Multi-task planning method and system for mechanical arm

Abstract

The invention provides a multi-task planning method and system for a mechanical arm, which are applied to the technical field of mechanical arm control and are used for realizing multi-task planning of the mechanical arm through intelligent identification of an uncertain area, two-party case deduction, multi-dimensional cost calculation and optimal scheme selection based on total cost minimization. The problems of task interruption, invalid motion increase and low efficiency caused by recheck requirements in the prior art are effectively solved. According to the method, the flexibility, the intelligent level and the overall detection efficiency of multi-task planning of the mechanical arm are remarkably improved, so that the mechanical arm can operate more stably and efficiently in a complex industrial detection environment, and the method has remarkable practical value and technical progress.

Description

Technical Field

[0001] This application relates to the field of robotic arm control technology, and in particular to a multi-task planning method and system for robotic arms. Background Technology

[0002] In modern industrial production, automated inspection is a crucial link in ensuring product quality and reliability. This is especially true in the manufacturing of large, complex equipment such as megawatt-class wind turbine blades, where meticulous internal quality checks are essential. Typically, a flexible robotic arm is deployed, with its end effector integrating various inspection tools, such as ultrasonic probes, high-resolution industrial cameras, and eddy current flaw detectors, to comprehensively inspect the blade's interior for potential problems, such as delamination of composite materials, defects in bond lines, or cracks in reinforcing ribs. These different inspection tasks place unique and stringent requirements on the robotic arm's motion accuracy, sensor operating parameter settings, and system resource consumption.

[0003] Currently, most task planning systems are designed based on a standard 3D digital model of the wind turbine blade, pre-planning the robotic arm's movement path, the sequence of tasks, and the sensor parameters required for each task. The main purpose of this planning is to improve detection efficiency and ensure that the robotic arm does not collide during operation. However, in actual industrial production environments, when the robotic arm works continuously inside the wind turbine blade for extended periods, it often encounters a thorny problem: the uncertainty of the detection results.

[0004] Because the initial detection results are unclear, the system needs to "re-inspect" this area. This re-inspection is not simply repeating the previous actions, but requires adjusting the detection strategy. This ad-hoc re-inspection requirement directly disrupts the robotic arm's original task planning. If the robotic arm has moved to the next detection area according to the original plan after completing the initial detection of the previous area, but the system subsequently reports that the detection result of the previous area is uncertain and requires immediate return for re-inspection, this will force the robotic arm to interrupt its current or upcoming subsequent tasks and replan a return route. This frequent "back and forth" or "task interruption" significantly increases the robotic arm's ineffective movement time, thereby reducing overall detection efficiency.

[0005] Existing task planning systems often lack the flexibility to handle such "temporary re-inspection" requests. They typically adhere strictly to a fixed task execution sequence, and when a re-inspection request arises, they either simply interrupt the current task and wait for manual intervention to determine how to handle it, or attempt to simply insert the re-inspection task into the current sequence without performing global optimization. This lack of feedback-based flexible adjustment significantly reduces the stability and efficiency of the entire multi-task inspection process when facing uncertainties in actual production.

[0006] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0007] In view of the shortcomings of the prior art, this application provides a multi-task planning method and system for robotic arms, which aims to solve the problem that existing robotic arm task planning systems lack flexible processing capabilities when facing uncertainty in detection results, resulting in task interruption, increased invalid movements, and thus reduced overall detection efficiency.

[0008] In a first aspect, a multi-task planning method for a robotic arm, the method comprising the following steps: S1: When the robotic arm performs the initial inspection of the inside of the blade, if there is an uncertain area with an "uncertain" inspection result, the re-inspection task request of the uncertain area, the current position of the robotic arm, the current sensor type, and the battery level are obtained. S2: Based on the re-inspection task request, the current position of the robotic arm, the current sensor type, and the battery level, respectively deduce the scheme of immediately executing the re-inspection task request and the scheme of delaying the execution of the re-inspection task request; S3: For each simulation scenario, calculate the total time, power consumption, and number of sensor switching operations; S4: Calculate the total cost by combining the total time, power consumption, and number of sensor switching, and select the simulation scheme with the lowest total cost to execute.

