Autonomous operation robot cluster cooperation method for nuclear power high-risk environment

Abstract

The invention discloses an autonomous operation robot cluster cooperation method for a nuclear power high-risk environment, and the method comprises the steps: enabling a cloud system to recognize and mark a change region in the environment relative to a map, estimating the specific position and posture of a target object in a three-dimensional space, and automatically generating a visual radiation field thermodynamic diagram. Identifying a high-risk area to trigger safe route re-planning and risk point position alarm; if it is judged that the current task exceeds the capacity limit of the single robot, cooperative task execution is carried out, and total data is transmitted back to the cloud system. According to the invention, an edge intelligent method integrating sensing, positioning, decision making and execution is established, real autonomous navigation, obstacle avoidance, target identification and fine operation are realized, dynamic real-time decision making can be carried out according to the risk of the radiation field, the dependence on manual operation is greatly reduced, and the strain capacity in a complex and uncertain environment is improved.

Description

Technical Field

[0001] This invention relates to the field of machine collaboration, and more specifically, to a method for collaborative operation of autonomous robot swarms in high-risk nuclear power environments. Background Technology

[0002] In high-risk scenarios such as routine inspections, emergency response, and decommissioning and dismantling of nuclear power plants, traditional operating robots have the following shortcomings: 1. Insufficient environmental adaptability: Traditional robots have insufficient radiation resistance, and key electronic components are prone to soft errors or permanent damage under cumulative doses, leading to system failure; in addition, existing protection solutions mostly use simple lead shielding, which is heavy and seriously restricts the robot's mobility and endurance. 2. Low level of autonomy and intelligence, relying on pre-programmed paths or remote teleoperation, poor adaptability in complex and dynamically changing nuclear environments, and the reliability of the vision system's perception results drops sharply under strong radiation, smoke, steam and other interference. 3. Limited functionality and lack of collaboration: Most equipment is dedicated to a single function (only for inspection or only for operation), lacking integrated multi-task operation capabilities. At the same time, it lacks real-time collaboration and knowledge sharing mechanisms among multiple robots, making it difficult to cope with large-scale emergency tasks.

[0003] Therefore, developing a robotic system and cluster collaboration method suitable for high-risk scenarios such as daily inspection, emergency response, and decommissioning and dismantling of nuclear power plants is of great value and significance. Summary of the Invention

[0004] To overcome or at least partially solve the aforementioned problems, this invention provides a method for the collaborative operation of autonomous robot swarms in high-risk nuclear power environments. This invention can largely replace human personnel entering high-risk environments, significantly reducing personnel exposure risks and labor costs. Furthermore, through multi-robot collaborative operation, autonomous navigation and operation time can be reduced by more than 90%, and multi-robot collaboration improves the efficiency of large-area area surveys by 300%.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for collaborative operation of autonomous robot swarms in high-risk nuclear power environments, characterized by comprising: S1. The cloud system controls the robot to obtain sensor data through multi-source data fusion positioning, and then fuses and matches the sensor data with a prior high-precision map to identify and mark the areas of change in the environment relative to the map. S2. The robot scans target objects within its field of view using a multispectral vision system, automatically identifies predefined target objects, and estimates the specific position and orientation of the target object in three-dimensional space using binocular vision or structured light technology after successful identification. S3. During the movement and perception process, the robot continuously collects radiation data at its location. The SLAM algorithm is used to mark the radiation data of each collection point on the environmental map that is currently being built or updated, automatically generating a visualized radiation field heat map, and sharing the radiation field heat map to the cluster network. S4. The cloud system automatically analyzes the radiation field heat map based on the preset safety threshold, identifies high-risk areas, and triggers alarms to replan the safety route and the location of risk points. S5. Upon reaching the work location, based on the initial instructions and the identified target, If it is determined that the current task does not exceed the capability limit of a single robot, the robot will independently perform the task and send the task status data to the cloud system in real time during the operation. If the robot determines that the current task has exceeded the capability limit of a single robot, it will send the determination information to the cloud system, which will then process it. Cloud-based collaborative decision-making algorithm The overall task is intelligently broken down into multiple sub-tasks, which are then distributed to other robots in the network cluster for collaborative task execution. S6. After the task is completed, each robot will send all the data generated in this task back to the cloud system.

