An embodied intelligent satellite large model application method and embodied intelligent satellite large model for autonomous perception, decision-making and execution

Abstract

The application relates to the technical field of satellite autonomous operation, and discloses a body-integrated intelligent satellite large model application method and a body-integrated intelligent satellite large model for autonomous perception, decision and execution integration, which acquire a remote sensing observation task, generate a first state perception instruction, acquire first state data of an on-orbit satellite based on the first state perception instruction, perform feasibility analysis on the on-orbit satellite by using the first state data to obtain feasibility data, perform task planning on the on-orbit satellite by using the feasibility data to generate an execution action plan, control the on-orbit satellite to execute the remote sensing observation task according to the execution action plan, acquire and feed back remote sensing observation data of the remote sensing observation task for execution condition evaluation, and the body-integrated intelligent satellite large model comprises an application agent module and contains multiple agents with specific functions and capable of being called. The body-integrated intelligent satellite large model has the beneficial effect of realizing autonomous task planning and decision of the on-orbit satellite and improving the autonomous operation capability and intelligent level of the on-orbit satellite.

Description

Technical Field

[0001] This application relates to the field of satellite autonomous operation technology, and in particular to an application method and a large-scale embodied intelligent satellite model for autonomous perception, decision-making and execution. Background Technology

[0002] When performing remote sensing observation missions using satellites in orbit, it is typically necessary to rely on ground stations to specify detailed mission plans and transmit them to the satellites. The satellites then perform observation actions according to the predetermined mission plan to complete the remote sensing observation mission. However, in related technologies, satellites in orbit generally lack the ability to adjust their plans based on current conditions. When conditions affecting remote sensing data, such as the environment or the satellite's own state, change, the satellites in orbit often cannot respond autonomously to these changes and must transmit the changes to the ground station, waiting for feedback and a replanning. This dependence on manual operation by ground stations often leads to missed opportunities for real-time responses, significantly negatively impacting the execution of remote sensing observation missions. Furthermore, it limits the autonomy of the satellites in orbit, resulting in low decision-making efficiency for remote sensing observation missions.

[0003] Among related technologies, the autonomous operation capability and intelligence level of on-orbit satellites still need to be improved. Summary of the Invention

[0004] This application provides a method for applying an embodied intelligent satellite large-scale model that integrates autonomous perception, decision-making, and execution, as well as the embodied intelligent satellite large-scale model itself. The embodied intelligent satellite large-scale model is deployed on the satellite, and multiple functional modules in the embodied intelligent satellite large-scale model are used to call intelligent agents to complete corresponding tasks. Through the collaborative work of multiple intelligent agents, autonomous mission planning and decision-making of the on-orbit satellite is realized, thereby improving the autonomous operation capability and intelligence level of the on-orbit satellite.

[0005] To achieve the above objectives, the main technical solutions adopted in this application include: In a first aspect, embodiments of this application provide a method for applying an embodied intelligent satellite large-scale model integrating autonomous perception, decision-making, and execution. The embodied intelligent satellite large-scale model includes a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application intelligent agent module. The method includes: The cognitive reasoning module acquires the remote sensing observation task and generates a corresponding first state perception command; the multimodal perception module acquires the first state data of the on-orbit satellite before it executes the remote sensing observation task based on the first state perception command. The cognitive reasoning module uses the first state data to perform a feasibility analysis on whether the on-orbit satellite is suitable for performing the remote sensing observation mission, and obtains the feasibility data of the on-orbit satellite before performing the remote sensing observation mission. The autonomous planning module uses the feasibility data to perform mission planning for the on-orbit satellite, generating an execution action plan for the on-orbit satellite for the remote sensing observation mission. The execution feedback module controls the on-orbit satellite to perform the remote sensing observation task according to the execution action plan; it acquires and feeds back the remote sensing observation data of the remote sensing observation task for performance evaluation. The application intelligent agent module includes multiple intelligent agents with specific functions, which can be called by other modules in the embodied intelligent satellite large model.

[0006] The embodiment of this application proposes an application method for an embodied intelligent satellite large-scale model integrating autonomous perception, decision-making, and execution. This method deploys an embodied intelligent satellite large-scale model on the satellite, including a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application intelligent agent module. After issuing a remote sensing observation task, the cognitive reasoning module obtains the task and instructs the multimodal perception module to acquire the first state data of the on-orbit satellite. The cognitive reasoning module performs a task feasibility analysis based on the first state data, serving as the data basis for the autonomous planning module's task planning, resulting in the on-orbit satellite's execution action plan. The execution feedback module controls the on-orbit satellite to execute the remote sensing observation task according to the execution action plan and acquires remote sensing observation data for performance evaluation. During application, these modules can call upon corresponding intelligent agents in the application intelligent agent module, enabling collaborative work among multiple intelligent agents. Compared with related technologies, this application utilizes multiple functional modules within an embodied intelligent satellite model to invoke intelligent agents. Through collaborative work among these agents, autonomous state perception, mission planning, and execution feedback are achieved on the satellite side, reducing the satellite's dependence on ground stations and enhancing its autonomous operation capabilities. Furthermore, the satellite side acquires multimodal state data from the on-orbit satellite via intelligent agents as the data foundation for mission planning. It also acquires remote sensing observation data from the on-orbit satellite to evaluate the execution of remote sensing missions, providing a data basis for mission optimization and supplementation. This embodied intelligence is realized on the satellite side, improving the overall intelligence level of the on-orbit satellite.

[0007] Optionally, the first state data includes first internal state data and first external environment data of the on-orbit satellite before it performs the remote sensing observation mission; the feasibility analysis of whether the on-orbit satellite is suitable for performing the remote sensing observation mission using the first state data through the cognitive reasoning module, to obtain feasibility data of the on-orbit satellite before performing the remote sensing observation mission, includes: Based on the first internal state data, an internal state analysis is performed on the on-orbit satellite to obtain the internal state assessment results of the on-orbit satellite before it performs the remote sensing observation mission. Based on the first external environment data, an external environment analysis is performed on the on-orbit satellite to obtain the external environment assessment results before the on-orbit satellite performs the remote sensing observation mission. The feasibility data is obtained based on the internal state assessment results and the external environment assessment results.

[0008] Optionally, the first internal state data includes the operational state data and execution capability data of the on-orbit satellite before it performs the remote sensing observation mission; the step of performing internal state analysis on the on-orbit satellite based on the first internal state data to obtain the internal state assessment result of the on-orbit satellite before it performs the remote sensing observation mission includes: Based on the operational status data, a health status analysis is performed on the on-orbit satellite to obtain a health status assessment result of the on-orbit satellite before it performs the remote sensing observation mission. An execution condition assessment is performed on the execution capability data and the mission standard data of the remote sensing observation mission to obtain the execution condition assessment results before the on-orbit satellite performs the remote sensing observation mission; wherein, the mission standard data represents the requirements of the on-orbit satellite in terms of operational resources and observation capabilities for performing the remote sensing observation mission; The internal status assessment result is obtained based on the health status assessment result and the execution condition assessment result.

[0009] Optionally, the step of using the feasibility data to perform mission planning for the on-orbit satellite through the autonomous planning module, and generating the execution action plan for the on-orbit satellite for the remote sensing observation mission, includes: Based on the feasibility data, motion planning is performed on the in-orbit satellite to obtain the initial motion plan of the in-orbit satellite; Based on the initial action plan, the action scheduling of the on-orbit satellite is optimized to obtain the execution action plan for carrying out the remote sensing observation mission.

[0010] Optionally, the application agent module includes an evaluation and diagnostic agent, and the remote sensing observation data includes intermediate observation data obtained by executing some actions in the execution action plan; the method further includes: During the execution of the remote sensing observation task, the evaluation and diagnostic agent evaluates the execution status of the intermediate observation data to obtain intermediate evaluation results; If the intermediate evaluation result does not meet the preset quality conditions, the abnormal cause that caused the intermediate evaluation result to not meet the preset quality conditions is determined by the evaluation and diagnostic agent. The cognitive reasoning module generates a second state perception command corresponding to the cause of the anomaly, so as to optimize and adjust the execution action planning of the on-orbit satellite during the execution of the remote sensing observation mission and eliminate the impact of the anomaly on the remote sensing observation mission.

[0011] Optionally, the step of generating a second state perception command corresponding to the cause of the anomaly through the cognitive reasoning module to optimize and adjust the execution action plan of the on-orbit satellite during the execution of the remote sensing observation task includes: The multimodal sensing module acquires second state data of the on-orbit satellite during the execution of the remote sensing observation mission based on the second state sensing command; The cognitive reasoning module uses the second state data to determine whether the on-orbit satellite is suitable for performing the remote sensing observation task, and obtains the standard matching data of the on-orbit satellite in the process of performing the remote sensing observation task. The autonomous planning module uses the standard matching data to optimize and adjust the execution action plan of the on-orbit satellite, thereby obtaining the optimized action plan of the on-orbit satellite.