[0009] Through this technical solution, this application can intelligently plan the re-inspection tasks of uncertain areas encountered by the robotic arm during the inspection process. By deduce two schemes, immediate execution and delayed execution, and comprehensively consider time, power consumption and sensor switching costs, the optimal scheme is selected. This effectively solves the problem of task interruption and low efficiency caused by re-inspection requirements in the prior art, and significantly improves the flexibility of multi-task planning of the robotic arm and the overall inspection efficiency.

[0010] Furthermore, step S1 includes: S11: When the robotic arm performs the initial inspection of the inside of the blade, acquire the detection data of multiple types of sensors in the detection area; S12: Generate a detection result based on the detection data. If the detection result is "uncertain", then determine the corresponding detection area as an uncertain area. S13: Obtain the location of the uncertain region, the required sensor type, the uncertainty cause label, and the generation time, and encapsulate them into a re-inspection task request for the uncertain region; S14: Obtain the current position of the robotic arm, the current sensor type, and the battery level information.

[0011] Through this technical solution, this application can obtain detailed information on the re-inspection task request for uncertain areas, including the area location, required sensor type, cause of uncertainty, and generation time. Combined with the real-time status information of the robotic arm, it provides a comprehensive and accurate data foundation for subsequent scheme deduction, ensuring the pertinence and effectiveness of the planning.

[0012] Furthermore, step S12 includes: S121: When the detection data is an ultrasonic signal, the signal-to-noise ratio and waveform of the ultrasonic signal are obtained; S122: Compare the signal-to-noise ratio with a preset signal-to-noise ratio threshold. When the signal-to-noise ratio is less than the preset signal-to-noise ratio threshold, the detection result is "uncertain". S123: Calculate the mean square error between the waveform and the preset standard waveform. When the mean square error is greater than the preset error threshold, the detection result is "uncertain". S124: When the detection data is image information, obtain the sharpness and grayscale standard deviation of the image information; S125: When the sharpness is less than a preset sharpness threshold, or the grayscale standard deviation is greater than a preset grayscale standard deviation threshold, the detection result is "uncertain".

[0013] Through this technical solution, this application can make multi-dimensional and refined judgments on the signal-to-noise ratio, waveform mean square error, and image clarity and grayscale standard deviation of ultrasonic signals, more accurately identify areas where the detection result is "uncertain", avoid misjudgment or omission, thereby improving the reliability of the initial detection results and providing a more accurate basis for the planning of subsequent re-inspection tasks.

[0014] Furthermore, step S2 includes: S21: Obtain the priority of the re-inspection task request, the current position of the robotic arm, the location of the uncertain area, the matching degree between the required sensor and the current sensor type, and the battery power information; S22: Based on the priority, the current position of the robotic arm, the location of the uncertain area, the matching degree, and the battery level, and in conjunction with a preset battery level threshold, generate a path and task sequence for immediately executing the re-inspection task request scheme, and an optimized task sequence and path for delaying the execution of the re-inspection task request scheme.

[0015] Through this technical solution, this application can comprehensively consider key factors such as the priority of the re-inspection task, the real-time position of the robotic arm, the sensor matching degree, and the battery power, and intelligently deduce two different re-inspection schemes: immediate execution and delayed execution. This lays the foundation for subsequent cost calculation and optimal scheme selection, and significantly improves the intelligence of planning and the scientific nature of decision-making.

[0016] Furthermore, in step S22, generating the path and task sequence for immediately executing the re-inspection task request scheme includes the following steps: S221: When the priority is a high-priority re-inspection request task and the battery power is greater than the preset battery power threshold, select the scheme to immediately execute the re-inspection task request. S222: Taking the current position of the robotic arm as the starting point and the location of the uncertain area as the target point, plan a collision-free shortest path in the preset blade internal CAD model; the shortest path is the path to immediately execute the re-inspection task request scheme; S223: When the matching degree is less than the preset matching degree threshold, obtain the sensor switching time; S224: Obtain the robotic arm's moving speed and the preset re-inspection task execution time, and combine them with the path length of the shortest path and the sensor switching time to generate a task sequence for immediately executing the re-inspection task request scheme.