[0006] Furthermore, in step S1, the step of the robot obtaining sensor data through multi-source data fusion localization includes: Once the robot is started, its lidar, vision camera, and inertial measurement unit immediately begin to work synchronously. LiDAR quickly scans the surrounding environment and generates point cloud data; Visual cameras capture environmental image features and generate image recognition data; The inertial measurement unit provides its own acceleration and angular velocity information in real time to compensate for the data delay between sensors.

[0007] Furthermore, in step S3, the radiation data includes radiation dose rate and nuclide energy spectrum data.

[0008] Furthermore, in step S4, the high-risk area triggering alarms for replanning safe routes and identifying risk points includes: When the robot system identifies a high-risk area, it is triggered to autonomously abandon its original planned path; The robot immediately sends a high-level alert to the cloud monitoring platform and other robots at the same level, reporting the location and details of the risk point; The robot system continues to recalculate a safe route that avoids high-risk areas and minimizes cumulative radiation dose.

[0009] Furthermore, in step S5, the tasks include: The cloud system plans an optimal movement path so that the robot's visual camera faces the device; it takes pictures of the device through the visual camera to obtain image data, then uses OCR technology to read the values, and automatically records the data and binds it with timestamps and location information; Furthermore, in step S5, the tasks include: By using a structured light scanner to perform close-range 3D scanning of the equipment, a precise 3D model is obtained. This allows the robot to calculate the optimal grasping posture and rotation direction of the operated equipment during the operation. Furthermore, torque monitoring is performed in real time throughout the operation, and the application of force is immediately stopped when the torque exceeds a set threshold.

[0010] Furthermore, in step S6, the full dataset includes raw images, videos, radiation dose spectra, all status parameters, updated environmental maps, and work logs.

[0011] Furthermore, in step S6, the cloud system receives all the data and performs in-depth fusion and analysis to generate a structured comprehensive report.

[0012] The present invention has at least the following advantages or beneficial effects: 1. This invention integrates multiple functions such as inspection, identification, operation, and emergency response into a single robot. Through a cluster collaboration mechanism based on cloud-based digital twins and reinforcement learning, it realizes task allocation, map sharing, and collaborative operation among multiple robots, forming a cluster intelligence effect of "1+1>2". The ability to cope with large-scale and complex tasks and the overall operation efficiency are multiplied.

[0013] 2. This invention establishes an edge intelligence method that integrates perception, positioning, decision-making, and execution, achieving true autonomous navigation, obstacle avoidance, target recognition, and precision operation. It can also make dynamic real-time decisions based on radiation field risks, greatly reducing reliance on manual operation and improving adaptability in complex and uncertain environments.

[0014] 3. This invention integrates and analyzes all data through a cloud system, automatically generating structured reports and visual charts, transforming raw data into knowledge that can directly support operation and maintenance and emergency decision-making, realizing a closed loop from "data collection" to "decision support", and significantly improving the level of intelligence in nuclear facility management.

[0015] 4. This invention can reduce the time for manual intervention in autonomous navigation and operation by more than 90% through multi-machine collaborative operation, and multi-machine collaboration can improve the efficiency of large-area survey by 300%.

[0016] 5. This invention can replace personnel in high-risk environments to the greatest extent possible, significantly reducing the risk of personnel exposure and labor costs.

[0017] 6. This invention can provide real-time, intuitive, and comprehensive on-site data and analysis results, supporting precise and efficient operation and maintenance and emergency decision-making. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0022] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0023] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, 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 limitations, 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.

[0024] In the description of the embodiments of the present invention, "multiple" means at least two.

