Urban road disease automatic processing method and device based on large model intelligent agent

Abstract

The invention provides an urban road disease automatic processing method and device based on a large model agent, and the method comprises the steps: obtaining disease report information, adding the disease report information to an existing disease file or building a new disease file through a super agent, and enabling each disease file to correspond to a disease and an autonomous agent; performing grading processing on diseases in the new disease file through an autonomous agent to obtain disease grades; when the disease level is that the disease needs to be treated, generating a disease work order; based on the plurality of work order distribution factors, sending the disease work order to the target object through the super agent; the disease maintenance information of the disease work order is obtained through the autonomous agent, the maintenance effect is evaluated based on the disease maintenance information, the maintenance effect evaluation result is obtained, and the service efficiency and the treatment capacity are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of AI technology, and in particular to an automated method, device, electronic device and storage medium for handling urban road defects based on a large model intelligent agent. Background Technology

[0002] With the rapid advancement of urbanization and the continuous expansion of modern transportation systems, the demand for routine maintenance and emergency repairs of urban roads and bridges, among other municipal infrastructure, is increasing dramatically. Urban road defects (such as potholes, cracks, and abnormal settlement) have increased significantly, affecting not only residents' travel safety but also the city's image and governance evaluation. Therefore, road inspection, defect identification, treatment decisions, and repair transfer have become core tasks of urban management. However, currently, the vast majority of urban road maintenance processes rely on manual labor, facing a series of intractable bottlenecks. Summary of the Invention

[0003] This application provides an automated processing method, apparatus, electronic device, and storage medium for urban road defects based on a large-model intelligent agent, aiming to at least partially solve one of the technical problems in related technologies. The technical solution disclosed herein is as follows: In a first aspect, embodiments of this application propose an automated method for handling urban road defects based on a large-scale intelligent agent, applied to an automated urban road maintenance management system. The system includes a super intelligent agent and multiple autonomous intelligent agents, characterized by the following steps: Obtain disease reporting information, which includes disease scene information, location information, and disease identification results; The super intelligent agent adds the disease reporting information to the existing disease file or creates a new disease file. Each disease file corresponds to one disease and one autonomous intelligent agent. The autonomous intelligent agent performs classification processing on the diseases in the new disease files to obtain disease levels; and when the disease level is required to be treated, a disease work order is generated. Based on multiple work order allocation factors, the super intelligent agent sends the defect work order to the target object. The autonomous intelligent agent obtains the defect repair information of the defect work order, evaluates the repair effect based on the defect repair information, and obtains the repair effect evaluation result. Secondly, embodiments of this application propose an automated urban road damage treatment device based on a large-scale intelligent agent. This device is configured within an automated urban road maintenance management system, which includes a super intelligent agent and multiple autonomous intelligent agents. The device comprises: The information reporting module is used to obtain disease reporting information, which includes disease scene information, location information, and disease identification results. The archiving module is used to add the disease reporting information to existing disease files or create new disease files through the super intelligent agent. Each disease file corresponds to one disease and one autonomous intelligent agent. The work order processing module is used to classify the diseases in the new disease files through the autonomous intelligent agent to obtain the disease level; and generate a disease work order when the disease level is required to be treated. The work order dispatch module is used to send the defect work order to the target object through the super intelligent agent based on multiple work order allocation factors. The work order review module is used to obtain the defect repair information of the defect work order through the autonomous intelligent agent, evaluate the repair effect based on the defect repair information, and obtain the repair effect evaluation result. Thirdly, embodiments of this application provide an electronic device, including: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method described in the first aspect.

[0004] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the method described in the first aspect.

[0005] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the method described in the first aspect.

[0006] The method, device, electronic equipment, and storage medium for automated processing of urban road defects based on large-model intelligent agents provided in this application, with a collaborative system consisting of a super intelligent agent driven by a large model and an autonomous intelligent agent, realizes the automated creation of intelligent archives for the entire life cycle of defects and the individualized maintenance mechanism of autonomous intelligent agents. It realizes closed-loop management of the entire life cycle of defects, from discovery, classification, dispatch, repair, review to long-term monitoring and data analysis. The system is highly automated and intelligent, and fully integrates multimodal data, dialogue collaboration, and business process visualization, significantly improving business efficiency and governance capabilities.

[0007] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0008] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart illustrating an automated urban road damage processing method based on a large model intelligent agent, provided in an embodiment of this application; Figure 2 A block diagram of an automated urban road damage processing device based on a large model intelligent agent provided in an embodiment of this application; Figure 3 This is a block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0009] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0010] Terminology Explanation: ChatGLM: A large language model.

