An unmanned aerial vehicle inspection system and method based on multi-agent collaboration

Abstract

The present invention belongs to the technical field of UAV inspection, and more specifically, relates to a UAV inspection system and method based on multi-agent collaboration. The system includes a preliminary preparation agent, a task planning agent, a monitoring and execution agent, a summary and induction agent, and a Memory module. Each agent contains a UAV tool set corresponding to its function. Each agent has a clear division of labor and can share information and make decision-making collaborations according to inspection requirements, improving the response speed and collaboration of inspection tasks and solving the information processing bottleneck in the traditional single-agent mode.

Description

Technical Field

[0001] The present invention belongs to the technical field of drone inspection, and more specifically, relates to a drone inspection system and method based on multi-agent collaboration. Background Art

[0002] Drones are widely used in fields such as power inspection, agricultural monitoring, and traffic management. Especially in infrastructure inspection, they have gradually become a common tool due to their flexibility and unmanned operation advantages. However, traditional drone inspection mission planning usually relies on a single system or intelligent agent to perform static planning of inspection routes, and lacks the ability to flexibly respond to environmental changes and sudden mission situations. In existing technologies, some solutions are based on fixed-route inspection methods, which usually use a single intelligent agent or a centralized control system for planning and management. For example, some power inspection drone systems use pre-planned paths when generating routes, or specify the inspection order through manual intervention, and perform simple monitoring feedback, making it difficult to adjust the path in real time or coordinate multiple drones to perform complex tasks. This method is easily affected by environmental changes, equipment failures, etc. during execution, making it difficult to complete the inspection task as expected.

[0003] The existing drone inspection planning technology has the following shortcomings:

[0004] Poor responsiveness: Traditional solutions lack the ability to dynamically adjust routes and often rely on fixed paths, making them incapable of responding to environmental changes or temporary mission requirements. This approach can easily lead to inspections being interrupted by external factors such as obstacles and weather changes.

[0005] Single-agent control: Most traditional solutions use centralized control, coordinating inspection tasks through only a single agent or control center. This makes it impossible to achieve division of labor and cooperation among multiple task roles, making it difficult for the system to effectively handle diverse task requirements and resulting in low monitoring and feedback efficiency.

[0006] Lack of distributed decision-making: In the event of drone failure or environmental changes, traditional methods cannot achieve distributed collaboration and decision-making among intelligent agents, thereby reducing the integrity and reliability of inspection tasks.

[0007] Chinese Patent Document CN111652460A discloses an intelligent optimization method and system for multi-UAV collaborative inspection of multiple transmission towers. The method includes: pre-establishing an information database for the inspection area; receiving an inspection task for a specified number of transmission towers within the inspection area; obtaining the capability information of the UAVs in the UAV base; according to the information database, the capability information of the UAVs and the inspection task, using a preset algorithm to determine the corresponding inspection plan, and issuing a driving instruction according to the inspection plan to drive the corresponding UAVs in the UAV base to depart and execute the inspection task. The system includes: an inspection control center module, a UAV control module, and an intelligent planning module. However, this invention completes the corresponding inspection task through the collaboration of multiple UAVs and uses a centralized control system for planning and management, lacking distributed decision-making.

[0008] In view of this, the present invention designs a UAV inspection system and method based on multi-agent collaboration. Summary of the Invention

[0009] The present invention aims to overcome at least one defect of the above-mentioned prior art and provides a UAV inspection system based on multi-agent collaboration to achieve dynamic adjustment, distributed collaboration, and efficient feedback of inspection tasks.

[0010] The present invention also discloses a UAV inspection method based on multi-agent collaboration.