[0012] Optionally, the application agent module includes an evaluation and diagnostic agent, and the remote sensing observation data includes the complete execution result obtained after completing the execution action planning; the method further includes: After completing the execution action planning, the evaluation and diagnostic agent evaluates the execution status of the complete execution result to obtain the completion evaluation result; If the completion evaluation result indicates that the complete execution result does not meet the preset quality conditions, the cognitive reasoning module performs supplementary reasoning on the complete execution result to generate improvement task suggestions, which are used to generate supplementary tasks to improve the complete execution result.

[0013] Optionally, the embodied intelligent satellite large model further includes an on-orbit computing module, and the application intelligent agent module includes an image computing intelligent agent; the method further includes: For the target area specified in the remote sensing observation mission, the remote sensing imaging data of the target area is obtained through the on-orbit computing module; Based on the remote sensing observation mission, the on-orbit computing module calls the image computing agent to perform image recognition and information annotation on the remote sensing imaging data, thereby obtaining image recognition information about the remote sensing observation mission, which is then used to generate the remote sensing observation results of the remote sensing observation mission.

[0014] Optionally, when the remote sensing observation task is to map urban congestion, the image recognition information includes urban road information and vehicle identification information of the target area, and the remote sensing observation result includes congestion data of the target area; the remote sensing observation result is obtained in the following way: The image computing agent performs road segmentation on the remote sensing imaging data to obtain the urban road information; based on the remote sensing imaging data, the image computing agent performs vehicle identification on the urban roads represented by the urban road information to obtain the vehicle identification information. The congestion data is generated based on the urban road information and the vehicle identification information.

[0015] Optionally, when the remote sensing observation task is natural disaster assessment, the image recognition information includes the natural disaster characteristics of the target area, and the remote sensing observation results include the disaster impact prediction results of the target area; the remote sensing observation results are obtained in the following ways: Based on the natural disaster feature database, the image computing agent performs disaster feature identification on the remote sensing imaging data to obtain the natural disaster features of the target area. Based on the characteristics of natural disasters at multiple time points, the image computing agent performs disaster trend extrapolation on the target area to obtain disaster impact prediction results for the target area, which can be used for disaster early warning.

[0016] Secondly, embodiments of this application provide an embodied intelligent satellite large model, including a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application intelligent agent module. The embodied intelligent satellite large model is used to execute the method described in any one of the above embodiments.

[0017] Thirdly, embodiments of this application provide a computer device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the method described in any of the above embodiments.

[0018] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer instructions, which are used to cause a computer to perform the method described in any one of the above embodiments.

[0019] Fifthly, embodiments of this application provide an on-orbit satellite, characterized in that it performs remote sensing observation tasks using the method described in any one of the above embodiments.

[0020] Sixthly, embodiments of this application provide an on-orbit constellation, characterized in that it includes on-orbit satellites as described in any of the above embodiments. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 A step diagram illustrating the application method of an embodied intelligent satellite large-scale model that integrates autonomous perception, decision-making, and execution, provided in an embodiment of this application. Figure 2 A block diagram of the large-scale embodied intelligent satellite model provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the intelligent agent module used in the embodiments of this application; Figure 4 This is a flowchart illustrating the steps involved in conducting a feasibility analysis of an in-orbit satellite in this application embodiment; Figure 5 This is a flowchart illustrating the steps involved in performing partial state analysis on an on-orbit satellite in an embodiment of this application. Figure 6 This is a flowchart illustrating the steps involved in mission planning for an on-orbit satellite in this application embodiment; Figure 7 This is a flowchart illustrating the steps involved in evaluating the performance of an on-orbit satellite and replanning actions during mission execution, as described in this application embodiment. Figure 8 This is a flowchart illustrating the steps involved in optimizing and adjusting the execution action planning of an on-orbit satellite in this embodiment of the application. Figure 9 This is a flowchart illustrating the steps involved in evaluating the performance of an on-orbit satellite and performing supplementary reasoning after completing the execution action planning, as described in this application embodiment. Figure 10 This is a flowchart illustrating the steps involved in generating remote sensing observation results for a remote sensing observation task in an embodiment of this application. Figure 11 This is a flowchart illustrating the operational process of an on-orbit satellite performing remote sensing observation tasks, as described in this application embodiment. Figure 12 This is a block diagram of the satellite subsystem on the satellite in orbit, as described in the embodiments of this application. Figure 13 This is a flowchart illustrating the steps involved in generating remote sensing observation results when the remote sensing observation task is to map urban congestion, as described in this application embodiment. Figure 14This is a workflow diagram for the case where the remote sensing observation task is to map urban congestion, as described in this application embodiment. Figure 15 This is a flowchart illustrating the steps involved in generating remote sensing observation results in the context of a natural disaster assessment task, as described in this application embodiment. Figure 16 This is a workflow diagram for the case where the remote sensing observation task is natural disaster assessment, as described in this application embodiment. Figure 17 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] When performing remote sensing observation missions using in-orbit satellites, it is typically necessary to rely on ground stations to specify detailed mission plans and transmit them to the satellites. The satellites then perform observation actions according to the predetermined mission plans to complete the remote sensing observation mission. However, in related technologies, in-orbit satellites generally lack the ability to adjust their plans based on current conditions. When conditions affecting remote sensing data, such as the environment or the satellite's own state, change, the satellites often cannot respond autonomously and must transmit these changes to the ground station, waiting for feedback and a replanning. This dependence on manual operation by ground stations often leads to missed opportunities for real-time responses, significantly negatively impacting the execution of remote sensing observation missions. It also limits the autonomy of in-orbit satellites, resulting in low decision-making efficiency for remote sensing observation missions. Therefore, the autonomous operation capabilities and intelligence levels of in-orbit satellites still need improvement.

[0025] To address the aforementioned issues, this application provides an application method and a large-scale embodied intelligent satellite model for autonomous perception, decision-making, and execution. The method involves: acquiring remote sensing observation tasks and generating first-state perception commands; acquiring first-state data of the on-orbit satellite based on the first-state perception commands; performing feasibility analysis on the on-orbit satellite using the first-state data to obtain feasibility data; using the feasibility data to perform task planning on the on-orbit satellite and generate execution action plans; controlling the on-orbit satellite to execute remote sensing observation tasks according to the execution action plans; acquiring and feeding back remote sensing observation data for the remote sensing observation tasks to evaluate the execution status. The large-scale embodied intelligent satellite model includes an application intelligent agent module, containing multiple intelligent agents with specific functions that can be invoked.

[0026] This application provides a method for applying an embodied intelligent satellite large-scale model that integrates autonomous perception, decision-making, and execution. The method deploys an embodied intelligent satellite large-scale model on the satellite, including a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application agent module. After a remote sensing observation task is issued, the cognitive reasoning module acquires the task and instructs the multimodal perception module to acquire the first state data of the on-orbit satellite. The cognitive reasoning module performs a task feasibility analysis based on the first state data, serving as the data basis for the autonomous planning module's task planning, resulting in the on-orbit satellite's execution action plan. The execution feedback module controls the on-orbit satellite to execute the remote sensing observation task according to the execution action plan and acquires remote sensing observation data for performance evaluation. During application, these modules can call upon corresponding agents in the application agent module, enabling collaborative work among multiple agents.

[0027] Compared with related technologies, this application utilizes multiple functional modules in the embodied intelligent satellite model to call intelligent agents separately. Through the collaborative work between multiple intelligent agents, the satellite can achieve autonomous state perception, mission planning, and execution feedback, thereby reducing the dependence of the on-orbit satellite on the ground station and improving the autonomous operation capability of the on-orbit satellite.

[0028] In addition, the satellite also acquires multimodal state data of the satellite in orbit through intelligent agents as the data basis for mission planning. It also acquires remote sensing observation data of the satellite in orbit to evaluate the execution of remote sensing observation missions, which serves as the data basis for mission optimization and supplementation. Embodied intelligence is realized at the satellite end, improving the intelligence level of the satellite in orbit.

[0029] According to the embodiments of this application, an embodiment of a method for applying a large-scale embodied intelligent satellite model that integrates autonomous perception, decision-making and execution is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0030] Reference Figure 1 As shown, this embodiment provides a method for applying an embodied intelligent satellite large-scale model that integrates autonomous perception, decision-making, and execution. The embodied intelligent satellite large-scale model includes a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application intelligent agent module. The method includes: S100. Obtain the remote sensing observation task through the cognitive reasoning module and generate the corresponding first state perception command; obtain the first state data of the on-orbit satellite before it performs the remote sensing observation task based on the first state perception command through the multimodal perception module.

[0031] S200. The cognitive reasoning module uses the first state data to conduct a feasibility analysis on whether the on-orbit satellite is suitable for performing remote sensing observation tasks, and obtains the feasibility data of the on-orbit satellite before performing remote sensing observation tasks.

