates user needs and practical constraints, making it a key factor in determining the completeness of design automation. This allows users to easily create customized housing spaces tailored to their individual needs, without requiring specialized expertise.

[0120]

[0121] FIG. 8 is a diagram illustrating a process for conducting a user survey according to one embodiment of the present invention. Referring to FIG. 8, an AI-based customized modular building design and BIM linkage device can obtain user input in the form of a survey. In this process, the user can interface with an agent (40) via a user terminal (10), and the agent (40) provides a survey form to the user (11) based on the user's initial input or request. The survey can be structured in a templated question-based format, and can include multiple-choice or descriptive questions such as, for example, "What is the most important residential element?", "How many floors do you prefer in a building?", and "Please tell me your budget range." In addition, score-based items for quantitatively understanding the user's subjective preferences can also be included. Specifically, for the question “How open would you like to be?” users can choose from a scale of 1 (closed) to 5 (very open), and for the question “How important is privacy or separation to you?” users can choose from a scale of 3 (not important) to 5 (very important).

[0122] These score-based questions are advantageous for quantifying users' emotional needs and organizing input data into a format that can be immediately utilized in the subsequent housing needs index calculation (S200) step. Furthermore, user response forms can be implemented in various UI formats, such as radio buttons, slider bars, and star ratings, allowing for the collection of precise data without compromising the user experience.

[0123] Survey questions can be in a predefined format, but can also be provided as a customized set of questions tailored to the user's input context, utilizing the LLM model (60). For example, if a user inputs "I really want a terrace," the LLM model (60) can generate an additional survey including a rating item related to openness and provide it to the user via the agent (40).

[0124] These various forms of survey items can serve as basic data for precisely understanding housing needs and reflecting them in design results.

[0125] Here, the agent (40) can request the generation of survey-related information from the LLM model (60) to provide a survey, and the LLM model (60) can reference a pre-learned context-based question set or dynamically generate question items based on the context of user input. Such survey configuration information can be generated in real time by reflecting updated design conditions or criteria from the database (30), and the LLM model (60) can be linked to the database (30) when necessary to check and reflect the latest information.

[0126] Once a user completes a survey, the response is collected by the agent (40), and the need for additional information is determined through the LLM model (60). If the determination reveals that required information is missing or ambiguous, the agent (40) may send a request for additional input to the user (11), and the user may respond by providing supplementary input.

[0127] Through this procedure, the user survey process of Fig. 8 serves as an initial means for clarifying and quantifying user requirements, and can collect basic information that can be utilized in subsequent stages such as calculating housing needs indicators (S200), determining specifications (S300), and designing space (S700).

[0128]

[0129] Figure 9 is a diagram illustrating a process for creating a space according to one embodiment of the present invention. Referring to Figure 9, an AI-based customized modular building design and BIM integration device calculates housing demand indicators based on user input, determines specifications based on these indicators, and then performs a series of procedures for generating spatial data.

[0130] First, a user (11) inputs housing-related requirements in natural language, and an agent (40) can transmit the input to an LLM model (60) to request the calculation of a housing demand index. The LLM model (60) can analyze the context of the input to calculate quantitative housing demand indices such as openness, privacy, and budget weighting, and return the results to the agent (40).

[0131] Next, the agent (40) performs a specification determination request based on the generated housing demand index, and the model (50) can utilize the specification mapping model to determine hard specifications (e.g., total area, number of floors, number of parking spaces) and soft specifications (e.g., openness, lighting requirements, etc.) and return the corresponding information.

[0132] In the subsequent process, the agent (40) collaborates with the LLM model (60) to determine whether any items currently lack information, and, if necessary, may request additional input from the user (11). If the user inputs supplementary information, the agent (40) reflects this information and updates the housing demand indicators and specifications, respectively, and finally requests the model (50) to generate spatial data based on the completed indicators and specifications.

[0133] The model (50) can utilize the space creation model to design the spatial layout, structural configuration, and traffic flow of a house according to updated specifications, and provide the resulting spatial data to the user (11) via the agent (40). This data can then be converted into a BIM format (IFC, etc.) and utilized for collaboration with architects or construction companies.