[0011] GPT4: A large language model.

[0012] Qwen: A large language model.

[0013] LoRA: Low-Rank Adaptation of Large Language Models, is a method for fine-tuning large language models.

[0014] Work orders for defects: documents related to structural or surface damage to urban infrastructure such as roads and bridges, and the corresponding repair and handling procedures.

[0015] Intelligent agent: An AI agent driven by a large model, possessing autonomous analysis, perception, reasoning, and decision-making capabilities.

[0016] Super agent: A high-level intelligent agent responsible for global decision-making, resource scheduling, comprehensive analysis and optimization of complex problems.

[0017] Autonomous intelligent agents: Low- to mid-level intelligent agents responsible for specific atomized tasks, which can be invoked as needed by superintelligent agents.

[0018] Disease lifecycle management: Digital monitoring and analysis management of the entire process and lifecycle of urban road diseases, from discovery, classification, dispatch, repair to long-term follow-up and re-inspection.

[0019] Multimodal large models: Deep learning large models capable of processing multiple data types such as text, images, speech, and even spatiotemporal data.

[0020] End-side deduplication: Quickly identify and filter duplicates at the source of disease data (such as intelligent inspection vehicles).

[0021] GIS map: A geographic information system map used to visualize the distribution and status of road defects.

[0022] Platform: refers to system architecture.

[0023] In related technologies, the vast majority of urban road maintenance processes rely on manual labor, facing a series of intractable bottlenecks: Imbalance between data volume and processing capacity: New generation of intelligent patrol vehicles, road sensors and other equipment can automatically collect massive amounts of disease data at high frequency, but the ability of manual identification, input, deduplication and task assignment is limited, resulting in a large amount of data being delayed or lost due to manual screening, causing the actual situation to be disconnected from the database, making it difficult to give full play to the value of intelligent hardware.

[0024] Process fragmentation and information silos: Disease discovery, registration, dispatch, tracking, and feedback are usually scattered across different systems (such as WeChat groups, DingTalk, stand-alone spreadsheets, work order apps, etc.) and people, lacking a unified data link, resulting in a large number of cases of being out of control or missed, and making it almost impossible to trace the history of diseases.

[0025] Inconsistent manual standards and loss of experience: Disease deduplication, classification, plan formulation, review and acceptance, etc., rely heavily on the individual experience of dispatchers and inspectors. The lack of quantitative standards leads to inconsistent standards, arbitrary judgments, poor error correction and risk management capabilities, and difficulty in accumulating experience into organizational assets.

[0026] Lack of closed-loop management: Focusing on work orders as the main line, paying attention to "the current problem has been fixed" while ignoring the long-term evolution and recurrence of the same diseased section, it is impossible to achieve closed-loop management from discovery to archiving and even to the next periodic inspection.

[0027] Insufficient intelligent analysis and management: Data is mainly used for transaction flow, and in-depth value mining (such as trend analysis, graph query, automated reporting, performance analysis, etc.) is lacking. Managers passively summarize and make decisions after the fact, making it difficult to achieve full-process data-driven decision-making.

[0028] Limitations of human-computer interaction and multimodal applications: Information flow and decision-making heavily rely on manual tracking via mobile phones and PCs, lacking intelligent human-computer collaboration capabilities such as semantic interaction, multi-turn dialogue, and automatic language supervision and explanation.

[0029] Furthermore, intelligent inspection, data entry, and work order processing rely heavily on manual labor, making it difficult to handle large volumes of data and cope with heterogeneous processes and team resource limitations. Business processes are fragmented, and there is a lack of full lifecycle management for diseases from discovery to treatment, review, and even recurrence. Disease information, historical data, and knowledge and experience are difficult to accumulate. The standards for disease judgment, deduplication, classification, work order dispatch, process monitoring, and result review are chaotic, time-consuming, and prone to subjective errors. Data silos and inefficient collaboration between systems mean that important processes are mostly handled via WeChat / DingTalk / offline, resulting in poor overall perception of business processes and decision-making, and slow response. The lack of intelligent support such as work order / disease attribution analysis, automatic data report generation, and leadership dashboards hinders the upgrading of operation and maintenance efficiency and smart governance.