[0032] S300. Through the autonomous planning module, feasibility data is used to perform mission planning for in-orbit satellites, generating action plans for in-orbit satellites to perform remote sensing observation tasks.

[0033] S400. The execution feedback module controls the on-orbit satellite to perform remote sensing observation tasks according to the execution action plan; it acquires and feeds back remote sensing observation data of the remote sensing observation tasks for use in performance evaluation.

[0034] The application agent module contains multiple agents with specific functions, which can be called by other modules in the embodied intelligent satellite model.

[0035] Reference Figure 2 As shown, this embodiment deploys an embodied intelligent satellite model at the satellite end for autonomous management and decision-making of one or more on-orbit satellites. The embodied intelligent satellite model includes modules such as a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application intelligent agent module. These modules are sequentially connected and connected to the application intelligent agent module to invoke corresponding intelligent agents from it.

[0036] The cognitive reasoning module can incorporate a large language model, enabling language comprehension. It receives remote sensing mission commands from the ground station to the satellite, acquires and parses the remote sensing observation tasks indicated by these commands, thus initiating the state awareness process. Furthermore, after receiving multimodal state data from the multimodal perception module, the cognitive reasoning module performs mission feasibility analysis on each on-orbit satellite based on this data, serving as the data foundation for autonomous mission planning. In some embodiments, the cognitive reasoning module can also detect anomalies in on-orbit satellites based on the results of performance evaluations, identifying the reasons why an on-orbit satellite cannot continue performing remote sensing observation tasks, and using this information for action replanning.

[0037] The multimodal perception module communicates with the on-orbit satellites and responds to state perception commands sent by the cognitive reasoning module. It acquires multimodal state data from each on-orbit satellite and feeds it back to the cognitive reasoning module, enabling the embodied intelligent satellite model to understand the state of the on-orbit satellites and their environment. For example, the multimodal state data may include multi-source remote sensing data such as optical, infrared, and radar data, as well as internal state information such as energy status and thermal control status.

[0038] The autonomous planning module is used to perform mission planning for all in-orbit satellites based on feasibility data, generating execution action plans for each in-orbit satellite to perform remote sensing observation tasks. It can be understood that execution action plans can include single-satellite actions such as attitude adjustment and remote sensing data acquisition for any single in-orbit satellite, as well as multi-satellite actions such as resource scheduling and collaborative work among satellites.

[0039] The execution feedback module receives the execution action plans generated by the autonomous planning module and distributes them to each in-orbit satellite to control the satellites in performing remote sensing observation tasks. The execution feedback module can also acquire and provide feedback on the remote sensing observation data from each in-orbit satellite during task execution and after the execution action plans are completed. This allows the embodied intelligent satellite large model to analyze the execution status of the remote sensing observation data and determine whether the data meets expectations. It is understood that the remote sensing observation data can be remote sensing image data acquired by the in-orbit satellites, or it can be the result of image processing and calculation based on remote sensing image data.

[0040] In some embodiments, if the remote sensing observation data does not meet expectations, the execution feedback module can send the situation to the cognitive reasoning module, which will then reason and correct the remote sensing observation data and update the decision through the autonomous planning module. This enables the on-orbit satellite to have real-time response capabilities and improves the autonomy of mission decision-making.

[0041] Reference Figure 3As shown, the application intelligent agent module contains multiple intelligent agents with specific functions, including at least an evaluation and diagnosis intelligent agent, a fault repair intelligent agent, a task planning intelligent agent, a resource scheduling intelligent agent, and an image computing intelligent agent. Different intelligent agents are connected to other modules through standardized interfaces so that they can be called by other modules in the embodied intelligent satellite big model.

[0042] The system comprises several intelligent modules: an assessment and diagnosis agent connected to the multimodal perception and cognitive reasoning modules; an assessment agent monitoring and diagnosing anomalies in orbit based on multimodal state data from the multimodal perception module; and a fault repair agent repairing anomalies and rescheduling resources for the embodied intelligent satellite model or existing satellites based on anomaly diagnosis results from the assessment and diagnosis agent, enabling rapid handling of satellite-side anomalies; a mission planning agent connected to the autonomous planning module; a mission planning agent generating execution action plans based on feasibility data from each orbiting satellite; a resource scheduling agent connected to each orbiting satellite; and an image processing agent coordinating resource allocation among them based on the execution action plans, including energy, power, thermal control, and attitude control, generating resource utilization margin assessment reports. Finally, an image processing agent connected to the execution feedback module receives remote sensing data from each orbiting satellite and performs image processing and calculations to obtain remote sensing data for the mission.

[0043] In some embodiments, the evaluation and diagnostic agent can also be used to perform performance analysis on remote sensing observation data, determine whether the remote sensing observation data meets expectations, and notify the cognitive reasoning module if the remote sensing observation data does not meet expectations, so as to trigger replanning for the execution action planning.

[0044] Specifically, in this embodiment, the aforementioned embodied intelligent satellite model is deployed on the satellite. The user sends a task instruction for the remote sensing observation mission to the satellite via a ground station. The cognitive reasoning module receives this task instruction, performs task understanding, and thus identifies the user's observation intention and obtains the task elements of the remote sensing observation mission. It is understood that the task elements of a remote sensing observation mission may include elements such as the target area, imaging requirements, mission timeliness requirements, and priority information. Imaging requirements may include imaging mode and resolution requirements. When the imaging requirements of the remote sensing observation mission are not explicitly given in the task instruction, the cognitive reasoning module can autonomously reason using a large language model combined with a model knowledge base and historical mission data to generate imaging requirements suitable for the remote sensing observation mission.

[0045] Furthermore, after obtaining the mission elements of the remote sensing observation task, the cognitive reasoning module generates corresponding first state perception commands based on these mission elements to assess whether the on-orbit satellite can meet the mission elements requirements for the remote sensing observation task. The cognitive reasoning module sends the first state perception commands to the multimodal perception module, which, in response, enables various sensors on the on-orbit satellite to acquire the current first state data of the on-orbit satellite. It is understood that the first state data may include the on-orbit satellite's energy status, attitude and orbit information, payload operating status, fault and abnormal states, and time windows for the target area. In some embodiments, the first state data may also include the external conditions of the on-orbit satellite's environment, including weather conditions and lighting conditions in the target area.

[0046] Furthermore, the multimodal sensing module feeds back the obtained first-state data to the cognitive reasoning module. The cognitive reasoning module uses this first-state data to assess the mission feasibility of the on-orbit satellites, determining whether each on-orbit satellite is suitable for performing remote sensing observation tasks, thus obtaining feasibility data for each on-orbit satellite for the remote sensing observation task. It is understood that feasibility data can include energy feasibility, payload feasibility, fault feasibility, time window feasibility, and external condition feasibility. Energy feasibility can be obtained based on the energy status of the on-orbit satellite, indicating whether the satellite's energy reserves can support its performance of the remote sensing observation task. Payload feasibility can be obtained based on the operational status of the on-orbit satellite's payload, indicating whether the payload equipment on the satellite can function normally and be used to perform the remote sensing observation task. Fault feasibility can be obtained based on the fault or abnormal status of the on-orbit satellite, indicating whether there are any faults affecting the remote sensing observation task. Time window feasibility can be obtained based on the time window of the on-orbit satellite for the target area, indicating whether the on-orbit satellite has the time conditions to observe the target area. External condition feasibility can be obtained based on the external conditions of the environment in which the on-orbit satellite is located, indicating whether the target area is currently suitable for remote sensing observation.

[0047] Furthermore, after obtaining feasibility data for each on-orbit satellite, the feasibility data is sent to the autonomous planning module. The autonomous planning module calls the mission planning agent to determine the on-orbit satellites capable of performing the remote sensing observation mission based on the feasibility data. For these on-orbit satellites capable of performing the remote sensing observation mission, mission planning is performed based on the feasibility data to determine the observation actions that each on-orbit satellite needs to perform, resulting in an execution action plan. It is understood that the execution action plan may include information such as the target satellites actually performing the remote sensing observation mission, the time point and duration of image acquisition by the target satellites, the payloads to be activated and their parameter configurations, the attitude and trajectory adjustment scheme of the target satellites, and multiple key checkpoints during mission execution. When there are multiple on-orbit satellites forming an on-orbit constellation, the execution action plan may include the individual action execution tables for each target satellite to control collaborative work among them. In some embodiments, during mission planning, the autonomous planning module may also call the resource scheduling agent to perform resource scheduling for the on-orbit satellites, allocating the remaining resources at the satellite end primarily to the satellites actually performing the remote sensing observation mission, ensuring that these satellites can operate normally.