[0134]

[0135] Although the present invention has been described above with reference to preferred embodiments thereof, it will be understood by those skilled in the art that various modifications and changes may be made to the present invention without departing from the spirit and scope of the present invention as set forth in the claims below.

Claims

1. As an electronic device, a memory containing at least one instruction; and At least one processor electrically connected to said memory and configured to perform said at least one instruction; At least one processor, Receiving relevant user input for generating housing space data, wherein said user input is in natural language; Applying the space demand analysis criteria to the above user input to produce a housing demand index, Based on the above housing demand indicators, hard and soft specifications are determined through a specification mapping model, If the above housing demand index cannot be calculated or if there are elements among the above hard specifications that cannot be calculated with the above housing demand index alone, the missing information is analyzed to determine the necessary additional demand information. Request the user to provide the information necessary to obtain the above additional required information, Sequentially update the housing demand indicators, hard specifications, and soft specifications by reflecting the additional inputs from the above users. Generates housing space data through a space generation model based on updated hard and soft specifications. Electronic devices.

2. In paragraph 1, The above housing demand indicators are: Featuring openness, storage ratio, privacy needs, energy efficiency and budget weighting. Electronic devices.

3. In paragraph 2, At least one processor, The housing demand index is calculated in the range of 0 to 10 points, and hard specifications and soft specifications are mapped according to the value of the housing demand index. Electronic devices.

4. In paragraph 3, At least one processor, If one or more of the housing demand indicators cannot be derived solely from the user inputs above, characterized in that it requests additional input from the user to produce the indicator; Electronic devices.

5. In paragraph 4, At least one processor, If the hard specifications cannot be determined based on the user input and housing demand indicators above, characterized in that it requests additional input from the user to produce the specifications; Electronic devices.

6. In paragraph 5, At least one processor, Refer to the building standards that include regulatory information in the hard specifications, Characterized in that the spatial layout is adjusted by applying the building coverage ratio, floor area ratio, floor limit, and sunlight right criteria to the housing space data according to the above-mentioned building standards. Electronic devices.

7. In paragraph 6, The above electronic device, Includes more modular block libraries, The above modular block library is, Includes information on the specifications, use, material, unit price, and edge conditions of each block. The above space creation model is, Generates house space data by calculating block combinations for space configuration based on the above modular block library. Electronic devices.

8. In paragraph 7, The above space creation model is, Characterized by converting the produced housing space data into BIM data in the Industry Foundation Classes (IFC) format. Electronic devices.

9. In paragraph 8, At least one processor, Characterized in that it provides a construction cost estimate based on the above BIM data. Electronic devices.

10. In paragraph 9, It is characterized by further including a pricing model that calculates the construction cost in real time by aggregating the quantity per block based on the BIM data and applying unit price information. Electronic devices.

11. By means of a memory and at least one processor electrically connected to the memory, A step of receiving user input relevant to generating housing space data, wherein the user input is composed of natural language; A step of calculating a housing demand index by applying a space demand analysis criterion to the above user input; A step of determining hard specifications and soft specifications through a specification mapping model based on the above housing demand indicators; A step of analyzing the insufficient information to determine the necessary additional demand information when the housing demand index cannot be calculated or when there is an element among the hard specifications that cannot be calculated using only the housing demand index; A step of requesting the user for information necessary to obtain the above additional required information; A step of sequentially updating the housing demand index, hard specifications, and soft specifications by reflecting the additional input of the above user; and Including a step of generating housing space data through a space generation model based on updated hard and soft specifications. method.

12. In paragraph 11, The above housing demand indicators are: Featuring openness, storage ratio, privacy needs, energy efficiency and budget weighting. method.

13. In paragraph 12, The housing demand index is calculated in the range of 0 to 10 points, and hard specifications and soft specifications are mapped according to the value of the housing demand index. method.

14. In paragraph 13, If one or more of the housing demand indicators cannot be derived solely from the user inputs above, characterized in that it requests additional input from the user to produce the indicator; method.

Images