[0011] The detailed technical solution of the present invention is as follows:

[0012] A UAV inspection system based on multi-agent collaboration, the system includes a pre-preparation agent, a task planning agent, a monitoring and execution agent, a summary and induction agent, and a Memory memory module; each agent contains a UAV tool set corresponding to the function;

[0013] The pre-preparation agent is responsible for helping the user complete the preliminary preparation work of the UAV inspection task;

[0014] The task planning agent is responsible for recommending the most suitable flight route for the user's UAV inspection task and dynamically adjusting it during the execution process;

[0015] The monitoring and execution agent is responsible for real-time monitoring of the execution status of the task during the UAV inspection process;

[0016] The summary and induction agent is responsible for confirming the task status and generating a detailed report after the task is completed;

[0017] The Memory memory module is responsible for information storage and information retrieval.

[0018] Preferably according to the present invention, the tool set of the preliminary preparation agent includes: control platform login, nest list query, and UAV query;

[0019] The tool set of the mission planning agent includes: route list query, mission status query, and mission creation;

[0020] The tool set of the monitoring and execution agent includes: UAV query, nest preparation, UAV takeoff, UAV point flight, and UAV action execution;

[0021] The tool set of the summary and induction agent includes: UAV return and mission archiving.

[0022] A UAV inspection method based on multi-agent collaboration, the method comprising the following steps:

[0023] S1. The user issues an instruction, and the preliminary preparation agent Setup Agent parses the user instruction and calls the nest list query tool and the UAV query tool to query the status of the nest and the UAV. If an abnormality is detected, an abnormality report is returned to the user, and the user is prompted to troubleshoot and provide new information in an interactive manner. After determining that the device status is normal, the execution right is transferred to the mission planning agent Dynamic Planning Agent;

[0024] S2. The mission planning agent Dynamic Planning Agent plans an initial route according to the inspection requirements and geographical environment provided by the user, and transmits the planning result to the monitoring and execution agent Task Management Agent; if information supplementation or confirmation is required during the planning process, the user is notified to supplement or confirm;

[0025] S3. The monitoring and execution agent Task Management Agent sequentially executes each step in the plan, periodically monitors the task execution situation, and when the task cannot continue to be executed, notifies the preliminary preparation agent Setup Agent to solve the device abnormality or notifies the mission planning agent Dynamic Planning Agent to re-plan until the flight is completed;

[0026] S4. After the inspection is completed, the summary and induction agent Completion Agent commands the UAV to return, and generates an inspection report according to the records during the inspection process.

[0027] Preferably according to the present invention, during the entire multi-agent interaction process, all agents can communicate with the Memory memory module to record or read important information related to their own stages.

[0028] Preferably according to the present invention, the technical implementation of all agents follows the design pattern of "planning - action - reflection", and formatted outputs are used in all three stages for easy parsing.

[0029] Preferably according to the present invention, the prompt words for the tasks executed by the preliminary preparation agent are as follows:

[0030] a. Log in to the drone control platform and confirm the user's permissions;

[0031] b. Check the status of the drone device to ensure the device is normal;

[0032] c. Confirm the availability of the take-off airport and the hangar to ensure that the drone can take off and return smoothly;

[0033] d. If the user encounters problems such as login failure or device anomalies, provide corresponding troubleshooting suggestions for the corresponding problems and assist in reconfiguration;

[0034] e. If the device status is normal and the airport is available, transfer to the task planning agent for task planning.

[0035] Preferably according to the present invention, the prompt words for the tasks executed by the task planning agent are as follows:

[0036] a. Plan an initial flight route according to the inspection requirements and geographical environment provided by the user;

[0037] b. If encountering obstacles, no-fly zones or weather changes, adjust the flight route in real time to ensure the safe flight of the drone;

[0038] c. During the task, if the user needs to modify the inspection area or task objectives, re-plan the flight route and provide an optimization plan;

[0039] d. Provide the flight route planning result and adjust according to the user's feedback;

[0040] e. Assist the user in customizing the flight route planning when necessary to ensure coverage of all inspection targets;

[0041] f. After the flight route planning is confirmed to be correct, transfer to the monitoring and execution agent.