[0017] Furthermore, in step S22, generating the optimized task sequence and path for the delayed execution of the re-inspection task request scheme includes the following steps: S225: When the priority is a non-high priority re-inspection request task, or the battery power is less than or equal to the preset battery power threshold, select the deductive delay execution scheme for the re-inspection task request. S226: Obtain the region locations of all pending regular tasks; S227: Based on the regional location of the regular task to be executed and the regional location of the uncertain region, perform a spatial clustering operation to aggregate regular tasks with a spatial distance less than a preset distance threshold and delayed re-examination tasks into one task; S228: Based on the current position of the robotic arm and the results of spatial clustering, generate a continuous task sequence and path that minimizes both the total moving distance and the number of sensor switching, as an optimized task sequence and path for delaying the execution of the re-inspection task request.

[0018] Furthermore, step S3 includes: S31: When the deduction scheme is to immediately execute the re-inspection task request scheme, calculate the movement time of the robotic arm to reach the uncertain area based on the length of the generated path and the moving speed of the robotic arm. S32: Add the movement time, sensor switching time, and preset re-inspection task execution time to obtain the total time; S33: Obtain the load on the robotic arm and calculate the power consumption based on the load on the robotic arm and the total time; S34: Obtain the number of sensor switching based on the task sequence in the immediate execution of the re-inspection task request scheme.

[0019] Furthermore, step S3 also includes: S35: When the deduction scheme is to delay the execution of the re-inspection task request scheme, the movement time is calculated based on the generated path and the moving speed of the robotic arm, and the movement time is added to the sensor switching time to obtain the total time. S36: Calculate the power consumption based on the robotic arm load and the total time; S37: Based on the optimized task sequence in the delayed execution of the re-inspection task request scheme, obtain the number of sensor switching times.

[0020] Furthermore, step S4 also includes: S41: Assign a preset first weight, a second weight, and a third weight to the total time, the power consumption, and the number of sensor switching times, respectively, wherein the second weight is greater than the first weight and the third weight; S42: The total time, power consumption, and number of sensor switching times are weighted and summed according to the first weight, the second weight, and the third weight to obtain a comprehensive score, which represents the total cost. S43: Select the simulation scheme with the smaller total cost and execute it.

[0021] Secondly, a multi-task planning system for a robotic arm, used to implement the method described in any of the above claims, the system comprising: Acquisition Module: When the robotic arm performs the initial inspection of the inside of the blade, if there is an uncertain area with an "uncertain" result, the module acquires the re-inspection task request for the uncertain area, the current position of the robotic arm, the current sensor type, and the battery level. The deduction module: Based on the re-inspection task request, the current position of the robotic arm, the current sensor type, and the battery level, it deduces the scheme of immediately executing the re-inspection task request and the scheme of delaying the execution of the re-inspection task request. Calculation module: For each simulation scenario, calculate the total time, power consumption, and number of sensor switching operations; Execution module: Calculates the total cost by combining the total time, power consumption, and number of sensor switching, and selects the simulation scheme with the lowest total cost to execute.

[0022] Beneficial Effects: The multi-task planning method and system for robotic arms proposed in this application effectively solves the problems of task interruption, increased invalid motion, and low efficiency caused by re-inspection requirements in existing technologies by intelligently identifying uncertain regions, performing dual-scheme deduction, calculating multi-dimensional costs, and selecting the optimal solution based on minimizing total cost. This method significantly improves the flexibility, intelligence level, and overall detection efficiency of multi-task planning for robotic arms, enabling them to operate more stably and efficiently in complex industrial inspection environments, demonstrating significant practical value and technological advancement. Attached Figure Description

[0023] Figure 1 This is a flowchart of a multi-task planning method for a robotic arm proposed in this application.

[0024] Figure 2 This is a structural diagram of a multi-task planning system for a robotic arm proposed in this application.

[0025] Figure 3 This is a simplified schematic diagram of a multi-task planning system for a robotic arm proposed in this application.

[0026] Labeling explanation: 201, Acquisition module; 202, Deduction module; 203, Calculation module; 204, Execution module. Detailed Implementation

[0027] 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 a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and marked in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0028] It should be noted that similar reference numerals 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. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0029] Please refer to Figure 1 A multi-task planning method for a robotic arm, the method comprising the following steps: S1: When the robotic arm performs the initial inspection of the inside of the blade, if there is an uncertain area with an "uncertain" result, then obtain the re-inspection task request for the uncertain area, the current position of the robotic arm, the current sensor type, and the battery level. S2: Based on the re-inspection task request, the current position of the robotic arm, the current sensor type, and the battery level, respectively deduce the scheme of immediately executing the re-inspection task request and the scheme of delaying the execution of the re-inspection task request; S3: For each simulation scenario, calculate the total time, power consumption, and number of sensor switching operations; S4: Calculate the total cost by combining the total time, power consumption, and number of sensor switching, and select the simulation scheme with the lowest total cost to execute.