[0025] Example 1: A method for collaborative operation of autonomous robot swarms in high-risk nuclear power environments includes: S1. The cloud system controls the robot to obtain sensor data through multi-source data fusion positioning, and then fuses and matches the sensor data with a prior high-precision map to identify and mark the areas of change in the environment relative to the map. S2. The robot scans target objects within its field of view using a multispectral vision system, automatically identifies predefined target objects, and estimates the specific position and orientation of the target object in three-dimensional space using binocular vision or structured light technology after successful identification. S3. During the movement and perception process, the robot continuously collects radiation data at its location. The SLAM algorithm is used to mark the radiation data of each collection point on the environmental map that is currently being built or updated, automatically generating a visualized radiation field heat map, and sharing the radiation field heat map to the cluster network. S4. The cloud system automatically analyzes the radiation field heat map based on the preset safety threshold, identifies high-risk areas, and triggers alarms to replan the safety route and the location of risk points. S5. Upon reaching the work location, based on the initial instructions and the identified target, If it is determined that the current task does not exceed the capability limit of a single robot, the robot will independently perform the task and send the task status data to the cloud system in real time during the operation. If the robot determines that the current task has exceeded the capability limit of a single robot, it will send the determination information to the cloud system, which will then process it. Cloud-based collaborative decision-making algorithm The overall task is intelligently broken down into multiple sub-tasks, which are then distributed to other robots in the network cluster for collaborative task execution. S6. After the task is completed, each robot will send all the data generated in this task back to the cloud system.

[0026] In step S1 of this embodiment, the step of the robot obtaining sensor data through multi-source data fusion positioning includes: Once the robot is started, its lidar, vision camera, and inertial measurement unit immediately begin to work synchronously. LiDAR quickly scans the surrounding environment to generate point cloud data, which is used to measure distances and construct geometry. Visual cameras capture environmental image features, identify known objects, and generate image recognition data; The inertial measurement unit provides its own acceleration and angular velocity information in real time to compensate for the data delay between sensors.

[0027] The cloud system integrates and matches the sensor data with the robot's built-in high-precision map to achieve real-time, centimeter-level accurate positioning. It can also identify and mark areas of change in the environment relative to the map, such as newly added obstacles or moving objects.

[0028] In step S2 of this embodiment, the target objects include a pressure gauge, a valve handle, a leak point, and a thermometer. The specific position and orientation of the target objects in three-dimensional space include the X-axis, Y-axis, Z-axis, and rotation angle, providing a data basis for the subsequent precise operation of the robotic arm.

[0029] Binocular vision and structured light technology are two different 3D imaging technologies, mainly used in fields such as robot navigation, industrial inspection, and smart devices.

[0030] The principle of binocular vision is to use two cameras to simulate the difference in perspective between the human eye for 3D measurement. Its principle is based on triangulation: two cameras (with a fixed baseline distance) capture the same scene, and the distance between the object and the cameras is deduced by calculating the positional difference (parallax) of corresponding points in the left and right images. The closer the object, the greater the parallax; the farther away, the smaller the parallax.

[0031] Structured light technology works by projecting specific light patterns (such as infrared dot matrix) onto the surface of an object to achieve 3D measurement. The system emits coded light, and a camera receives the deformation of the light patterns reflected from the object's surface. Algorithms are then used to analyze the object's shape and distance. This technology is less sensitive to material properties and can penetrate fabrics and other materials prone to misinterpretation.

[0032] In step S3 of this embodiment, the radiation data includes radiation dose rate and nuclide energy spectrum data. The CeBr3 spectrometer and Geiger-Miller (GM) counter on the robot work continuously to collect radiation dose rate and nuclide energy spectrum data at the location without interruption.

[0033] The SLAM algorithms used include: S1, SLAM mathematical description (objective): The goal of SLAM is to estimate the joint posterior probability of the trajectory and the map: p(x_{1:t}, m | z_{1:t}, u_{1:t}) Where: x_t is the robot's pose at time t, m is the map (e.g., landmarks or point cloud), z_t is the observation, and u_t is the control input; S2, EKF-SLAM (Extended Kalman Filter): State vector: x = [x_R^T, m_1^T, ..., m_N^T]^T Forecast (time updated): x^- = f(x, u) P^- = F_x P F_x^T + F_u Q F_u^T Where: F_x is the Jacobian of the motion model with respect to the state, F_u is the Jacobian of the motion model with respect to the control, and Q is the control noise covariance; Update (Observation Update): K = P^- H^T (HP^- H^T + R)^{-1} x = x^- + K (z - h(x^-)) P = (I - KH) P^- Where: H is the Jacobian of the observation model, R is the observation noise covariance, and K is the Kalman gain.