[0042] Preferably according to the present invention, the prompt words for the tasks executed by the monitoring and execution agent are as follows:

[0043] a. Monitor the inspection progress of the drone;

[0044] b. During the task, if a flight failure, out-of-range, device problem or any abnormal situation occurs, immediately notify the user and provide a corresponding solution;

[0045] c. Trigger necessary operation tasks to ensure the smooth completion of the tasks;

[0046] d. Feed back all intermediate outputs of the agent to the user in real time to ensure they understand the task progress and potential risks;

[0047] e. In case of task anomalies, it can be transferred to the pre - prepared agent or task planning agent for fault repair or route adjustment;

[0048] f. After the task is completed, transfer to the summary and induction agent for return flight and task archiving.

[0049] Preferably according to the present invention, the prompting words for the tasks performed by the summary and induction agent are as follows:

[0050] a. Confirm whether the task has been completed, including whether all inspection points have been covered, whether the flight is smooth, and whether the equipment status is normal;

[0051] b. Summarize all inspection data;

[0052] c. Generate a detailed inspection report based on the information collected. The report content should include task progress, key events and solutions, equipment inspection results, etc.;

[0053] d. After the report is generated, prompt the user that the task has ended and provide a task summary;

[0054] e. If there are any omissions in the report, remind the user to supplement relevant information to ensure the integrity of the report.

[0055] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0056] (1) The present invention introduces four key agents to realize the division of labor and cooperation in the inspection tasks. The task planning agent can flexibly adjust the flight route according to the real - time environment and task requirements to adapt to complex inspection needs; the monitoring and execution agent monitors the task progress in real time and cooperates with external modules to improve the safety of the inspection process; the summary and induction agent automatically generates an inspection report and ensures the integrity of task records through the Memory module, improving the accuracy of data processing and the efficiency of task completion; each agent has a clear division of labor, and can share information and make decision - making cooperation according to inspection needs, improving the response speed and coordination of inspection tasks, and solving the information - processing bottleneck in the traditional single - agent method.

[0057] (2) The application of the "Planning - Action - Reflection" mode of the present invention enables the agent to have the ability to dynamically adapt to the task environment. At the same time, through the formatted phased output, it ensures the traceability and efficient optimization of the task execution process. Meanwhile, the communication of the Memory module is set up: it supports the information storage and sharing between agents, enabling the important information related to each stage to be stored and called across agents, and enhancing the integrity and traceability of the task process. Description of the Drawings

[0058] Figure 1 is the flowchart of a method for unmanned aerial vehicle inspection based on multi - agent cooperation according to the present invention.

[0059] Figure 2 is the "Planning - Action - Reflection" flowchart of the monitoring and execution agent in the embodiment of the present invention. Detailed Embodiment

[0060] The following further describes the present disclosure in conjunction with the drawings and embodiments.

[0061] Embodiment 1

[0062] This embodiment provides a system for unmanned aerial vehicle inspection based on multi - agent cooperation. The system includes a pre - preparation agent setup_agent, a task - planning agent dynamic_planning_agent, a monitoring and execution agent task_management_agent, a summarization agent completion_agent, and a Memory module; each agent contains a set of unmanned aerial vehicle tools corresponding to its function.

[0063] The pre - preparation agent is responsible for helping the user complete the pre - preparation work of the unmanned aerial vehicle inspection task.

[0064] The task - planning agent is responsible for recommending the most suitable flight path for the user's unmanned aerial vehicle inspection task and dynamically adjusting it during the execution process.

[0065] The monitoring and execution agent is responsible for monitoring the execution status of the task in real - time during the unmanned aerial vehicle inspection process.

[0066] The summarization agent is responsible for confirming the task status and generating a detailed report after the task is completed.

[0067] The Memory module is responsible for information storage and retrieval. For example, after one execution cycle of setup_agent ends, the memory summarizes the text content within this cycle and stores it in the database in the form of "title - content". At the start of the next execution of setup_agent, information saved from other agents can also be retrieved through the memory module to help itself understand the context.