[0048] Furthermore, the autonomous planning module sends the action plan to the execution feedback module, which then distributes the action execution table from the action plan to the corresponding on-orbit satellites. This enables the on-orbit satellites to perform observation actions at the appropriate time, obtaining remote sensing observation data for the remote sensing observation mission, and deriving remote sensing observation data for the mission based on the remote sensing observation data. In some embodiments, multiple satellite agents can be deployed on the on-orbit satellites, including but not limited to attitude control agents and mission execution agents. The attitude control agent can be used to precisely control the attitude and orientation of the on-orbit satellite and stabilize its attitude at the imaging position when the satellite enters the imaging orbit segment. The mission execution agent can be used to invoke one or more imaging sensors on the on-orbit satellite for image acquisition, including but not limited to visible light cameras, infrared imagers, and synthetic aperture radar. It is understood that during mission execution, the evaluation and diagnostic agent monitors the status of the on-orbit satellites in real time and reports any abnormalities in any on-orbit satellite in a timely manner, thereby ensuring the effective execution of the remote sensing observation mission.

[0049] In some embodiments, after obtaining the remote sensing observation data for a remote sensing observation mission, an evaluation and diagnostic agent can be invoked to perform an execution analysis on the remote sensing observation data to determine whether the data meets expectations. The execution analysis may include area pointing accuracy analysis, image quality analysis, and payload operational status analysis. Area pointing accuracy analysis can be used to determine whether the accuracy of the on-orbit satellite pointing to the target area meets the requirements. Image quality analysis can be used to determine whether the image data acquired by the on-orbit satellite is clear and corresponds to appropriate exposure conditions, meeting imaging requirements. Payload operational status analysis can be used to determine whether each payload on the on-orbit satellite is operating normally, ensuring the validity of the remote sensing observation data. If any result of the execution analysis indicates that the remote sensing observation data does not meet expectations, this situation is notified to the cognitive reasoning module to trigger a replanning of the execution action plan, ensuring that the remote sensing observation data meets the expected requirements.

[0050] This embodiment provides a method for applying an embodied intelligent satellite large-scale model that integrates autonomous perception, decision-making, and execution. The embodied intelligent satellite large-scale model is deployed on the satellite, including a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application intelligent agent module. After a remote sensing observation task is issued, the cognitive reasoning module obtains the task and instructs the multimodal perception module to acquire the first state data of the on-orbit satellite. The cognitive reasoning module performs a task feasibility analysis based on the first state data, serving as the data basis for the autonomous planning module to plan the task and obtain the execution action plan for the on-orbit satellite. The execution feedback module controls the on-orbit satellite to execute the remote sensing observation task according to the execution action plan and acquires remote sensing observation data for performance evaluation. During application, these modules can call upon corresponding intelligent agents in the application intelligent agent module to collaborate through multiple intelligent agents.

[0051] Compared with related technologies, this application utilizes multiple functional modules in the embodied intelligent satellite model to call intelligent agents separately. Through the collaborative work between multiple intelligent agents, the satellite can achieve autonomous state perception, mission planning, and execution feedback, thereby reducing the dependence of the on-orbit satellite on the ground station and improving the autonomous operation capability of the on-orbit satellite.

[0052] In addition, the satellite also acquires multimodal state data of the satellite in orbit through intelligent agents as the data basis for mission planning. It also acquires remote sensing observation data of the satellite in orbit to evaluate the execution of remote sensing observation missions, which serves as the data basis for mission optimization and supplementation. Embodied intelligence is realized at the satellite end, improving the intelligence level of the satellite in orbit.

[0053] Reference Figure 4As shown, in one embodiment of this application, the first state data includes first internal state data and first external environment data of the satellite before it performs a remote sensing observation mission; the cognitive reasoning module uses the first state data to perform a feasibility analysis on whether the satellite is suitable for performing a remote sensing observation mission, obtaining feasibility data of the satellite before it performs a remote sensing observation mission, including: S210. Based on the first internal state data, perform internal state analysis on the on-orbit satellite to obtain the internal state assessment results of the on-orbit satellite before it performs remote sensing observation missions.

[0054] S220. Based on the first external environment data, perform external environment analysis on the on-orbit satellite to obtain the external environment assessment results before the on-orbit satellite performs remote sensing observation missions.

[0055] S230. Obtain feasibility data based on the results of internal status assessment and external environment assessment.

[0056] The first state data includes first internal state data and first external environment data of the satellite before it performs remote sensing observation missions. The first internal state data can be the internal state data of the satellite in orbit, used to describe the system health status and mission execution capabilities of the satellite, including but not limited to the satellite's energy status, attitude and orbit information, payload operating status, fault and abnormal status, and time window for the target area. The first external environment data can be the external conditions of the environment in which the satellite is located, used to describe the degree of influence of the target area's environment on the remote sensing observation mission, including the weather conditions and lighting conditions of the target area.

[0057] Specifically, the first internal state data of each on-orbit satellite is obtained. Based on the mission elements of the remote sensing observation mission and the first internal state data, the internal state analysis of each on-orbit satellite is performed to determine whether each on-orbit satellite has the ability to perform the remote sensing observation mission and the ability to fully perform the corresponding part of the remote sensing observation mission, thereby obtaining the internal state assessment results of the on-orbit satellite before performing the remote sensing observation mission.

[0058] Furthermore, initial external environment data for each in-orbit satellite is acquired. Based on the mission elements of the remote sensing observation task and the initial external environment data, an external environment analysis is performed on each in-orbit satellite to determine whether its external environment is suitable for performing the remote sensing observation task. This yields an external environment assessment result before the in-orbit satellites perform the remote sensing observation task. It is understood that the external environment analysis includes space environment analysis and target area environment analysis. Space environment analysis can be used to determine whether the space environment in which the in-orbit satellites are located will affect the execution process of the remote sensing observation task; target area environment analysis can be used to determine whether the target area is suitable for remote sensing observation within the time window for performing the remote sensing observation task.

[0059] Furthermore, based on the internal state assessment results and the external environment assessment results, feasibility data is obtained. If both the internal state assessment results and the external environment assessment results of any on-orbit satellite meet the requirements of the remote sensing observation mission, then that on-orbit satellite is determined to be capable of performing the remote sensing observation mission and can be selected as the target satellite. If either the internal state assessment result or the external environment assessment result of any on-orbit satellite does not meet the requirements of the remote sensing observation mission, then that on-orbit satellite is determined to be unable to perform the remote sensing observation mission and will not be considered during the mission planning process. It is understood that the feasibility data also includes first internal state data and first external environment data, which serve as the data basis for the autonomous planning module's mission planning.

[0060] Reference Figure 5 As shown, in one embodiment of this application, the first internal state data includes operational state data and execution capability data of the satellite before it performs a remote sensing observation mission; the internal state analysis of the satellite before it performs the remote sensing observation mission is performed based on the first internal state data to obtain the internal state assessment result of the satellite before it performs the remote sensing observation mission, including: S212. Based on the operational status data, perform a health status analysis on the satellites in orbit to obtain the health status assessment results of the satellites in orbit before they perform remote sensing observation missions.

[0061] S214. Evaluate the execution conditions of the execution capability data and the mission standard data of the remote sensing observation mission to obtain the execution condition evaluation results before the on-orbit satellite performs the remote sensing observation mission; wherein, the mission standard data represents the requirements of the on-orbit satellite in terms of operational resources and observation capabilities for performing the remote sensing observation mission.

[0062] S216. The internal status assessment results are obtained based on the health status assessment results and the performance condition assessment results.

[0063] The first set of internal status data includes operational status data and execution capability data of the satellite in orbit before it performs remote sensing observation missions. Operational status data represents the actual working status of the satellite in orbit, used to determine whether the satellite is in a healthy state. This data may include the satellite's energy status, payload operational status, and fault / abnormal status. Execution capability data represents the satellite's ability to perform remote sensing observation missions, used to determine whether the satellite can complete the missions and obtain the expected results. This data may include the satellite's attitude and orbit information, observation capability information, and time windows for the target area. The mission standard data for the remote sensing observation mission may be the remote sensing capability requirements of the satellite in orbit, including the required imaging modes and resolution requirements.

[0064] Specifically, for any given satellite in orbit, a health status analysis is performed based on its operational status data to obtain a health status assessment result before it performs a remote sensing observation mission. It is understood that the health status analysis may include energy feasibility analysis, payload feasibility analysis, and fault feasibility analysis. Energy feasibility analysis determines whether the satellite's energy reserves are sufficient to support the remote sensing observation mission; payload feasibility analysis determines whether the satellite's payload equipment can function properly and perform the mission; and fault feasibility analysis determines whether the satellite has any faults that could affect the remote sensing observation mission.

[0065] In some embodiments, health status analysis can be performed based on a satellite situation map of all on-orbit satellites. This situation map represents the current status of each on-orbit satellite, including operational status data, attitude and orbit information, and interrelationships with other on-orbit satellites. The satellite situation map can be updated in real-time based on the operational status of each on-orbit satellite, enabling the embodied intelligent satellite model to grasp the real-time status of each satellite. This allows for mission planning based on the real-time status of the on-orbit satellites, achieving autonomous perception and intelligent decision-making by the embodied intelligent satellite model and improving the intelligence level of the on-orbit satellites.