[0034] S3, Graph-SLAM: Objective optimization function: min Σ ||e_i||^2_Ω Where: e_i is the factor residual, and Ω=R^{-1} is the information matrix; Linearization solution: H Δx = -b x ← x + Δx Where: H is the Hessian matrix, and Δx is the state increment; S4, Odometer Model (Differential Motion Model): Differential kinematics: x' = x + v dt cosθ y' = y + v dt sinθ θ' = θ + ω dt Where: v is the linear velocity, ω is the angular velocity, and dt is the time interval; S5, Laser and Visual Observation Model: Laser Landmarks: r = sqrt((x_i - x)^2 + (y_i - y)^2) φ = atan2(y_i - y, x_i - x) - θ Where: r is the measured distance, and φ is the measured azimuth angle; Camera (pinhole) model: u = fx * X / Z + cx v = fy * Y / Z + cy Where: fx, fy are the camera focal lengths; cx, cy are the principal point coordinates; X, Y, Z are the coordinates of the spatial point in the camera coordinate system.

[0035] S6. Data Association (Mahanobis Distance): Mahalanobis distance: d^2 = (z - ẑ)^TS^{-1} (z - ẑ) S = HPH^T + R Where: S is the innovation covariance, and z is the predicted observation; Matching conditions: d^2 < χ 2 _{k, α} Where: χ 2 _{k, α} is the chi-square distribution threshold.

[0036] In step S4 of this embodiment, the cloud system automatically analyzes the heat map based on the preset safety threshold to identify high-risk areas, such as sudden exceedance of the dose rate or the appearance of specific dangerous nuclides.

[0037] The high-risk areas trigger alarms for replanning safe routes and identifying risk points, including: When the robot system identifies a high-risk area, it is triggered to autonomously abandon its original planned path; The robot immediately sends a high-level alert to the cloud monitoring platform and other robots at the same level, reporting the location and details of the risk point; The robot system continues to recalculate a safe route that avoids high-risk areas and minimizes cumulative radiation dose.

[0038] In step S5 of this embodiment, the cloud system (central system) or the robot itself will determine whether the current task exceeds the capability limit of a single robot. For example, the task range may be too large, heavy objects may need to be moved, or multiple angles may need to be operated simultaneously.

[0039] If collaboration is required: The cloud-based collaborative decision-making algorithm intelligently breaks down the overall task into multiple sub-tasks and distributes them to other robots in the cluster via the network.

[0040] The cloud-based collaborative decision-making algorithms used include: S1. Cloud-based Task Decomposition: Task T1 is broken down into multiple subtasks: T_sub = {T1^1, ..., T1^k} = f_decompose(T1) Where: T1 is the original task, T1^i is the decomposed subtask, k is the number of subtasks, and f_decompose is the task decomposition function.

[0041] S2, Cloud Task Graph: The task dependency structure is represented as follows: G_T = (V, E) Where V is the set of task nodes (e.g., T1^i), and E is the dependency edge between tasks.

[0042] S3, Cloud-based Task Allocation Optimization Optimization goal: min Σ C(T1^i, R_assign(i)) Where: R_assign(i) is the robot assigned task T1^i; Cost function: C = α d + β E + γ t Where: d is the distance between the robot and the task location, E is the energy consumption estimate for performing the task, t is the expected execution time; α, β, γ are cost weights.

[0043] S4, Global Fusion: Global SLAM pose fusion: X_global = Fuse(x1, ..., xN) Where: Fuse() is the cloud fusion function, and x1...xN are the current local poses / maps uploaded by each robot; Global graph optimization: x̂ = argmin Σ ||e_ij(x)||^2_{Ω_ij} Where: e_ij(x) is the factor residual, Ω_ij is the corresponding information matrix, and x is the optimized global state.