[0068] The toolset of the preliminary preparation agent includes: control platform login, nest list query, and UAV query;

[0069] The toolset of the task planning agent includes: flight route list query, task status query, and task creation;

[0070] The toolset of the monitoring and execution agent includes: UAV query, nest preparation, UAV takeoff, UAV waypoint flight, and UAV action execution;

[0071] The toolset of the summarization agent includes: UAV return and task archiving.

[0072] The UAV tool design is shown in Table 1:

[0073] Table 1 UAV tool design table

[0074] Tool Name Tool Description Parameter Description Return Result Control Platform Login The user logs in to the UAV control platform to obtain API access rights Token: Login token Login Result (Success or Failure) Nest List Query Used to query the status of all nests None All nest IDs, nest status, etc. UAV Query Used to query the status of all UAVs None All UAV IDs, UAV types, affiliated nests, health status, task status (standby, in patrol, returning) Route List Query Used to query all patrol routes None All route IDs, route names, ordered waypoints, etc. Task Status Query Used to query the status of all tasks None Task ID, status (in progress, ended), used route, used UAV, etc. Task Creation Create a new patrol task Route ID, nest ID, UAV ID, etc. Task Creation Result (Success + Task ID or Failure + Failure Reason) Nest Preparation The UAV rises and the top of the nest opens Nest ID, UAV ID Nest Preparation Result (Success or Failure) UAV Takeoff The UAV leaves the nest and takes off UAV ID UAV Takeoff Result (Success or Failure) UAV Point-to-Point Flight Instruct the UAV to fly to a specified location UAV ID, target longitude and latitude Point-to-Point Flight (Success or Failure) UAV Action Execution Instruct the UAV to execute an action UAV ID, action type (alarm, voice broadcast, take pictures, etc.) Action Execution Result UAV Return UAV return command UAV ID Return Execution Result Task Archiving Summarize all patrol information Task ID, patrol record Archiving Result (Success or Failure)

[0075] Example 2

[0076] As Figure 1 shown, this embodiment provides a UAV inspection method based on multi - agent cooperation. The method includes:

[0077] S1. The User Agent issues an instruction, and the Setup Agent parses the user instruction and calls tools to query the status of the nest and UAV. If an anomaly is detected, an anomaly report is returned to the user, and the user is prompted in an interactive manner to troubleshoot and provide new information. After determining that the device status is normal, the execution right is transferred to the Dynamic Planning Agent.

[0078] S2. The Dynamic Planning Agent plans an initial flight route according to the inspection requirements and geographical environment provided by the user, and transfers the planning result to the Task Management Agent; during the planning process, if information supplementation or confirmation is required, the user is notified to supplement or confirm.

[0079] S3. The Monitoring and Execution Agent (Task Management Agent) sequentially executes each step in the plan, acting as a "co-pilot" during the UAV's mission execution (the "pilot" role is played by external modules such as the UAV's real-time obstacle avoidance). It periodically monitors the mission execution status. When the mission cannot continue, it notifies the Preparatory Agent (Setup Agent) to resolve equipment anomalies or notifies the Mission Planning Agent (Dynamic Planning Agent) to re-plan until the flight is completed;

[0080] S4. After the inspection is completed, the Summarization Agent (Completion Agent) commands the UAV to return and generates an inspection report based on the records during the inspection.

[0081] During the entire multi-agent interaction process, all agents can communicate with the Memory module to record or read important information related to their own stages.

[0082] The technical implementation of all agents follows the "Plan-Act-Reflect" design pattern, and all three stages use formatted outputs for easy parsing.

[0083] The "Plan-Act-Reflect" design pattern enables agents to have the ability to dynamically adapt to the task environment, as follows:

[0084] The "Plan" step constrains the agent to first output its thinking content when facing complex problems, and then output more sub-step guidelines. Complex problems often need to be disassembled through planning. Agents lacking planning tend to talk nonsense when facing complex problems.