[0030] The multi-task planning method for robotic arms proposed in this application aims to optimize the efficiency and resource utilization of robotic arms in complex inspection tasks, especially in scenarios where uncertain inspection results require re-inspection. A robotic arm is a programmable, multi-functional automated mechanical device, typically composed of multiple joints and links, capable of simulating the functions of a human hand to perform operations such as grasping, moving, and inspection. In this application, the robotic arm is primarily used to carry various sensors to inspect the interior of wind turbine blades.

[0031] Sensor type refers to the different detection tools integrated into the end effector of the robotic arm, such as ultrasonic probes, high-resolution industrial cameras, and eddy current flaw detectors. Different sensors are suitable for detecting different types of defects and may require different configurations and switching times during operation. Battery power refers to the current power status of the robotic arm's internal power supply system, which directly affects the robotic arm's endurance and the continuity of task execution. A re-inspection task request is a task instruction automatically generated by the system or manually triggered when the robotic arm finds the detection result of a certain area to be "uncertain" during the initial inspection, ordering a second inspection of that area. This request typically includes information such as the location of the uncertain area, the required sensor type, and the reason for the uncertainty.

[0032] Total time refers to the total time required to complete a simulation plan, including the robotic arm's movement time, sensor switching time, and actual task execution time. Power consumption refers to the battery power consumed by the robotic arm during the simulation plan's execution. Sensor switching count refers to the number of times the robotic arm needs to change to different types of sensors during the execution of a simulation plan. Total cost is a quantitative indicator calculated by weighting the total time, power consumption, and sensor switching count, used to evaluate the merits of different simulation plans.

[0033] In practice, firstly, when the robotic arm performs an initial inspection of the inside of the blade, if there is an uncertain area with an "uncertain" result, the system will obtain a re-inspection request for the uncertain area, the current position of the robotic arm, the current sensor type, and the battery level.

[0034] For example, when a robotic arm performs an initial inspection of the inside of a wind turbine blade, it may use ultrasonic sensors to scan the internal structure. If the ultrasonic signal waveform in a certain area is abnormal or the signal-to-noise ratio is too low, the system may not be able to clearly determine whether a defect exists, thus marking that area as an "uncertain area." In this case, the system will automatically generate a re-inspection task request, which may include the precise coordinates of the uncertain area, the recommended type of re-inspection sensor (e.g., a higher-precision ultrasonic probe or visual re-inspection combined with an industrial camera may be needed), and the preliminary cause of the uncertainty (e.g., "signal attenuation" or "waveform blur"). Simultaneously, the system will acquire the robotic arm's current position information (e.g., via an encoder or vision positioning system), the type of sensor currently in use (e.g., ultrasonic sensor), and the robotic arm's battery level information in real time. This information can be acquired through the robotic arm's control system, sensor management module, and power management module.

[0035] Secondly, based on the re-inspection task request, the current position of the robotic arm, the current sensor type, and the battery level, we deduce the options for immediately executing the re-inspection task request and delaying its execution.

[0036] For example, after acquiring the above information, the system will simulate two possible re-inspection schemes in parallel. For the immediate execution scheme, the system will assess how long it would take, how much power it would consume, whether sensor switching would be necessary, and the number of switching attempts if the robotic arm immediately interrupted its current task and returned to the uncertain area for re-inspection. This involves shortest path planning from the current position to the uncertain area, the time and power required for sensor switching, and the execution time and power consumption of the re-inspection task itself. For the delayed execution scheme, the system will consider integrating the re-inspection task with other scheduled tasks. For example, if the robotic arm is currently performing initial inspection tasks on other areas of the blade, and these tasks are spatially close to the re-inspection task in the uncertain area, or can use the same sensor type, the system will attempt to insert the re-inspection task into the subsequent sequence of regular tasks to reduce unnecessary movement and sensor switching. This may involve re-optimizing the existing task sequence to find an optimized task sequence and path that minimizes both the total movement distance and the number of sensor switching attempts.