[0044] S5, Cloud-based Command Issuance and Robot Execution Commands are issued from the cloud: Cmd_i = {T1^i, X_global, constraints} Where: constraints include velocity constraints, safety distances, etc. Robot execution strategy: π_i = f_local(Cmd_i, s_i) Where: s_i is the robot's own state (battery level, position, etc.), and f_local is the local policy planning function.

[0045] S6, Cloud-Robot Loop: End-to-end communication closed loop: S61 robot R_i → Cloud C: Report status s_i S62 Cloud C: Performs task decomposition and allocation S63 Cloud C → R_i: Issue subtask Cmd_i S64 robot R_i: Executes the task and returns to the new state s_i' Where: s_i' represents the feedback status after execution, and C represents the cloud collaboration center (Cloud).

[0046] When robot A enters the unknown area first, its core task is to explore and build a map, and then share the completed real-time map with the cluster network. Robots B and C receive the map information shared by robot A, and without having to explore again, they can directly use the existing map to efficiently perform subsequent inspection or operation tasks, greatly improving efficiency.

[0047] If no collaboration is required: Robot A will continue to autonomously execute the subsequent single-machine task sequence.

[0048] In step S5 of this embodiment, the task operation includes: The cloud system plans an optimal movement path so that the robot's visual camera faces the device; it takes pictures of the device through the visual camera to obtain image data, then uses OCR technology to read the values, and automatically records the data and binds it with timestamps and location information; For the "Read Instruments" task: Based on the initial instructions and the identified target, the robot begins to perform "reading instruments" tasks, such as "reading pressure gauge values".

[0049] The cloud system plans an optimal movement path that allows the robotic arm's vision camera to be clearly aligned with the dial.

[0050] Once the robotic arm moves to the predetermined position, the vision camera takes a picture, identifies the pointer and scale, uses OCR technology to read the value, and automatically records the data and binds it with the timestamp and location information.

[0051] In step S5 of this embodiment, the task operation includes: By using a structured light scanner to perform close-range 3D scanning of the equipment, a precise 3D model is obtained. This allows the robot to calculate the optimal grasping posture and rotation direction of the operated equipment during the operation. Furthermore, torque monitoring is performed in real time throughout the operation, and the application of force is immediately stopped when the torque exceeds a set threshold.

[0052] For the "operate valve" task: First, a structured light scanner performs a close-up 3D scan of the valve to obtain its accurate 3D model, thereby calculating the optimal gripping posture and rotation direction of the robotic arm's end effector (gripper).

[0053] The robotic arm moves in force control mode, which means that torque feedback is monitored in real time when grasping and rotating valves; when the torque exceeds the set threshold, the force is stopped immediately to prevent damage to the valve or the robotic arm itself due to jamming.

[0054] During the "reading instruments" and "operating valves" tasks, all status data of the robotic arm, including its status, position, reading results, and whether the operation was successful, are transmitted back to the backend system in real time via a secure link to ensure that the operation is transparent and controllable.

[0055] Example 2: The difference from Implementation 1 is that in step S6, the cloud system receives all the data and performs in-depth fusion and analysis to generate a structured comprehensive report; the full data includes the original images, videos, radiation dose spectra, all status parameters, updated environmental maps and work records.

[0056] After the mission is completed, the robot will send all the data generated during the mission (including original images, videos, radiation dose spectra, all status parameters, updated environmental maps and work records) back to the cloud-based digital twin platform via a secure and stable communication link.

[0057] After receiving the data, the digital twin platform deeply integrates and analyzes it with historical data, equipment models, environmental parameters, etc.

[0058] The platform automatically generates structured comprehensive reports, such as: "Radiation Level Assessment Report of Building B, Floor 2", "Inspection Results and Anomaly Warning of Equipment X", and "Summary of This Task Execution". The reports include key data, charts, conclusions and recommendations.

[0059] The final report is presented to human operators or decision-makers in a visual format, providing comprehensive, accurate, and intuitive data support for their subsequent decisions on equipment maintenance, radiation protection, and emergency response, truly realizing an intelligent model of "machines operating autonomously while humans supervise decision-making."