[0085] The "Act" step provides the agent with the ability to call tools. For example, for the question "Please tell me the current time", without external tools, the agent cannot answer this question. With the "time query tool", the agent's answer history may be "Please tell me the current time -> To answer this question, I need to call the tool check_time() -> The result of calling check_time: 2025 / 1 / 2 -> I now know the answer, and the current time is 2025 / 1 / 2".

[0086] The "Reflect" step is used to constrain the agent to perform checks and summaries, providing the agent with the ability to "look back" when the agent makes a final mistake due to an error in a previous step, increasing the robustness of the response result.

[0087] Take Figure 2Taking the specific implementation of the monitoring and execution agent, the Task Management Agent, as an example, first, the Task Management Agent combines the task (system prompt) with the input information (flight route planning result) for planning. Then, it selects tools to complete the subtasks in the planned content. Finally, it observes the updated context after the action, that is, the real-time inspection information, reflects, and outputs the next action, whether to continue execution, finish execution, or transfer to other agents. For other agents, the tasks, input information, action / reflection context, available tool list, and reflection return type in the Task Management Agent flowchart can be replaced to complete the migration.

[0088] The prompt of the aforementioned preliminary preparation agent is as follows:

[0089] You are an intelligent assistant responsible for helping users complete the preliminary preparation work of the drone inspection task. You can use the following tools:

[0090] {tool_descs}

[0091] You will assist users in completing the following tasks:

[0092] a. Log in to the drone management and control platform and confirm the user's permissions;

[0093] b. Check the status of devices such as the drone battery, motor, and sensor to ensure the normal operation of the devices;

[0094] c. Confirm the availability of the take-off airport and the hangar to ensure that the drone can take off and return smoothly;

[0095] d. If the user encounters a login failure or device problem, provide corresponding troubleshooting suggestions for the problem and assist in reconfiguration;

[0096] e. If the device status is normal and the airport is available, transfer to the task planning agent for task planning.

[0097] The prompt of the aforementioned task planning agent is as follows:

[0098] You are a flight route planning expert responsible for recommending the most suitable flight route for the user's drone inspection task and dynamically adjusting it during the execution. You can use the following tools:

[0099] {tool_descs}

[0100] Your tasks include:

[0101] a. Plan the initial flight route according to the inspection requirements and geographical environment provided by the user;

[0102] b. If an obstacle, no - fly zone, or weather change is encountered, adjust the flight path in real - time to ensure the safe flight of the drone;

[0103] c. During the mission, if the user needs to modify the inspection area or mission objective, re - plan the flight path and provide an optimization plan;

[0104] d. Provide the flight path planning result and adjust it according to the user's feedback;

[0105] e. When necessary, assist the user in customizing the flight path planning to ensure that all inspection objectives are covered;

[0106] f. After the flight path planning is confirmed to be correct, transfer to the monitoring and execution agent.

[0107] The prompt words of the said monitoring and execution agent are as follows:

[0108] You are a task management assistant, responsible for monitoring the execution status of the mission in real - time during the drone inspection. You can use the following tools:

[0109] {tool_descs}

[0110] Your tasks include:

[0111] a. Monitor the inspection progress of the drone, including the current inspection point, flight status, battery power, etc.;

[0112] b. During the mission, if a flight failure, out - of - range situation, equipment problem, or any abnormal situation occurs, immediately notify the user and provide the corresponding solution;

[0113] c. Trigger necessary operation tasks, such as adjusting the flight path, restarting the mission, conducting temporary inspections, etc., to ensure the smooth completion of the mission;

[0114] d. Provide real - time feedback of all intermediate outputs of the agent to the user to ensure that they understand the mission progress and potential risks; the said intermediate outputs include planning, tool calls, summary and reflection, transfer of speaking rights, etc.;

[0115] e. In case of mission anomalies, transfer to the pre - mission preparation agent or mission planning agent for fault repair or flight path adjustment;

[0116] f. After the mission is completed, transfer to the summary and induction agent for returning and mission archiving.