[0066] Furthermore, based on the mission elements of the remote sensing observation mission, mission standard data is obtained. Then, based on the mission standard data and the execution capability data of any given in-orbit satellite, an execution condition assessment is performed on that satellite to obtain the execution condition assessment results before it performs the remote sensing observation mission. It is understood that the execution condition assessment may include capability matching analysis and time window feasibility analysis. Capability matching analysis can be used to determine whether the satellite possesses the sensors to perform the remote sensing observation mission, or whether the accuracy of the onboard sensors meets the requirements of the mission. Time window feasibility analysis can be used to determine whether the satellite has the time conditions to observe the target area.

[0067] In some embodiments, capability matching analysis may be performed based on satellite capability maps of all in-orbit satellites. A satellite capability map can be a multi-level visualization representing the capability range, performance boundaries, and constraints of in-orbit satellites for performing remote sensing observation missions. A search and matching process is performed on the satellite capability map based on the mission standard data for the remote sensing observation mission to determine the degree of matching between each in-orbit satellite and the mission standard data, thus obtaining the execution condition assessment results.

[0068] Reference Figure 6As shown, in one embodiment of this application, the autonomous planning module utilizes feasibility data to perform mission planning for the on-orbit satellite, generating an execution action plan for the on-orbit satellite for the remote sensing observation mission, including: S310. Based on feasibility data, perform motion planning for the in-orbit satellite to obtain the initial motion plan for the in-orbit satellite.

[0069] S320. Based on the initial action plan, optimize the action scheduling of the satellite in orbit to obtain the execution action plan for performing remote sensing observation tasks.

[0070] Specifically, based on feasibility data, including the first internal state data and the first external environment data, action planning is performed on all in-orbit satellites to obtain the initial action plan for each in-orbit satellite. The action planning process may include: decomposing the remote sensing observation task to obtain multiple sub-tasks and their corresponding timetables; determining feasible satellites capable of performing each sub-task based on the task elements of each sub-task; performing task planning on the corresponding feasible satellites within the time period of any sub-task to obtain the sub-task action plan; and arranging and integrating the sub-task action plans of all sub-tasks according to the timetable to obtain the initial action plan.

[0071] Furthermore, based on the initial action plan, the action scheduling of each in-orbit satellite is optimized to maximize the execution benefits of the remote sensing observation mission and minimize the energy consumption caused by executing the remote sensing observation mission, resulting in an execution action plan for performing the remote sensing observation mission. In some embodiments, the method for optimizing the action scheduling may include simulated annealing algorithm, particle swarm optimization, greedy algorithm, heuristic genetic algorithm, and reinforcement learning, etc. The object of the action scheduling optimization may be the target satellites that actually perform remote sensing observation missions in the initial action plan, the observation actions performed by each target satellite, and the execution time and order of the observation actions.

[0072] Reference Figure 7 As shown, in one embodiment of this application, the application agent module includes an evaluation and diagnostic agent, and the remote sensing observation data includes intermediate observation data obtained by executing some actions in the above-described action planning; the method further includes: S510. During the execution of remote sensing observation tasks, the intermediate observation data are evaluated by an assessment and diagnostic agent to obtain intermediate evaluation results.

[0073] S520. If the intermediate evaluation results do not meet the preset quality conditions, the abnormal reasons for the intermediate evaluation results not meeting the preset quality conditions shall be determined by the evaluation and diagnostic agent.

[0074] S530. Generates a second state perception command corresponding to the cause of the anomaly through the cognitive reasoning module, so as to optimize and adjust the execution action planning of the on-orbit satellite during the execution of the remote sensing observation mission and eliminate the impact of the anomaly on the remote sensing observation mission.

[0075] Specifically, at any intermediate moment in the execution process, after each on-orbit satellite performs a portion of the actions planned in the aforementioned execution action plan, it feeds back the remote sensing data acquired for that portion of the action to the execution feedback module as intermediate observation data. Upon receiving the intermediate observation data, the execution feedback module invokes the evaluation and diagnostic agent to evaluate the execution status of the intermediate observation data, thereby determining the observation quality of the intermediate observation data and obtaining intermediate evaluation results.

[0076] In some embodiments, performance evaluation may include image quality evaluation and mission progress evaluation, wherein image quality evaluation may assess the quality of intermediate observation data to determine whether the intermediate observation data meets the quality standards of the remote sensing observation mission. Image quality evaluation may include image quality assessment, image occlusion assessment, and recognition confidence assessment, wherein image quality evaluation may assess the image resolution of intermediate observation data to determine the image acquisition quality of the on-orbit satellite. Image occlusion assessment may assess the cloud occlusion level in intermediate observation data to determine the degree of remote sensing acquisition of the target area by the on-orbit satellite. Recognition confidence assessment may perform target recognition based on mission execution results to obtain the recognition confidence of the observed targets in the target area, thereby determining the recognition capability of the intermediate observation data for the observed targets.

[0077] Task progress assessment can be used for lifecycle management of remote sensing observation tasks, determining metrics such as time progress and task progress at any given intermediate point in time. Time progress can be obtained by comparing the current time taken to the planned time for intermediate observation data, indicating whether there are any time lags in the execution of the remote sensing observation task. Task progress can be obtained by comparing intermediate observation data with the total workload, indicating the completion status of the remote sensing observation task.

[0078] In other embodiments, the performance evaluation may also include satellite status evaluation, which may include: real-time status monitoring of each on-orbit satellite to obtain status evaluation results and determine whether each on-orbit satellite and its payload are operating within normal ranges. It is understood that real-time status monitoring can determine the status boundaries of each on-orbit satellite, thereby providing early warning of potential failures before any on-orbit satellite malfunctions, reducing the lifespan loss of on-orbit satellites, and ensuring their survival and safety.

[0079] Furthermore, if any intermediate evaluation result fails to meet the preset quality conditions, an evaluation diagnostic agent performs anomaly analysis on the intermediate evaluation result to determine the cause of the anomaly. For example, if the intermediate evaluation result indicates that the image quality of the intermediate observation data does not meet the quality standard, it suggests that the on-orbit satellite's sensors may be experiencing performance degradation or malfunction, or that the on-orbit satellite's attitude control system may be experiencing jitter.

[0080] Furthermore, the assessment and diagnostic agent sends the identified causes of anomalies to the cognitive reasoning module. The cognitive reasoning module generates corresponding second-state perception instructions based on the causes of anomalies to assess whether the on-orbit satellite is suitable to continue performing remote sensing observation tasks. Based on the state perception results, it optimizes and adjusts the on-orbit satellite's execution action plan to eliminate the impact of the anomalies on the remote sensing observation task. It is understandable that the above process can be repeated during the execution of the remote sensing observation task to respond promptly to changes in the satellite's state and trigger timely replanning of the execution action plan, ensuring that the remote sensing observation data meets the expected requirements.

[0081] Reference Figure 8 As shown, in one embodiment of this application, a second state perception command corresponding to the cause of the anomaly is generated through a cognitive reasoning module to optimize and adjust the execution action planning of the on-orbit satellite during the execution of a remote sensing observation mission, including: S532. The second state data of the on-orbit satellite during the remote sensing observation mission is obtained by the multimodal sensing module based on the second state sensing command.

[0082] S534. The cognitive reasoning module uses the second state data to determine whether the on-orbit satellite is suitable for performing remote sensing observation tasks, and obtains the standard matching data of the on-orbit satellite in the process of performing remote sensing observation tasks.

[0083] S536. The autonomous planning module optimizes and adjusts the execution action planning of the on-orbit satellite using standard matching data to obtain the optimized action plan of the on-orbit satellite.

[0084] Specifically, the cognitive reasoning module sends the second state perception command to the multimodal perception module. In response to the second state perception command, the multimodal perception module enables various sensors on the satellite in orbit to acquire the current second state data of the satellite in orbit and feed it back to the cognitive reasoning module.

[0085] Furthermore, the cognitive reasoning module uses the second-state data to assess the mission feasibility of the on-orbit satellites, determining whether each satellite is suitable to continue performing remote sensing observation missions, and obtaining standard matching data for the on-orbit satellites during the execution of remote sensing observation missions. It can be understood that standard matching data represents the degree of matching between the on-orbit satellites and the mission requirements of the remote sensing observation mission at the current intermediate moment, and can be used as a data basis for optimizing and adjusting the execution action planning.