[0030] Some leading domestic and international companies and urban management departments are attempting to improve urban maintenance by adopting smart work order systems and automated intelligent patrol vehicle solutions. For example, integrating a work order system with automated patrol vehicles involves the patrol vehicles equipped with cameras and sensors to automatically identify and collect road surface defects, uploading images, location data, etc., to the cloud or management platform via mobile internet. Manual or simple rule engines then perform initial screening of the defect data in the background, assigning it to the work order system and pushing it to relevant personnel for repair. Repair personnel upload photos and progress updates via an app, while managers view the work order status in real time through the platform. AI-based defect identification is being introduced, using AI image recognition models (such as CNN) to identify the type and severity of defects in uploaded images and assisting in generating automatic work order suggestions. Some platforms can also achieve batch work order review capabilities. The main implementation process is: patrol vehicle data collection → AI model image recognition → automatic generation of defect work orders → app push of repair information → manual scanning / image verification and archiving of repair results. The main limitations and shortcomings of this solution include: Limited automation: The intelligence is limited to the initial identification of disease images and the generation of work orders. It is difficult to achieve full automation or high-level intelligence in disease deduplication from multiple sources, intelligent classification and full-process tracking of difficult cases, and work order merging / scheduling.

[0031] The system cannot form an archive-style management system that tracks the entire life cycle of a disease: it still uses work orders as the unit of circulation, ignoring the multiple recurrences, state evolution, and full-cycle analysis of the same disease, making it difficult to close the loop in decision-making and knowledge accumulation.

[0032] Lack of intelligent agent decision-making and autonomous reasoning capabilities in complex scenarios: tasks such as "packaging and dispatching", "scheme optimization", "confidence control" and "boundary anomaly scenarios" require manual review, and it is unable to perform global optimal scheduling and closed-loop management of complex business flows involving both humans and machines.

[0033] Lack of multimodal and multi-turn human-computer dialogue: It only processes static images and structured data, and the communication between users and the system mainly relies on traditional APP forms, which cannot support multimodal input (natural language + images + map data) and flexible multi-turn dialogue decision-making.

[0034] Difficulty in automatically generating comprehensive analysis reports and visual situational awareness: Reports mostly rely on manual statistics in the background, lacking the ability to automatically generate structured disease maps, statistical analysis, and intelligent PPT / reports, resulting in low management efficiency.

[0035] Untraceability and weak data assetization: Most systems are single-function closed loops, making it difficult to connect disease data to historical records, resulting in low knowledge accumulation and reuse capabilities, and serious loss of experience.

[0036] The aforementioned problems have severely hampered the city's ability to efficiently treat and meticulously maintain road damage, becoming one of the core obstacles restricting the modernization and upgrading of smart city management and maintenance.

[0037] To address the above issues, this application provides an automated urban road defect processing method and platform based on a large-model intelligent agent. The method is applied to the platform, whose core is a collaborative system of a super intelligent agent driven by a large model and an autonomous intelligent agent. This system enables closed-loop management of the entire road defect lifecycle, from defect discovery, classification, dispatching, repair, review to long-term monitoring and data analysis. The system is highly automated and intelligent, and fully integrates multimodal data, dialogue collaboration, and business process visualization, significantly improving business efficiency and governance capabilities.

[0038] The following description, with reference to the accompanying drawings, outlines an automated method, apparatus, and device for processing urban road defects based on a large-scale intelligent agent, representing embodiments of this application.

[0039] Figure 1 This is a flowchart illustrating an automated method for handling urban road defects based on a large model intelligent agent, as provided in an embodiment of this application.

[0040] It should be noted that the executing entity of the automated urban road damage processing method based on large model intelligent agents in this application embodiment is the automated urban road damage processing device based on large model intelligent agents in this application embodiment. The automated urban road damage processing device based on large model intelligent agents can be configured in an electronic device so that the electronic device can perform the automated urban road damage processing function based on large model intelligent agents.

[0041] like Figure 1 As shown, the automated urban road damage treatment method based on a large model intelligent agent includes the following steps: Step S101: Obtain disease reporting information, which includes disease scene information, location information, and disease identification results.

[0042] The automated urban road defect treatment method of this embodiment is applied to an automated urban road maintenance management system, which includes a super intelligent agent and multiple autonomous intelligent agents.

[0043] In some embodiments, the super agent and multiple autonomous agents may select different large models (e.g., ChatGLM, Qwen, GPT-4, etc.), or adopt proprietary large models or open-source multimodal models.

[0044] In some embodiments, the method for obtaining disease reporting information includes: collecting relevant information about the inspection object, the relevant information of the inspection object including scene information and location information presented through video or pictures; identifying diseases based on the scene information using an edge-side multimodal AI model to obtain disease identification results, the disease identification results including disease type and severity.