[0117] The prompt words of the said summary and induction agent are as follows:

[0118] You are an inspection report generation expert, responsible for confirming the mission status and generating a detailed report after the mission is completed. You can use the following tools:

[0119] {tool_descs}

[0120] Your tasks include:

[0121] a. Confirm whether the task has been completed, including whether all inspection points have been covered, whether the flight is smooth, and whether the equipment status is normal.

[0122] b. Summarize all inspection data, including route details, UAV status, task history records, exception handling situations, etc.

[0123] c. Generate a detailed inspection report based on the information collected. The report content should include task progress, key events and solutions, equipment inspection results, etc.; the information collected includes: execution results of the upstream agent, retrieval results of the memory module, information provided by the user, results of the agent's call to the tool, etc.

[0124] d. After the report is generated, prompt the user that the task has ended and provide a task summary.

[0125] e. If there are any omissions in the report, remind the user to supplement relevant information to ensure the integrity of the report.

[0126] Based on the multi-agent UAV inspection system and method of the present invention, three experimental tests as shown in Table 2 were carried out:

[0127] Table 2 Experimental test table of multi-agent UAV inspection system and method

[0128]

[0129] The experimental results show that the system and method of the present invention have achieved excellent performance in substation inspection tasks, transmission line inspection tasks, and water conservancy dam inspection tasks, improving the inspection efficiency.

Claims

1. An unmanned aerial vehicle inspection system based on multi-agent collaboration, characterized in that The system includes a preliminary preparation agent, a task planning agent, a monitoring and execution agent, a summarization agent, and a Memory memory module; each agent contains a UAV toolset corresponding to its function; The preliminary preparation agent is responsible for helping the user complete the preliminary preparation work for the UAV inspection task; The toolset of the preliminary preparation agent includes: control platform login, nest list query, and UAV query; the preliminary preparation agent is used to parse the user's instructions and call the nest list query tool and the UAV query tool to query the status of the nest and the UAV. If an abnormality is detected, an abnormality report is returned to the user, and the user is prompted to troubleshoot and provide new information in an interactive manner. After determining that the device status is normal, the execution right is transferred to the task planning agent; The task planning agent is responsible for recommending the most suitable flight path for the user's UAV inspection task and dynamically adjusting it during the execution process; the toolset of the task planning agent includes: flight path list query, task status query, and task creation; the task planning agent is used to plan the initial flight path according to the inspection requirements and geographical environment provided by the user, and transfer the planning result to the monitoring and execution agent; if information supplementation or confirmation is required during the planning process, the user is notified to supplement or confirm; The monitoring and execution agent is responsible for monitoring the execution status of the task in real time during the UAV inspection process; the toolset of the monitoring and execution agent includes: UAV query, nest preparation, UAV takeoff, UAV waypoint flight, and UAV action execution; the monitoring and execution agent is used to execute each step in the plan in sequence, periodically monitor the task execution status, and when the task cannot continue to be executed, notify the preliminary preparation agent to solve the device abnormality or notify the task planning agent to re-plan until the flight is completed; The summarization agent is responsible for confirming the task status and generating a detailed report after the task is completed; the toolset of the summarization agent includes: UAV return and task archiving; the summarization agent is used to command the UAV to return and generate an inspection report based on the records during the inspection process; The technical implementation of all agents follows the "planning - action - reflection" design pattern, and all three stages use formatted output for easy parsing; The Memory memory module is responsible for information storage and information retrieval.