[0086] Furthermore, the cognitive reasoning module sends the standard matching data to the autonomous planning module. Upon receiving the standard matching data, the autonomous planning module optimizes and adjusts the execution action plan of the on-orbit satellite using this data, resulting in an optimized action plan. For example, if the standard matching data indicates that the on-orbit satellite can eliminate the cause of the anomaly through simple fault repair, it means that the on-orbit satellite is still suitable to continue performing remote sensing observation tasks, and its corresponding execution action plan does not need to be optimized or adjusted. If the standard matching data indicates that the on-orbit satellite's resource reserves can complete the allocated sub-tasks, but its observation capabilities cannot meet the requirements of the remote sensing observation task, and the quality requirements of the remote sensing observation task can be adjusted, the quality requirements can be downgraded, and the execution action plan of the on-orbit satellite can be optimized and adjusted based on the downgraded quality requirements. If the standard matching data indicates that both the on-orbit satellite's resource reserves and observation capabilities cannot meet the requirements of the remote sensing observation task, it means that the on-orbit satellite is not suitable to continue performing the remote sensing observation task, and the task planning is re-performed using other on-orbit satellites as the target, updating the execution action plan.

[0087] Reference Figure 9 As shown, in one embodiment of this application, the application agent module includes an evaluation and diagnostic agent; the method further includes: S540. After completing the execution action planning, the evaluation and diagnostic agent evaluates the execution status of the complete execution of the remote sensing observation task to obtain the completion evaluation result.

[0088] S550. If the evaluation results indicate that the complete execution result does not meet the preset quality conditions, the cognitive reasoning module performs supplementary reasoning on the complete execution result to generate suggestions for improving the complete execution result.

[0089] The complete execution results may include remote sensing imaging data fed back by each on-orbit satellite, as well as image recognition information obtained after image processing and target recognition based on the above observation data.

[0090] Specifically, after completing the execution action planning and obtaining the complete execution result, the embodied intelligent satellite big model calls the evaluation and diagnostic agent to evaluate the execution status of the obtained complete execution result, thus obtaining a completion evaluation result. In some embodiments, the execution status evaluation process for the complete execution result may include imaging integrity evaluation and imaging quality evaluation, wherein imaging integrity evaluation may assess the coverage of the target area by the complete execution result to determine whether the complete execution result contains complete image information of the target area. The imaging integrity evaluation method may be the same as the image occlusion evaluation, which determines the coverage of the target area by evaluating the cloud occlusion level in the complete execution result. When the cloud occlusion in the complete execution result is relatively severe, it is determined that the complete execution result does not meet the preset quality conditions, and the areas with severe cloud occlusion are recorded and marked for use in generating supplementary tasks. Imaging quality evaluation may assess and analyze the image quality of the complete execution result, which may include the image sharpness and spectral information of the complete execution result, etc. If any of these indicators does not meet the expected standard, it is determined that the complete execution result does not meet the preset quality conditions.

[0091] In some embodiments, the complete execution result may further include infrared imagery. In this case, the evaluation process for the execution status of the complete execution result may also include thermal distribution anomaly assessment. Thermal distribution anomaly assessment may involve performing thermal distribution analysis on the infrared imagery to determine the thermal distribution data of the target area and comparing it with standard thermal distribution data to determine whether the temperature distribution in the target area is abnormal and whether further monitoring is required. It is understood that if there is a significant difference between the thermal distribution data of the target area and the standard thermal distribution data, it can be determined that the complete execution result does not meet the preset quality conditions, and supplementary tasks need to be generated to improve the complete execution result.

[0092] Furthermore, if the evaluation result indicates that the complete execution result does not meet the preset quality conditions, the complete execution result and the evaluation result are sent to the cognitive reasoning module. The large language model in the cognitive reasoning module analyzes and supplements the complete execution result and evaluation result to determine a second anomaly that causes the complete execution result to fail to meet the preset quality conditions. Based on this second anomaly, a suggested improvement task is generated. It is understood that after obtaining the improved task suggestion, it can be added to the satellite's task list. The embodied intelligent satellite large model then generates supplementary tasks based on the improved task suggestion, and the on-orbit satellite performs task planning based on these supplementary tasks to improve and optimize the complete execution result. In other embodiments, the improved task suggestion can also be sent to the ground station, where the user generates supplementary tasks to improve the complete execution result based on the improved task suggestion.

[0093] In some embodiments, the process of planning missions for on-orbit satellites according to supplementary tasks may include: generating corresponding third state perception instructions based on a second anomaly cause through a cognitive reasoning module, so as to perform state perception on the state of each on-orbit satellite after completing the remote sensing observation task through a multimodal perception module, and obtaining third state data of each on-orbit satellite; performing a feasibility analysis on whether the on-orbit satellite can perform supplementary observation tasks through the cognitive reasoning module using the third state data, and obtaining supplementary feasibility data of the on-orbit satellite after completing the remote sensing observation task; sending the supplementary feasibility data to an autonomous planning module, and performing supplementary mission planning for the on-orbit satellites through the autonomous planning module using the supplementary feasibility data, and generating supplementary action plans for the on-orbit satellites.

[0094] Reference Figure 10 As shown in the figure, as one embodiment of this application, the embodied intelligent satellite large model also includes an on-orbit computing module, and the application intelligent agent module includes an image computing intelligent agent; the method also includes: S610. For the target area specified in the remote sensing observation mission, remote sensing imaging data of the target area is obtained through the on-orbit computing module.

[0095] S620. Based on the remote sensing observation mission, the image computing agent is called by the on-orbit computing module to perform image recognition and information annotation on the remote sensing imaging data, so as to obtain image recognition information about the remote sensing observation mission, which is used to generate the remote sensing observation results of the remote sensing observation mission.

[0096] Specifically, after the on-orbit satellite completes its action planning and obtains remote sensing imaging data of the target area, it sends the remote sensing imaging data and the remote sensing observation task to the image computing agent. The image computing agent performs image recognition on the remote sensing imaging data according to the remote sensing observation task, obtains the target object pointed to by the remote sensing observation task, and annotates the remote sensing imaging data with information based on the target object recognition results, thus obtaining image recognition information about the remote sensing observation task. It can be understood that remote sensing imaging data and / or image recognition information can serve as the remote sensing observation result of the remote sensing observation task, and the remote sensing observation result can include remote sensing imaging data, image recognition information, or a combination of remote sensing imaging data and image recognition information.

[0097] Reference Figure 11As shown, in practice, the ground station receives monitoring requests from various scenarios, including monitoring traffic congestion in Hangzhou, ongoing fires, and illegal maritime activities. These monitoring requests are then treated as remote sensing observation tasks and transmitted to the embodied intelligent satellite model via the space-based operating system. Within the embodied intelligent satellite model, a cognitive reasoning module analyzes the remote sensing observation tasks received from the space-based operating system, understands the specific requirements of the tasks, and generates corresponding first-state perception commands. Based on these requirements, a multimodal perception module sends these first-state perception commands to the satellite subsystems within the on-orbit satellite to obtain first-state data before the satellite can execute the remote sensing observation task. The multimodal perception module feeds this first-state data back to the cognitive reasoning module, which uses this data to conduct a feasibility analysis of the on-orbit satellite's suitability for the remote sensing observation task, obtaining feasibility data for the satellite before execution. Finally, an autonomous planning module uses the feasibility data and the remote sensing observation task to plan the on-orbit satellite's actions for the task, generating an execution plan that includes the execution steps and resource requirements. The embodied intelligent satellite model also decomposes the execution action plan through a task distributor, obtaining an individual action execution list for each in-orbit satellite, and sending the action execution list to the corresponding in-orbit satellite. For any given in-orbit satellite, the execution feedback module controls that satellite to perform a remote sensing observation task according to the action execution list, acquiring and feeding back remote sensing observation data for performance evaluation. This enables real-time monitoring and feedback of the remote sensing observation task execution process, ensuring that the remote sensing observation task is carried out as planned.

[0098] Reference Figure 12 As shown, in some embodiments, the satellite subsystem in an on-orbit satellite may include a state awareness module, a resource assessment module, a system operation and maintenance module, and a multi-agent module. The state awareness module is connected to the resource assessment module and the system operation and maintenance module, respectively. The resource assessment module and the system operation and maintenance module are each connected to the multi-agent module. The resource assessment module and the system operation and maintenance module are also connected to an embodied intelligent satellite large model at the satellite end. The state awareness module can connect to various sensors within the on-orbit satellite and acquire the satellite's state data by calling the corresponding types of sensors. For example, the state data may include energy and thermal control state data, historical data, application scheduling data, environmental and orbital parameters, telemetry and remote control data, and payload logs, etc.

[0099] The resource assessment module can be used to perform health status assessments based on the status data of on-orbit satellites, ensuring that the satellites are in normal operating condition. Understandably, if the on-orbit satellite is determined to be in normal operating condition, the resource assessment module can also feed the status data of the on-orbit satellite back to the embodied intelligent satellite big data model for mission planning. For example, the health status assessment may include energy system assessment and thermal control system assessment. The energy system assessment may include evaluation items such as charging efficiency, battery health, and load balancing, while the thermal control system assessment may include evaluation items such as temperature stability and heater efficiency.