[0045] In one example, during the daily operation of urban roads, intelligent patrol vehicles are equipped with edge-side multimodal AI models to conduct 24-hour dynamic patrols of the roads. The onboard system collects road surface images, videos, and location data in real time. Through the edge-side multimodal AI model, various road surface defects such as cracks and potholes are automatically identified, and the type and severity of the defects are preliminarily determined. The identification results, along with the original images, videos, and geographical location information, are uploaded to the backend business system through the intelligent maintenance platform API.

[0046] In some embodiments, obtaining disease reporting information includes: obtaining disease reporting information uploaded by manual inspections and third-party repair channels.

[0047] In one example, the backend business system synchronously receives disease reports from other reporting channels (such as manual inspections, third-party repair requests, etc.) and standardizes all disease reports for storage.

[0048] Step S102: Add the disease reporting information to the existing disease file or create a new disease file through the super intelligent agent. Each disease file corresponds to one disease and one autonomous intelligent agent.

[0049] In some embodiments, the method of adding disease reporting information to an existing disease file or a new disease file includes: performing a full comparison of the first disease in the disease reporting information with all diseases in the disease file database using multiple matching algorithms; if the first disease corresponding to the disease reporting information is consistent with the second disease in the disease file database, adding information to the disease file corresponding to the second disease based on the disease reporting information; if the first disease in the disease reporting information is inconsistent with all diseases in the disease file database, creating a new disease file for the first disease based on the disease reporting information.

[0050] In some embodiments, the various matching algorithms may include spatiotemporal clustering, image feature matching, and keyword semantic reasoning.

[0051] In some embodiments, before performing a full comparison between the first disease in the disease report information and all diseases in the disease archive, the process includes: verifying the disease report information based on disease scenario information; and after the verification is passed, standardizing the relevant information of the first disease corresponding to the disease report information.

[0052] In one example, the reported road damage information is suspected damage data and requires further verification. After the suspected damage data is uploaded to the backend business system, the supercomputer will automatically perform secondary confirmation of the damage on the backend. For example, based on a multimodal model of text and image information, the supercomputer will determine whether it is a road damage and automatically verify and complete the damage information, such as location information, damage type, size and other key information. If necessary, it will also go back to the historical damage archive for comparison and correction to achieve standardization of damage information. After the disease is accurately identified and the information is standardized, the system platform enters the "global disease deduplication and file collection" stage. The super intelligent agent uses multiple algorithms such as spatiotemporal clustering, image feature matching, and keyword semantic reasoning to perform a full comparison between the reported disease and all registered diseases in the system. If the same or highly overlapping diseases are found, the reported information is automatically merged into the disease file with the unique ID of the corresponding disease, and new discoveries and development status are supplemented in the disease file for the entire life cycle of the disease. If it is determined to be a new disease, a new disease file with a unique ID is automatically created. The autonomous intelligent agent is responsible for the full maintenance of the disease file. All subsequent discoveries, treatments and monitoring are included in the disease file, and no longer based on a single temporary work order.

[0053] Step S103: The autonomous intelligent agent performs classification processing on the diseases in the new disease files to obtain the disease level; and when the disease level is that it needs to be treated, a disease work order is generated.

[0054] In some embodiments, a method for classifying diseases in a new disease file to obtain disease levels, and generating a disease work order when the disease level requires treatment, includes: classifying diseases in a new disease file based on relevant information of the diseases in the new disease file to obtain disease levels, and adding the disease levels to the new disease file; generating a maintenance plan when the disease level requires treatment; obtaining recommended maintenance solutions, and generating a disease work order based on the maintenance plan and the recommended maintenance solutions.

[0055] In some embodiments, when the disease level is under continuous monitoring, a time-triggered mechanism is used to conduct subsequent inspections and periodic follow-ups on the first disease.

[0056] In one example, the autonomous AI agent on the system platform classifies each new defect. Based on multiple factors such as reported image content, defect size, defect type, geographical location, road grade, and past development data, the agent automatically assesses the severity and urgency of the defect, categorizing it into different levels such as "requires immediate action," "included in the plan," and "continuous monitoring," and writes the classification results into the defect file. For monitored defects, the platform establishes a time-triggered mechanism to continuously monitor subsequent inspections and periodic follow-ups, constantly updating the defect file and providing timely warnings of abnormal changes. For defects requiring immediate action, the autonomous AI agent automatically generates a maintenance plan in the system platform's road and bridge work order system, intelligently recommending maintenance solutions (such as selecting optimal processes, materials, and equipment), automatically filling in information such as department affiliation and risk level, and forming a defect work order to be dispatched.

[0057] In some embodiments, the disease identification and grading algorithm can be implemented by a small edge model or a cloud-based multimodal model, and the deduplication method can be implemented based on rules or data-driven methods (graph matching / vector analysis).