2. A method for unmanned aerial vehicle (UAV) inspection based on multi-agent collaboration, using the UAV inspection system based on multi-agent collaboration as described in claim 1, characterized in that, The method includes the following steps: S1. The user issues an instruction, and the preliminary preparation agent parses the user's instruction and calls the nest list query tool and the UAV query tool to query the status of the nest and the UAV. If an abnormality is detected, an abnormality report is returned to the user, and the user is prompted to troubleshoot and provide new information in an interactive manner. After determining that the device status is normal, the execution right is transferred to the task planning agent; S2. The task planning agent plans the initial flight path according to the inspection requirements and geographical environment provided by the user, and transfers the planning result to the monitoring and execution agent; if information supplementation or confirmation is required during the planning process, the user is notified to supplement or confirm; S3. The monitoring and execution agent sequentially executes each step in the plan, periodically monitors the task execution status, and when the task cannot continue to be executed, notifies the preliminary preparation agent to resolve equipment anomalies or notifies the task planning agent to re-plan until the flight is completed; S4. After the inspection is completed, the summarization agent commands the UAV to return, and generates an inspection report based on the records during the inspection process.

3. The method for unmanned aerial vehicle inspection based on multi-agent collaboration according to claim 2, wherein During the entire multi-agent interaction process, all agents can communicate with the Memory module to record or read important information related to their own stages.

4. The method for UAV inspection based on multi-agent collaboration according to claim 2, wherein, The technical implementation of all agents follows the "planning-action-reflection" design pattern, and all three stages use formatted output for easy parsing.

5. The method for unmanned aerial vehicle inspection based on multi-agent collaboration according to claim 2, wherein, The prompt words for the tasks executed by the preliminary preparation agent are as follows: a. Log in to the UAV control platform and confirm user permissions; b. Check the status of the UAV equipment to ensure normal operation; c. Confirm the availability of the take-off airport and the nest to ensure that the UAV can take off and return smoothly; d. If the user encounters problems such as login failure or equipment anomalies, provide corresponding troubleshooting suggestions for the corresponding problems and assist in reconfiguration; e. If the equipment status is normal and the airport is available, transfer to the task planning agent for task planning.

6. The method for UAV inspection based on multi-agent collaboration according to claim 2, characterized in that, The prompt words for the tasks executed by the task planning agent are as follows: a. Plan the initial flight path according to the inspection requirements and geographical environment provided by the user; b. If encountering obstacles, no-fly zones or weather changes, adjust the flight path in real time to ensure the safe flight of the UAV; c. During the task, if the user needs to modify the inspection area or task objectives, re-plan the flight path and provide an optimization plan; d. Provide the flight path planning result and adjust according to the user's feedback; e. Assist the user in customizing the flight path planning when necessary to ensure coverage of all inspection targets; f. After the flight path planning is confirmed to be correct, transfer to the monitoring and execution agent.

7. The method for UAV inspection based on multi-agent collaboration according to claim 2, wherein The prompt words for the tasks executed by the monitoring and execution agent are as follows: a. Monitor the inspection progress of the UAV; b. During the task, if a flight failure, out-of-range, equipment problem or any abnormal situation occurs, immediately notify the user and provide corresponding solutions; c. Trigger necessary operation tasks to ensure the smooth completion of the task; d. Provide real-time feedback of all intermediate outputs of the agent to the user to ensure that they understand the task progress and potential risks; e. When the task is abnormal, it can transfer to the preliminary preparation agent or the task planning agent for fault repair or flight path adjustment; f. After the task is completed, transfer to the summarization agent for return and task archiving.

8. A method for UAV inspection based on multi-agent collaboration according to claim 2, characterized in that The prompt words for the tasks executed by the summarization agent are as follows: a. Confirm whether the task has been completed, including whether all inspection points have been covered, whether the flight is smooth, and whether the equipment status is normal; b. Summarize all inspection data; c. Generate a detailed inspection report based on the information collected, and the report content should include task progress, key events and solutions, equipment inspection results, etc.; d. After the report is generated, notify the user that the task has ended and provide a task summary; e. If there are any omissions in the report, remind the user to supplement relevant information to ensure the completeness of the report.

Images