[0100] The system operation and maintenance module can be used for the operation and maintenance management and fault diagnosis of satellite systems in orbit, promptly identifying and resolving potential problems to ensure the stable operation of the satellite system. Upon receiving the execution action plan, the system operation and maintenance module can also optimize resource allocation for in-orbit satellites based on the execution action plan, pre-allocating resource reserves among in-orbit satellites, improving resource utilization, and shortening the total execution time of the execution action plan, thereby effectively improving the execution efficiency of remote sensing observation missions. Furthermore, the system operation and maintenance module can also provide feedback on the fault status of in-orbit satellites to the embodied intelligent satellite model, enabling the embodied intelligent satellite model to perform mission planning based on the fault status of each in-orbit satellite.

[0101] The multi-agent module contains multiple agents with specific functions, which can be invoked by other modules in the on-orbit satellite to implement their respective functions. For example, the multi-agent module may include a resource optimization agent. After receiving the execution action plan, the satellite subsystem invokes the resource optimization agent to optimize resource allocation within the satellite subsystem according to the execution action plan, resulting in a mission execution plan for the satellite subsystem. It should be noted that the resource allocation optimization process may include adjusting the order of subtasks or the allocated satellite resources in the execution action plan, so that the execution order of the subtasks matches the current resource status of the satellite subsystem. This reduces redundant actions used to ensure satellite resources during remote sensing observation missions, such as charging the satellite subsystem's energy system. It can be understood that by optimizing resource allocation within the satellite subsystem, the utilization rate of existing satellite resources within the satellite subsystem is improved, the total execution time of the mission execution plan is shortened, thereby effectively improving the execution efficiency of remote sensing observation missions.

[0102] In other embodiments, the image computation agent can be a visual model with image analysis capabilities, used to process and analyze remote sensing observation data according to the task requirements of the remote sensing observation mission, and obtain the remote sensing observation results required by the user. It is understood that the image computation agent can be deployed synchronously at both the satellite and ground stations, used for image computation and analysis at the satellite end. The ground station collaboratively evaluates the synchronously deployed image computation agent based on multiple feedback remote sensing observation results, and optimizes the model of the image computation agent if it meets the model optimization conditions, thereby updating the image computation agent. After updating the image computation agent, the ground station synchronizes the updated image computation agent to the satellite in the next communication time window, thereby achieving timely updates of the image computation agent.

[0103] Reference Figure 13 As shown in the embodiment of this application, when the remote sensing observation task is to map urban congestion, the image recognition information includes urban road information and vehicle identification information of the target area, and the remote sensing observation results include congestion data of the target area; the remote sensing observation results are obtained in the following ways: S622. Use an image computing agent to segment roads in remote sensing imaging data to obtain urban road information; based on the remote sensing imaging data, use an image computing agent to identify vehicles on urban roads represented by the urban road information to obtain vehicle identification information.

[0104] S624. Generate congestion data based on urban road information and vehicle identification information.

[0105] Reference Figure 14 As shown, in the case of a remote sensing observation task involving urban congestion mapping, taking Hangzhou city as an example, the ground station sends a task instruction regarding urban congestion mapping to the satellite. The specific content of the task instruction could be "to perform multimodal imaging and collect data of Hangzhou city." After receiving the task instruction, the embodied intelligent satellite model at the satellite end first uses a cognitive reasoning module to understand the task instruction, determine that the target area is Hangzhou, and obtain the task priority and other task elements for this remote sensing observation task. In some embodiments, the cognitive reasoning module can connect to an on-board vector database, which stores urban vector data for multiple regions, representing the geographical boundaries of each region. By calling the on-board vector database, the cognitive reasoning module determines the latitude and longitude range of the target area, obtains the geographical boundaries of the target area, and thus determines the area requiring remote sensing observation.

[0106] Secondly, the cognitive reasoning module generates a first-state perception command based on the task understanding results and sends it to the multimodal perception module. The multimodal perception module then performs environmental and resource perception on the on-orbit satellites, acquiring first-state data for each satellite, including battery level, camera status, and weather conditions in the target area. Based on this first-state data, the cognitive reasoning module performs a feasibility analysis, obtaining feasibility data for each on-orbit satellite, and sends this feasibility data, along with the task understanding results, to the autonomous planning module.

[0107] Next, after receiving the feasibility data and mission understanding results, the autonomous planning module divides the remote sensing mission into multiple sub-tasks and their timelines based on the mission understanding results. It then performs mission planning and scheme formulation for each sub-task based on the feasibility data of each on-orbit satellite, resulting in an action execution table for each on-orbit satellite to perform the remote sensing observation mission, forming an execution action plan. The execution feedback module receives the execution action plan sent by the autonomous planning module and distributes the action execution table to each on-orbit satellite. The on-orbit satellites generate tool invocation instructions and action execution instructions based on the action execution table to perform observation actions through their onboard payloads and satellite tools, acquiring remote sensing observation data for the remote sensing observation mission.

[0108] Finally, the on-orbit computing module receives remote sensing observation data from the execution feedback module and invokes the image computing agent to perform image recognition and information annotation on the remote sensing imaging data, obtaining image recognition information about the remote sensing observation task. In the case of a remote sensing observation task involving urban congestion mapping, the image computing agent can segment roads in the target area based on the remote sensing imaging data, generating a mask region file of urban roads to obtain urban road information within the target area. Based on the urban roads represented by this information, vehicle identification is performed on the remote sensing imaging data to determine the number of vehicles on the urban roads and the confidence level of each vehicle identification result, thus obtaining vehicle identification information.

[0109] Understandably, after obtaining urban road information and vehicle identification information, the satellite sends this information to the ground station, where the ground station generates congestion data based on the urban road information and vehicle identification information to represent the road congestion status of the target area.

[0110] Reference Figure 15 As shown, in one embodiment of this application, when the remote sensing observation task is natural disaster assessment, the image recognition information includes the natural disaster characteristics of the target area, and the remote sensing observation results include the disaster impact prediction results of the target area; the remote sensing observation results are obtained in the following ways: S626. Based on the natural disaster feature database, an image computing agent is used to identify disaster features in remote sensing imaging data to obtain the natural disaster features of the target area.

[0111] S628. Based on the characteristics of natural disasters at multiple time points, conduct disaster trend simulations for the target area to obtain disaster impact prediction results for the target area, which can be used for disaster early warning.

[0112] The natural disaster feature database can be a database that stores the features of different natural disasters. These features can be remote sensing image features, including but not limited to fire location information, fire spread paths, and flood-inundated areas.

[0113] Reference Figure 16 As shown, when the remote sensing observation task is to assess natural disasters, the assessment objects can include forest fires, landslides, and floods. The remote sensing observation task can be initiated after a natural disaster occurs to monitor its development trend, or it can be initiated before a natural disaster occurs to predict and identify whether a natural disaster will occur.

[0114] First, observation tasks are initiated at the ground station by users or the management system. These tasks include routine tasks and immediate tasks. Routine tasks may be for observing high-risk disaster areas where no natural disaster has yet occurred, while immediate tasks may be for observing disaster-stricken areas where natural disasters have already taken place. Immediate tasks have higher priority and urgency than routine tasks. Understandingly, when initiating a routine task, the ground station calls an interface to obtain regional information about high-risk disaster areas, such as key fire-prone areas. Based on this information, it outputs a list of key monitoring areas and generates task instructions for the remote sensing observation mission. When initiating an immediate task, the ground station calls an interface to generate and output the specific target area for observation based on the ground location of the disaster-stricken area, thus generating task instructions for the remote sensing observation mission.

[0115] Secondly, the mission instructions for remote sensing observation are sent to the satellite. An embodied intelligent model then schedules and plans the satellites according to these instructions, enabling the remote sensing observation mission to be executed using available on-orbit satellites and to acquire remote sensing data. After obtaining the remote sensing data, the on-orbit computing module calls an image computing agent to perform image processing on the data. Taking a forest fire as an example, the image computing agent and a natural disaster feature database are used to identify fire points in the remote sensing data based on the fire point features stored in the database. If no fire point exists in the target area, a no-fire report is generated and sent to the ground station. If a fire point exists in the target area, it is located, and the location information is reported to the ground station.

[0116] Next, if fire points exist in the target area, the ground station, after obtaining fire point location information at multiple time points, performs fire intensity analysis and development trend prediction in the target area based on this information, obtaining the disaster impact prediction results for the target area. It is understandable that regardless of whether fire points exist in the target area, the ground station generates a natural disaster analysis report based on the obtained disaster impact prediction results or data such as no fire reports, and sends the natural disaster analysis report to relevant departments, thereby enabling timely early warning in the event of a natural disaster.