[0058] Step S104: Based on multiple work order allocation factors, the defect work order is sent to the target object through the super intelligent agent.

[0059] In one example, after a work order for a fault is generated, the super-intelligent agent comprehensively considers the maintenance team's geographical location, professional skills, and current workload, automatically optimizing the work order assignment and placing it in the most suitable team. Simultaneously, the system platform's road and bridge dispatching system pushes dispatch messages to the instant messaging apps of relevant personnel and team members, ensuring timely communication of task information. The system platform also continuously monitors the timeout status of work orders, providing real-time progress reminders to teams to prevent work order delays. Instant messaging apps can include WeChat / DingTalk, Lark, and enterprise-specific IM.

[0060] Step S105: Obtain the defect repair information of the defect work order through the autonomous intelligent agent, evaluate the repair effect based on the defect repair information, and obtain the repair effect evaluation result.

[0061] In some embodiments, the defect repair information includes pre- and post-construction image information uploaded by repair personnel according to guidance information. A method for evaluating the repair effect based on the defect repair information and obtaining a repair effect evaluation result includes: determining whether the image information is qualified through a multimodal large model; if it is not qualified, notifying the corresponding personnel to resubmit the image information; if it is qualified, comparing the relevant modal information before construction and the relevant modal information after construction, and obtaining the repair effect evaluation result based on the comparison result; if the repair effect evaluation result does not meet the review requirements, reporting to the super agent, and instructing the super agent to take countermeasures through the corresponding personnel.

[0062] In one example, during on-site maintenance, the maintenance team uploads photos before, during, and after the work, following platform instructions, and submits data such as work trajectory and working hours. The autonomous intelligent agent uses a multimodal large model to analyze each uploaded photo in real time, automatically determining whether the image content is qualified and meets the repair standards. If it finds suspected non-standard practices, duplicate shots, or missing photos, it immediately provides feedback to the team for reshooting or rectification. After the maintenance is completed, the autonomous intelligent agent automatically compares all multimodal data before and after the maintenance, comprehensively analyzes the maintenance effect, and provides a preliminary conclusion on whether the work order passes the review. For work orders with low review confidence, inconsistencies, or complex issues, the autonomous intelligent agent system will proactively report to the super intelligent agent. The super intelligent agent will then push notifications to the dispatcher or manager as needed, accelerating decision-making and ensuring quality through human-machine collaboration. Throughout the process, if the autonomous intelligent agent encounters knowledge gaps, uncertainties, or data anomalies at any time, the super intelligent agent will automatically conduct further summarization and decision-making. If the problem still cannot be resolved automatically, it will be pushed to the system's intelligent agent interaction platform or human terminal for online decision-making assistance from managers. All complex decision-making processes are fully logged and archived in the defect file. The entire lifecycle of disease information and all historical data are dynamically maintained in the database. System administrators or road and bridge decision-makers can view the list of all diseases at any time on the intelligent agent interaction platform and filter them according to multiple dimensions such as time, disease level, and current status. Each disease detail page brings together interactive graphics and text, key considerations, and action logs of the entire process from initial discovery, each repair, result review to inspection and monitoring.

[0063] In some embodiments, disease-related information is obtained from disease archives and projected onto a GIS visualization map; the disease-related information includes disease lifecycle status and agent decision chain, and the disease lifecycle status includes any one of pending, continuous monitoring and processed.

[0064] In one example, the platform projects the lifecycle status, key indicators, and agent decision chains of all road defects onto a GIS visualization map (such as a large screen in a road and bridge cockpit) in real time. Different types and states of defects are dynamically displayed with different colors and icons, allowing managers to intuitively grasp the health status of the entire city's road network at a glance and check details as needed.

[0065] In some embodiments, based on the received intelligent report request, one or more of the following can be generated through automated data querying and AI analysis capabilities: multidimensional charts, analytical text, and PPT presentations.

[0066] In one example, regarding data analysis and information services, managers can initiate intelligent report requests using natural language or preset templates through the intelligent agent interaction platform. Leveraging automated data querying and AI analysis capabilities, the system supports the instant generation of multi-dimensional charts and analytical texts, including disease statistics, maintenance cycles, resource allocation, and trend analysis. It can also generate PPT presentations with a single click, automatically pushing them to management's DingTalk or email accounts, significantly improving reporting efficiency. The platform also supports real-time human-computer dialogue via a conversational interface, allowing managers to issue commands to the intelligent agent, inquire about specific disease progress, and intervene in scheduling processes. The intelligent agent possesses multi-turn dialogue and contextual memory capabilities, enabling seamless completion of common management operations and complex analysis requests through natural language.