[0060] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0061] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for collaborative operation of autonomous robot swarms in high-risk nuclear power environments, characterized in that... include: S1. The cloud system controls the robot to obtain sensor data through multi-source data fusion positioning, and then fuses and matches the sensor data with a prior high-precision map to identify and mark the areas of change in the environment relative to the map. S2. The robot scans target objects within its field of view using a multispectral vision system, automatically identifies predefined target objects, and estimates the specific position and orientation of the target object in three-dimensional space using binocular vision or structured light technology after successful identification. S3. During the movement and perception process, the robot continuously collects radiation data at its location. The SLAM algorithm is used to mark the radiation data of each collection point on the environmental map that is currently being built or updated, automatically generating a visualized radiation field heat map, and sharing the radiation field heat map to the cluster network. S4. The cloud system automatically analyzes the radiation field heat map based on the preset safety threshold, identifies high-risk areas, and triggers alarms to replan the safety route and the location of risk points. S5. Upon arrival at the work location, assess the work capability based on the initial instructions and the identified target; If it is determined that the current task does not exceed the capability limit of a single robot, the robot will independently perform the task and send task status data to the cloud system in real time during the operation. If the robot determines that the current task has exceeded the capability limit of a single robot, the robot will send the determination information to the cloud system. The cloud system will intelligently decompose the total task into multiple sub-tasks through the cloud collaborative decision-making algorithm, and distribute them to other robots in the network cluster for collaborative task execution. S6. After the task is completed, each robot will send all the data generated in this task back to the cloud system.

2. The method for autonomous robot swarm collaboration in a high-risk nuclear power environment according to claim 1, characterized in that: In step S1, the step of the robot obtaining sensor data through multi-source data fusion localization includes: Once the robot is started, its lidar, vision camera, and inertial measurement unit immediately begin to work synchronously. LiDAR quickly scans the surrounding environment and generates point cloud data; Visual cameras capture environmental image features and generate image recognition data; The inertial measurement unit provides its own acceleration and angular velocity information in real time to compensate for the data delay between sensors.

3. The method for autonomous robot swarm collaboration in a high-risk nuclear power environment according to claim 1, characterized in that: In step S3, the radiation data includes radiation dose rate and nuclide energy spectrum data.

4. The method for autonomous robot swarm collaboration in a high-risk nuclear power environment according to claim 1, characterized in that: In step S4, the high-risk area triggers alarms for replanning safe routes and identifying risk points, including: When the robot system identifies a high-risk area, it is triggered to autonomously abandon its original planned path; The robot immediately sends a high-level alert to the cloud monitoring platform and other robots at the same level, reporting the location and details of the risk point; The robot system continues to recalculate a safe route that avoids high-risk areas and minimizes cumulative radiation dose.

5. The method for autonomous robot swarm collaboration in a high-risk nuclear power environment according to claim 1, characterized in that: The tasks in step S5 include: The cloud system plans an optimal movement path so that the robot's visual camera is facing the device; it takes pictures of the device through the visual camera to obtain image data, then uses OCR technology to read the values, and automatically records the data and binds it with timestamps and location information.

6. The method for autonomous robot swarm collaboration in a high-risk nuclear power environment according to claim 1, characterized in that: The tasks in step S5 include: By using a structured light scanner to perform close-range 3D scanning of the equipment, a precise 3D model is obtained. This allows the robot to calculate the optimal grasping posture and rotation direction of the operated equipment during the operation. Furthermore, torque monitoring is performed in real time throughout the operation, and the application of force is immediately stopped when the torque exceeds a set threshold.

7. The method for autonomous robot swarm collaboration in a high-risk nuclear power environment according to claim 1, characterized in that: In step S6, the cloud system receives all the data and performs in-depth fusion and analysis to generate a structured comprehensive report.

8. A method for collaborative operation of autonomous robot swarms in high-risk nuclear power environments according to claim 1 or 7, characterized in that: In step S6, the full dataset includes raw images, videos, radiation dose spectra, all status parameters, updated environmental maps, and work logs.

Images