[0037] Next, for each simulation scenario, calculate the total time, power consumption, and number of sensor switching operations.

[0038] For example, for an immediate execution plan, the total time can be calculated as: the time it takes for the robotic arm to move from its current position to an uncertain area + sensor switching time (if needed) + re-inspection task execution time. Power consumption can be estimated based on the robotic arm's moving load, moving distance, sensor operating power, and operating time. The number of sensor switching times is determined directly based on whether sensor switching is required in the plan. For a delayed execution plan, since the task sequence may be optimized, the calculation of total time, power consumption, and the number of sensor switching times becomes more complex, requiring consideration of the movement paths of all integrated tasks, sensor switching points, and their respective execution times. For example, the total time will be the total execution time of the optimized continuous task sequence, the power consumption will be the total power consumed in executing the sequence, and the number of sensor switching times will be the total number of sensor switching times in the optimized sequence.

[0039] Finally, the total cost is calculated by combining the total time, power consumption, and number of sensor switching, and the simulation scheme with the lowest total cost is selected for execution.

[0040] For example, after calculating the various metrics for both solutions, the system assigns preset weights to total time, power consumption, and the number of sensor switches. For instance, the weight of power consumption might be higher than that of total time, and the weight of total time might be higher than that of the number of sensor switches, reflecting that in some application scenarios, battery life may be more critical than task efficiency. By weighted summation, a comprehensive score, i.e., the total cost, can be obtained for each solution. The system compares the total costs of the two solutions and selects the solution with the lowest total cost as the optimal execution strategy. For example, if the total cost of the immediate execution solution is 100, while the total cost of the delayed execution solution is 80, the system will select the delayed execution solution and instruct the robotic arm to continue executing the task according to the optimized task sequence and path planned for that solution.

[0041] Compared to existing technologies, the method in this application is no longer limited to a fixed task sequence, but possesses the ability to dynamically adjust and optimize task planning. It can effectively balance task efficiency, energy consumption, and sensor switching costs, thereby maximizing the robotic arm's operational efficiency and resource utilization while ensuring inspection quality. This flexible and intelligent planning method provides a more efficient and reliable solution for the automated inspection of large and complex equipment such as wind turbine blades.

[0042] Furthermore, step S1 includes: S11: When the robotic arm performs the initial inspection of the inside of the blade, acquire the detection data of multiple types of sensors in the detection area; S12: Generate detection results based on the detection data. If the detection result is "uncertain", then determine the corresponding detection area as an uncertain area. S13: Obtain the location of the uncertain region, the required sensor type, the uncertainty cause label, and the generation time, and encapsulate it into a re-inspection task request for the uncertain region; S14: Obtain the current position of the robotic arm, the current sensor type, and the battery level information.

[0043] Specifically, in step S11, the detection data consists of raw signal or image data acquired by various types of sensors during the detection process.

[0044] In step S12, the detection result is a conclusion drawn from the analysis and processing of the detection data obtained in step S11, which may include states such as "normal," "defective," or "uncertain." When the detection result is determined to be "uncertain," the corresponding detection area is defined as the "uncertain area." The "uncertain area" refers to the internal area of ​​the blade whose state (normal or defective) cannot be clearly determined through preliminary detection due to various factors (such as signal interference, image blurring, incomplete data, etc.).

[0045] In step S13, once the uncertain region is identified, the system will further acquire detailed information about that region. The region's location can refer to its three-dimensional coordinates inside the blade, a region ID, or other spatial identifier. The required sensor type refers to the type of sensor needed for a more accurate re-inspection of the uncertain region. For example, if the initial detection is a blurred image detected by a vision sensor, a re-inspection might require a higher-resolution vision sensor or an ultrasonic sensor to penetrate the material for detection.

[0046] The "Uncertainty Reason" label provides a preliminary assessment of why the detection result is "uncertain," such as "low signal-to-noise ratio," "insufficient image clarity," or "abnormal waveform." These labels aid in the targeted planning of subsequent re-inspection tasks. The generation time is the timestamp at which the re-inspection task request was created. All of the above information will be encapsulated into a structured data packet, namely the "Re-inspection Task Request," for processing and scheduling by the subsequent task planning module.