[0067] In one example, all business modules of the platform are provided as API services, supporting integration with road and bridge dispatching systems, databases, map systems, enterprise messaging platforms, etc., and providing a reusable foundation for future deployment in smart city management, highways, bridges and tunnels, and other scenarios. The overall technical solution realizes a new paradigm of smart road maintenance management, characterized by intelligent perception, automatic identification, accurate decision-making, full-process tracking, automatic push notifications, data archiving, and continuous optimization, greatly reducing manual labor and improving the digitalization and intelligence level of urban road operation and maintenance.

[0068] Compared with existing technologies, this application integrates a disease management automatic decision-making framework and its multi-agent communication mechanism that combines a super intelligent agent and an autonomous intelligent agent collaborative system. It has a multi-modal model-driven AI closed-loop capability for automatic inspection identification, deduplication, classification, and compliance review throughout the entire process. It also features integrated innovations in conversational intelligent collaboration and instant push mechanisms. It realizes the intelligent agent's proactive wake-up, confidence judgment, automatic reporting, and human-machine hybrid decision-making flow in uncertain scenarios. It supports automated support for multi-terminal visualization, panoramic view of disease electronic archives, map-based source tracing, and one-click generation of intelligent reports and PPTs. It achieves high-granularity and refined management with "individual disease" as the life cycle as the main line, combined with disease archives with unique IDs and individualized maintenance by autonomous intelligent agents. It can automatically record the discovery, diagnosis, treatment, tracking, recurrence, and all historical maintenance content of diseases, forming traceable data assets and promoting knowledge discovery. Secondly, the platform adopts a multi-agent collaborative architecture. A super-agent is responsible for overall analysis and key decisions, while autonomous agents handle specific tasks. Both are highly autonomous and support message communication. When encountering complex problems or decisions with low confidence, human intervention can be automatically invoked, balancing intelligent automation with management security. Thirdly, the system comprehensively incorporates a multimodal large-scale model, supporting intelligent recognition, deduplication, classification, and compliance verification of multi-source inputs such as patrol vehicle images and structured data, significantly improving data processing accuracy and efficiency. Furthermore, the platform has a built-in conversational collaboration and instant messaging push mechanism, deeply integrated with enterprise IM platforms such as DingTalk or WeChat, enabling multi-terminal push and multi-round dialogue collaboration for disease work orders, approvals, notifications, and reports, significantly improving workflow efficiency and the flexible human-machine collaboration experience. Finally, the system supports global disease electronic archive map visualization and intelligent analysis, facilitating multi-dimensional queries by disease, time, and geographical location for managers. It also has the ability to generate intelligent reports and PPTs with one click, greatly freeing up management and decision-making manpower.

[0069] The automated urban road damage processing method based on large-model intelligent agents in this application uses a collaborative system composed of a super intelligent agent driven by a large model and an autonomous intelligent agent to realize the automated creation of intelligent archives for the entire life cycle of damage and the individualized maintenance mechanism of autonomous intelligent agents. It realizes closed-loop management of the entire life cycle of damage, from discovery, classification, dispatch, repair, review to long-term monitoring and data analysis. The system is highly automated and intelligent, and fully integrates multimodal data, dialogue collaboration and business process visualization, significantly improving business efficiency and governance capabilities.

[0070] To achieve the above embodiments, this application also proposes an automated urban road damage processing device based on a large model intelligent agent. Figure 2 This is a block diagram of an automated urban road defect processing device based on a large-model intelligent agent, provided as an embodiment of this application. Figure 2As shown, the automated urban road defect processing device based on a large model intelligent agent may include: an information reporting module 201, an archiving module 202, a work order processing module 203, a work order dispatching module 204, and a work order review module 205.

[0071] Among them, the information reporting module 201 is used to obtain disease reporting information, which includes disease scene information, location information and disease identification results; The archiving module 202 is used to add disease reporting information to existing disease files or create new disease files through a super intelligent agent. Each disease file corresponds to one disease and one autonomous intelligent agent. The work order processing module 203 is used to classify the diseases in the new disease files through an autonomous intelligent agent to obtain the disease level; and generate a disease work order when the disease level is that it needs to be treated. The work order dispatch module 204 is used to send defect work orders to the target object through a super intelligent agent based on multiple work order allocation factors. The work order review module 205 is used to obtain the defect repair information of the defect work order through an autonomous intelligent agent, and evaluate the repair effect based on the defect repair information to obtain the repair effect evaluation result. Furthermore, in one possible implementation of this application embodiment, the information reporting module 201 is specifically used for: Collect relevant information about the objects being inspected, including scene and location information presented through videos or pictures; The edge-side multimodal AI model is used to identify diseases in the scene information, and the disease identification results are obtained, including the disease type and severity.