[0117] In some embodiments, the on-orbit computing module also invokes an evaluation and diagnostic agent to evaluate the execution status of the remote sensing observation data and obtain a completion evaluation result. If the completion evaluation result indicates that the remote sensing observation data meets preset quality conditions, no further processing is performed on the remote sensing observation data, and the normal execution flow of the remote sensing observation task is entered. If the completion evaluation result indicates that the remote sensing observation data does not meet preset quality conditions, the remote sensing observation data and the completion evaluation result are sent to the cognitive reasoning module. The cognitive reasoning module performs supplementary reasoning to generate supplementary tasks to improve the complete execution result or to update the parameters of the embodied intelligent satellite large model.

[0118] This application provides a large-scale embodied intelligent satellite model, including a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application intelligent agent module, used in the method described in any of the above embodiments. Its structure can be referred to... Figure 2 As shown.

[0119] Please see Figure 17 , Figure 17 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application, such as... Figure 17 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 17 Take a processor 10 as an example.

[0120] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0121] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0122] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0123] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0124] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0125] This application also provides a computer-readable storage medium. The methods described in this application can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the methods shown in the above embodiments are implemented.

[0126] This application also provides an on-orbit satellite that performs remote sensing observation tasks using the embodied intelligent satellite large model application method for autonomous perception, decision-making and execution as described in any of the above embodiments.

[0127] This application also provides an in-orbit constellation, comprising multiple in-orbit satellites as described in any of the above embodiments.

[0128] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0129] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0130] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0131] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0132] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0133] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0134] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0135] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0136] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

[0137] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for applying a large-scale embodied intelligent satellite model integrating autonomous perception, decision-making, and execution, characterized in that: The embodied intelligent satellite large model includes a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application intelligent agent module; the method includes: The cognitive reasoning module acquires the remote sensing observation task and generates a corresponding first state perception command; the multimodal perception module acquires the first state data of the on-orbit satellite before it executes the remote sensing observation task based on the first state perception command. The cognitive reasoning module uses the first state data to perform a feasibility analysis on whether the on-orbit satellite is suitable for performing the remote sensing observation mission, and obtains the feasibility data of the on-orbit satellite before performing the remote sensing observation mission. The autonomous planning module uses the feasibility data to perform mission planning for the on-orbit satellite, generating an execution action plan for the on-orbit satellite for the remote sensing observation mission. The execution feedback module controls the on-orbit satellite to perform the remote sensing observation task according to the execution action plan; it acquires and feeds back the remote sensing observation data of the remote sensing observation task for performance evaluation. The application intelligent agent module includes multiple intelligent agents with specific functions, which can be called by other modules in the embodied intelligent satellite large model.

2. The method according to claim 1, characterized in that, The first state data includes first internal state data and first external environment data of the on-orbit satellite before it performs the remote sensing observation mission; the feasibility analysis of whether the on-orbit satellite is suitable for performing the remote sensing observation mission using the first state data through the cognitive reasoning module, to obtain feasibility data of the on-orbit satellite before performing the remote sensing observation mission, includes: Based on the first internal state data, an internal state analysis is performed on the on-orbit satellite to obtain the internal state assessment results of the on-orbit satellite before it performs the remote sensing observation mission. Based on the first external environment data, an external environment analysis is performed on the on-orbit satellite to obtain the external environment assessment results before the on-orbit satellite performs the remote sensing observation mission. The feasibility data is obtained based on the internal state assessment results and the external environment assessment results.

3. The method according to claim 2, characterized in that, The first internal state data includes the operational state data and execution capability data of the on-orbit satellite before it performs the remote sensing observation mission; the step of performing internal state analysis on the on-orbit satellite based on the first internal state data to obtain the internal state assessment result of the on-orbit satellite before it performs the remote sensing observation mission includes: Based on the operational status data, a health status analysis is performed on the on-orbit satellite to obtain a health status assessment result of the on-orbit satellite before it performs the remote sensing observation mission. An execution condition assessment is performed on the execution capability data and the mission standard data of the remote sensing observation mission to obtain the execution condition assessment results before the on-orbit satellite performs the remote sensing observation mission; wherein, the mission standard data represents the requirements of the on-orbit satellite in terms of operational resources and observation capabilities for performing the remote sensing observation mission; The internal status assessment result is obtained based on the health status assessment result and the execution condition assessment result.

4. The method according to claim 1, characterized in that, The step of using the feasibility data through the autonomous planning module to perform mission planning for the on-orbit satellite and generate the execution action plan for the on-orbit satellite for the remote sensing observation mission includes: Based on the feasibility data, motion planning is performed on the in-orbit satellite to obtain the initial motion plan of the in-orbit satellite; Based on the initial action plan, the action scheduling of the on-orbit satellite is optimized to obtain the execution action plan for carrying out the remote sensing observation mission.

5. The method according to claim 1, characterized in that, The application agent module includes an evaluation and diagnostic agent, and the remote sensing observation data includes intermediate observation data obtained by executing some actions in the execution action plan; the method further includes: During the execution of the remote sensing observation task, the evaluation and diagnostic agent evaluates the execution status of the intermediate observation data to obtain intermediate evaluation results; If the intermediate evaluation result does not meet the preset quality conditions, the abnormal cause that caused the intermediate evaluation result to not meet the preset quality conditions is determined by the evaluation and diagnostic agent. The cognitive reasoning module generates a second state perception command corresponding to the cause of the anomaly, so as to optimize and adjust the execution action planning of the on-orbit satellite during the execution of the remote sensing observation mission and eliminate the impact of the anomaly on the remote sensing observation mission.

6. The method according to claim 5, characterized in that, The step of generating a second state perception command corresponding to the cause of the anomaly through the cognitive reasoning module, in order to optimize and adjust the execution action plan of the on-orbit satellite during the execution of the remote sensing observation mission, includes: The multimodal sensing module acquires second state data of the on-orbit satellite during the execution of the remote sensing observation mission based on the second state sensing command; The cognitive reasoning module uses the second state data to determine whether the on-orbit satellite is suitable for performing the remote sensing observation task, and obtains the standard matching data of the on-orbit satellite in the process of performing the remote sensing observation task. The autonomous planning module uses the standard matching data to optimize and adjust the execution action plan of the on-orbit satellite, thereby obtaining the optimized action plan of the on-orbit satellite.

7. The method according to claim 1, characterized in that, The application agent module includes an evaluation and diagnostic agent, and the remote sensing observation data includes the complete execution result obtained after completing the execution action planning; the method further includes: After completing the execution action planning, the evaluation and diagnostic agent evaluates the execution status of the complete execution result to obtain the completion evaluation result; If the completion evaluation result indicates that the complete execution result does not meet the preset quality conditions, the cognitive reasoning module performs supplementary reasoning on the complete execution result to generate improvement task suggestions, which are used to generate supplementary tasks to improve the complete execution result.

8. The method according to claim 1, characterized in that, The embodied intelligent satellite model also includes an on-orbit computing module, and the application intelligent agent module includes an image computing intelligent agent; the method also includes: For the target area specified in the remote sensing observation mission, the remote sensing imaging data of the target area is obtained through the on-orbit computing module; Based on the remote sensing observation mission, the on-orbit computing module calls the image computing agent to perform image recognition and information annotation on the remote sensing imaging data, thereby obtaining image recognition information about the remote sensing observation mission, which is then used to generate the remote sensing observation results of the remote sensing observation mission.

9. The method according to claim 8, characterized in that, When the remote sensing observation task is to map urban congestion, the image recognition information includes urban road information and vehicle identification information of the target area, and the remote sensing observation results include congestion data of the target area. The remote sensing observation results were obtained in the following manner: The image computing agent performs road segmentation on the remote sensing imaging data to obtain the urban road information; based on the remote sensing imaging data, the image computing agent performs vehicle identification on the urban roads represented by the urban road information to obtain the vehicle identification information. The congestion data is generated based on the urban road information and the vehicle identification information.

10. The method according to claim 8, characterized in that, When the remote sensing observation task is natural disaster assessment, the image recognition information includes the natural disaster characteristics of the target area, and the remote sensing observation results include the disaster impact prediction results of the target area; the remote sensing observation results are obtained in the following ways: Based on the natural disaster feature database, the image computing agent performs disaster feature identification on the remote sensing imaging data to obtain the natural disaster features of the target area. Based on the characteristics of natural disasters at multiple time points, the image computing agent performs disaster trend extrapolation on the target area to obtain disaster impact prediction results for the target area, which can be used for disaster early warning.

11. A large-scale embodied intelligent satellite model, characterized in that, It includes a cognitive reasoning module, a multimodal perception module, an autonomous planning module, an execution feedback module, and an application intelligent agent module. The embodied intelligent satellite large model is used to execute the method of any one of claims 1 to 10.

12. A computer device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method of any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the method of any one of claims 1 to 10.

14. An in-orbit satellite, characterized in that, Perform remote sensing observation tasks using the method described in any one of claims 1 to 10.

15. An in-orbit constellation, characterized in that, Including the on-orbit satellite as described in claim 14.

Images