[0047] In step S14, the current position of the robotic arm refers to its real-time coordinates in three-dimensional space, which is crucial for planning the movement path. The current sensor type refers to the type of sensor the robotic arm is currently using, which helps determine whether a sensor switch is necessary. Battery power information is the remaining percentage of battery power or available working time of the robotic arm; these are key parameters for assessing task execution feasibility and planning power consumption.

[0048] The above technical solution ensures that when the robotic arm performs multi-task planning, the source, content, and status of the re-inspection task requests are acquired more comprehensively, accurately, and promptly. This not only provides a solid data foundation for subsequent simulation plans (immediate execution or delayed execution), but also makes task planning more targeted by clarifying the characteristics and re-inspection needs of uncertain areas. This effectively avoids planning deviations caused by insufficient or ambiguous information, thereby improving the overall efficiency and reliability of task planning.

[0049] Furthermore, step S12 includes: S121: When the detection data is an ultrasonic signal, the signal-to-noise ratio and waveform of the ultrasonic signal are obtained; S122: Compare the signal-to-noise ratio with a preset signal-to-noise ratio threshold. When the signal-to-noise ratio is less than the preset signal-to-noise ratio threshold, the detection result is "uncertain". S123: Calculate the mean square error between the waveform and the preset standard waveform. When the mean square error is greater than the preset error threshold, the detection result is "uncertain". S124: When the detection data is image information, obtain the image sharpness and grayscale standard deviation; S125: When the sharpness is less than the preset sharpness threshold, or the grayscale standard deviation is greater than the preset grayscale standard deviation threshold, the detection result is "uncertain".

[0050] Specifically, the detection data can be understood as the raw data collected by the robotic arm using various types of sensors during the initial inspection of the blade's interior. This data can be ultrasonic signals, used to detect defects or structural anomalies within the material; or it can be image information, used to identify surface defects or perform visual inspection.

[0051] In this context, when the detected data is an ultrasonic signal, the signal-to-noise ratio (SNR) is the ratio of signal strength to noise strength, reflecting the quality of the ultrasonic signal. A preset SNR threshold is a reference value determined based on experience or experimentation, used to determine if the signal is sufficiently clear and reliable. When the SNR is lower than this preset threshold, it indicates severe interference to the signal, which may not accurately reflect the internal condition of the blade; therefore, the detection result is judged as "uncertain." Waveform refers to the time-domain or frequency-domain representation of the ultrasonic signal. The preset standard waveform is the ultrasonic response waveform of an ideal or healthy blade's internal structure. Mean square error (MSE) quantifies the degree of difference between the actual waveform and the preset standard waveform. When the MSE is greater than the preset error threshold, it indicates a significant deviation between the actual waveform and the standard waveform, potentially indicating internal defects or anomalies, thus leading to an "uncertain" detection result.

[0052] When the detection data is image information, sharpness refers to the image's sharpness and detail representation. A preset sharpness threshold is used to determine whether the image is sharp enough for effective analysis. Gray-level standard deviation reflects the dispersion of image pixel gray-level values ​​and is typically used to measure image texture richness and contrast. A preset gray-level standard deviation threshold is used to determine whether the image's gray-level distribution is abnormal. When the image sharpness is lower than the preset sharpness threshold, or the gray-level standard deviation is greater than the preset gray-level standard deviation threshold, the image quality may be insufficient to accurately determine the leaf condition, and therefore the detection result is judged as "uncertain."

[0053] This application's solution, by setting specific judgment criteria for different types of detection data (such as ultrasonic signals and image information), can more precisely identify areas where the detection result is "uncertain." For ultrasonic signals, the dual judgment of signal-to-noise ratio and waveform mean square error can effectively avoid misjudgments caused by poor signal quality or internal structural anomalies. For image information, image quality is evaluated through sharpness and grayscale standard deviation, ensuring that only high-quality image data is used for subsequent analysis. This multi-dimensional and multi-type judgment mechanism enables the robotic arm to more accurately identify uncertain areas requiring re-inspection during the initial detection, thus providing reliable input for subsequent re-inspection task planning.

[0054] Furthermore, step S2 includes: S21: Obtain the priority of the re-inspection task request, the current position of the robotic arm, the location of the uncertain area, the matching degree between the required sensor and the current sensor type, and the battery power information; S22: Based on priority, current position of the robotic arm, location of the uncertain area, matching degree, and battery level, and in conjunction with a preset battery level threshold, generate the path and task sequence for the immediate execution of the re-inspection task request scheme, and the optimized task sequence and path for the delayed execution of the re-inspection task request scheme.