[0072] Furthermore, in one possible implementation of this application embodiment, the archiving processing module 202 is specifically used for: Using multiple matching algorithms, the first disease in the disease report is compared with all diseases in the disease archive. If the first disease corresponding to the disease report information is consistent with the second disease in the disease archive, add information to the disease archive corresponding to the second disease based on the disease report information; If the first disease in the disease report is inconsistent with all diseases in the disease archive, a new disease file will be created for the first disease based on the disease report information.

[0073] Furthermore, in one possible implementation of this application embodiment, before performing a full comparison between the first disease in the disease reporting information and all diseases in the disease archive, the archiving processing module 202 is further configured to: Based on the disease scenario information, the reported disease information is verified; After verification, the relevant information of the first disease corresponding to the disease report is standardized.

[0074] Furthermore, in one possible implementation of this application embodiment, the work order dispatch module 204 is also used for: When the disease level is under continuous monitoring, a time-triggered mechanism is used to conduct subsequent inspections and regular follow-ups on the first disease.

[0075] Furthermore, in one possible implementation of this application embodiment, the work order processing module 203 is specifically used for: Based on the relevant information of the diseases in the new disease files, the diseases in the new disease files are classified to obtain disease levels, and the disease levels are added to the new disease files. When the disease level is deemed to require treatment, a maintenance plan is generated; Obtain recommended repair solutions, and generate defect work orders based on the repair plan and recommended repair solutions.

[0076] Furthermore, in one possible implementation of this application embodiment, the work order review module 205 is specifically used for: Determine whether the image information is qualified by using a multimodal large model; If the image is not satisfactory, notify the relevant personnel to provide the image information again. If qualified, compare the relevant modal information before construction with the relevant modal information after construction, and obtain the maintenance effect evaluation result based on the comparison results; If the maintenance effectiveness assessment results do not meet the review requirements, the matter is reported to the super intelligent agent, which then takes appropriate measures through the corresponding personnel.

[0077] Furthermore, in one possible implementation of this application embodiment, the device further includes a map projection module, used for: Obtain disease-related information from disease records and project it onto a GIS visualization map. Disease-related information includes disease lifecycle status and agent decision chain. Disease lifecycle status includes any one of pending, continuous monitoring, and processed status.

[0078] Furthermore, in one possible implementation of this application embodiment, the apparatus further includes a report generation module, used for: Based on the received intelligent report request, one or more of the following can be generated through automated data querying and AI analysis capabilities: multi-dimensional charts, analytical text, and PPT presentations.

[0079] It should be noted that the foregoing explanation of the embodiment of the automated urban road damage processing method based on large model intelligent agents also applies to the automated urban road damage processing device based on large model intelligent agents in this embodiment, and will not be repeated here.

[0080] To implement the above embodiments, this application also proposes an electronic device. Please see [link to relevant documentation]. Figure 3 , Figure 3 This is a block diagram of the electronic device provided in an embodiment of this application. For example... Figure 3 As shown, the electronic device 300 includes: a processor 301 and a memory 302 communicatively connected to the processor 301; the memory 302 stores computer execution instructions; the processor 301 executes the computer execution instructions stored in the memory to implement the method provided in the foregoing embodiments.

[0081] To implement the above embodiments, this application also proposes a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the methods provided in the foregoing embodiments.

[0082] To implement the above embodiments, this application also proposes a computer program product, including a computer program that, when executed by a processor, implements the methods provided in the foregoing embodiments.

[0083] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0084] It should be noted that personal information collected from users should be used for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. Furthermore, such collection / sharing should only be conducted after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign an agreement / authorization that includes authorization of relevant user information before the user uses the function. In addition, any necessary steps must be taken to protect and safeguard access to such personal information data and ensure that others with access to personal information data comply with their privacy policies and procedures.

[0085] This application is intended to provide an implementation scheme for users to selectively prevent the use or access to their personal information data. Specifically, this disclosure is intended to provide hardware and / or software to prevent or block access to such personal information data. Once personal information data is no longer needed, risks can be minimized by restricting data collection and deleting data. Furthermore, where applicable, such personal information is de-identified to protect user privacy.

[0086] In the foregoing descriptions of the embodiments, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0087] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0088] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0089] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0090] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0091] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0092] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0093] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. An automated method for handling urban road defects based on a large-scale intelligent agent, applied to an automated urban road maintenance management system, wherein the system comprises a super intelligent agent and multiple autonomous intelligent agents, characterized in that... Includes the following steps: Obtain disease reporting information, which includes disease scene information, location information, and disease identification results; The super intelligent agent adds the disease reporting information to the existing disease file or creates a new disease file. Each disease file corresponds to one disease and one autonomous intelligent agent. The autonomous intelligent agent performs classification processing on the diseases in the new disease files to obtain disease levels; and when the disease level is required to be treated, a disease work order is generated. Based on multiple work order allocation factors, the super intelligent agent sends the defect work order to the target object. The autonomous intelligent agent obtains the defect repair information of the defect work order, evaluates the repair effect based on the defect repair information, and obtains the repair effect evaluation result.

2. The method according to claim 1, characterized in that, The acquisition of disease reporting information includes: Collect relevant information about the patrol targets, including scene information and location information presented through videos or pictures; The scene information is used to identify diseases based on the edge-side multimodal AI model to obtain disease identification results, which include disease type and severity.

3. The method according to claim 1, characterized in that, Adding the reported disease information to an existing disease file or a new disease file includes: Using multiple matching algorithms, the first disease in the reported disease information is compared with all diseases in the disease archive. If the first disease corresponding to the disease reporting information is consistent with the second disease in the disease archive, information is added to the disease archive corresponding to the second disease based on the disease reporting information; If the first disease in the disease reporting information is inconsistent with all diseases in the disease archive, a new disease archive is created for the first disease based on the disease reporting information.

4. The method according to claim 3, characterized in that, Before performing a full comparison of the first disease in the reported disease information with all diseases in the disease archive, the following steps are included: Based on the disease scenario information, the disease reporting information is verified; After verification, the relevant information of the first disease corresponding to the reported disease information is standardized.

5. The method according to claim 1, characterized in that, The process involves classifying the diseases in the new disease files to obtain disease levels; and when the disease level requires treatment, a disease work order is generated, including: Based on the relevant information of the diseases in the new disease file, the diseases in the new disease file are classified to obtain disease levels, and the disease levels are added to the new disease file. When the disease level is deemed to require treatment, a maintenance plan is generated; Obtain recommended repair solutions, and generate defect work orders based on the repair plan and recommended repair solutions.

6. The method according to claim 1, characterized in that, The method further includes: When the disease level is under continuous monitoring, the first disease is subject to subsequent inspections and regular follow-up examinations through a time-triggered mechanism.

7. The method according to claim 1, characterized in that, The defect repair information includes pre- and post-construction image information uploaded by repair personnel according to guidance information. The evaluation of repair effectiveness based on the defect repair information, to obtain the repair effectiveness evaluation result, includes: The image information is deemed qualified using a multimodal large model. If the image is not satisfactory, notify the relevant personnel to provide the image information again. If qualified, compare the relevant modal information before construction with the relevant modal information after construction, and obtain the maintenance effect evaluation result based on the comparison result; If the maintenance effect evaluation results do not meet the review requirements, the report is submitted to the super intelligent agent, which then takes countermeasures through the corresponding personnel.

8. The method according to claim 1, characterized in that, The method further includes: Obtain disease-related information from the disease archives and project the disease-related information onto a GIS visualization map; the disease-related information includes disease life cycle status and intelligent agent decision chain, and the disease life cycle status includes any one of pending processing, continuous monitoring, and processed.

9. The method according to claim 1, characterized in that, The method further includes: Based on the received intelligent report request, one or more of the following can be generated through automated data querying and AI analysis capabilities: multi-dimensional charts, analytical text, and PPT presentations.

10. An automated urban road damage treatment device based on a large-scale intelligent agent, the device being configured in an automated urban road maintenance management system, the system comprising a super intelligent agent and multiple autonomous intelligent agents, the device comprising: The information reporting module is used to obtain disease reporting information, which includes disease scene information, location information, and disease identification results. The archiving module is used to add the disease reporting information to existing disease files or create new disease files through the super intelligent agent. Each disease file corresponds to one disease and one autonomous intelligent agent. The work order processing module is used to classify the diseases in the new disease files through the autonomous intelligent agent to obtain the disease level; and generate a disease work order when the disease level is required to be treated. The work order dispatch module is used to send the defect work order to the target object through the super intelligent agent based on multiple work order allocation factors. The work order review module is used to obtain the defect repair information of the defect work order through the autonomous intelligent agent, evaluate the repair effect based on the defect repair information, and obtain the repair effect evaluation result.

Images