[0055] Specifically, in step S21, the priority of the re-inspection task request is used to indicate the urgency and importance of the task, for example, it can be divided into high priority and normal priority; the current position of the robotic arm and the area position of the uncertain area are used to determine the moving distance and path planning of the robotic arm; the matching degree between the required sensor and the current sensor type is used to assess whether sensor switching is needed and the cost of switching; the battery power information is directly related to the endurance of the robotic arm and the feasibility of the task.

[0056] Among them, high-priority tasks include at least: initial detection results of uncertain regions indicate the presence of cracks, large-area delamination, and bonding defects in critical parts that could lead to structural failure; the "uncertain regions" are located in critical parts of the blade where stress is concentrated, such as the main spars, flanges, and roots.

[0057] The current position of the robotic arm is obtained through its own encoder system. Each joint of the robotic arm is typically equipped with a high-precision rotary encoder or linear encoder. These encoders can measure the angle or displacement of each joint in real time. By combining these joint measurements with the kinematic model of the robotic arm (including forward and inverse kinematics), the Cartesian coordinates of the robotic arm's end effector (or its onboard sensors) in three-dimensional space can be accurately calculated. This method is the primary means of internal positioning of robotic arms in existing technology, offering high precision and real-time performance.

[0058] The location of an uncertain region typically refers to the three-dimensional coordinates inside the blade, which can be precisely determined by combining the robotic arm's own positioning system (e.g., through an encoder, inertial measurement unit, or external vision positioning system) with the blade's CAD model. For example, during the inspection process, the robotic arm records the real-time position of its end effector (i.e., the sensor) inside the blade, and when the inspection result is "uncertain," this positional information is associated with the uncertain region.

[0059] The matching degree between the required sensor and the current sensor type is obtained by determining whether the required sensor type and the current sensor type are completely identical. If they are exactly the same, the matching degree can be set to the highest value (e.g., 1). This means that no sensor switching is required, and the matching degree is high.

[0060] If the required sensor type is not exactly the same as the current sensor type, but the current sensor has the core or some functions required to perform the re-inspection task, a lower matching degree (between 0.1 and 0.9) can be set based on its compatibility. For example, if the re-inspection requires a high-resolution camera, but the current camera is of ordinary resolution, although it is not a perfect match, it still has a certain visual inspection capability, and the matching degree can be set to a medium value.

[0061] If the required sensor type is completely different from the current sensor type and cannot be replaced functionally, and a sensor switch is required for re-inspection, then the matching degree is set to 0.

[0062] In step S22, based on the information obtained above—namely, priority, current position of the robotic arm, location of the uncertain area, matching degree, and battery level—and combined with a preset battery level threshold, paths and task sequences for immediately executing the re-inspection task request scheme, and optimized task sequences and paths for delaying the execution of the re-inspection task request scheme, can be generated. The preset battery level threshold is used to determine whether the current battery level is sufficient to support immediate task execution, or whether delayed execution should be considered for charging or merging with other tasks. By comprehensively considering these factors, specific and operable paths and task sequences can be generated for two different execution strategies (immediate execution and delayed execution). Here, the path refers to the trajectory of the robotic arm moving from its current position to the target area, and the task sequence refers to a series of actions and operations that the robotic arm needs to complete during task execution, such as movement, sensor switching, and detection.

[0063] Furthermore, in step S22, generating the path and task sequence for the immediate execution of the re-inspection task request scheme includes the following steps: S221: When the priority is a high-priority re-inspection request task and the battery power is greater than the preset battery power threshold, select the scheme of immediately executing the re-inspection task request. S222: Starting from the current position of the robotic arm and with the location of the uncertain area as the target point, plan a collision-free shortest path in the preset CAD model inside the blade; the shortest path is the path of the immediate execution of the re-inspection task request scheme; S223: When the matching degree is less than the preset matching degree threshold, obtain the sensor switching time; S224: Obtain the robotic arm's moving speed and the preset re-inspection task execution time, and combine the path length of the shortest path and the sensor switching time to generate a task sequence for executing the immediate re-inspection task request scheme.

[0064] Specifically, in step S221, the conditions for selecting the immediate execution of the re-inspection task request scheme are clearly defined. Priority can be understood as the urgency or importance of the re-inspection task, for example, it can be divided into high-priority tasks and normal-priority tasks. Battery power refers to the current available power of the robotic arm. The preset battery power threshold is a benchmark value used to determine whether the robotic arm has the power required to execute the immediate task. For example, it can be set as the minimum power required for the robotic arm to complete one re-inspection task, or a safety margin value. Only when the re-inspection task has high priority and the robotic arm's battery power is sufficient will the immediate execution of the re-inspection task be considered, ensuring the rationality and feasibility of immediate execution.

[0065] In step S222, the determined region location refers to the coordinates of the specific region requiring re-inspection. The preset blade internal CAD model is a three-dimensional digital model of the blade's internal structure, providing precise geometric information and obstacle distribution for path planning. Planning a collision-free shortest path means finding a path in the CAD model from the current position of the robotic arm to the region location of the uncertain area that minimizes both collision with the blade's internal structure and the movement distance. This shortest path planning can be implemented using the RRT (Fast Random Tree) algorithm, which aims to minimize the robotic arm's movement time and ensure the safety of the movement process.

[0066] Specifically, the RRT (Fast Random Tree) algorithm is a commonly used sampling algorithm in the field of path planning. It can effectively search for a collision-free path from the starting point to the target point in a complex high-dimensional space.

[0067] The specific process of implementing shortest path planning using the RRT algorithm can be understood as follows: First, in the CAD model of the blade's interior, the current position of the robotic arm is set as the starting point, and the location of the uncertain area is set as the target point. The RRT algorithm constructs a random tree starting from the starting point. Then, a sampling point is randomly generated in the free space defined by the CAD model of the blade's interior. This sampling point represents a possible position that the robotic arm can reach inside the blade.

[0068] Next, in the currently generated random tree, find the node closest to the sampling point. This closest node is an existing reachable position for the robotic arm in the random tree. Then, extend a short distance from the closest node towards the random sampling point to generate a new node. When generating the new node, ensure that the path from the closest node to the new node is collision-free in the CAD model inside the blade. If a collision occurs, discard the sampling point and re-sample randomly.

[0069] Collision detection is performed each time a new node is generated by expanding the random tree. This involves checking whether the robotic arm model (including its end effector sensors) interferes with any obstacles (such as blade structures, stiffeners, etc.) in the internal CAD model of the blade as it moves from the nearest node to the new node. If a collision is detected, the expansion is invalid and resampling is required.

[0070] Finally, repeat the steps of random sampling, finding the nearest node, expanding the random tree, and collision detection until the newly generated node is close enough to the target point (e.g., less than a preset distance threshold) or directly connected to the target point. Once the target point is reached, it means that a collision-free path from the starting point to the target point has been found.

[0071] In step S223, the matching degree refers to the compatibility or consistency between the current sensor type and the sensor type required for the re-inspection task. For example, if the two are completely identical, the matching degree is 1; if they are not completely identical, the matching degree is [0.1, 0.9]; if they are completely different, the matching degree is 0. The preset matching degree threshold is the boundary value for determining whether sensor switching is necessary. When the current sensor cannot meet the requirements of the re-inspection task (i.e., the matching degree is less than the preset threshold of 0.5), it is necessary to obtain the sensor switching time. This time is the time required for the robotic arm to switch from the current sensor to the required sensor. The purpose is to incorporate the sensor switching time into the consideration of the task sequence.

[0072] In step S224, a task sequence is generated to execute the immediate re-inspection task request scheme. The robotic arm's movement speed is its average movement speed while performing the task. The preset re-inspection task execution time is the standard time required to complete one re-inspection task. By combining the path length of the shortest path, the robotic arm's movement speed (used to calculate the movement time), and the sensor switching time (if any) and the preset re-inspection task execution time, a complete task sequence can be constructed. This sequence includes the various stages of the robotic arm moving from its current position to the target area, performing sensor switching (if necessary), and executing the re-inspection task, along with their required times. For example, in a specific embodiment, the system checks the matching degree between the sensor type currently mounted on the robotic arm (e.g., a common vision sensor) and the sensor type required for the re-inspection task (e.g., a high-precision ultrasonic sensor). If the matching degree is lower than a preset threshold, the system retrieves the sensor switching time from the database; for example, switching to a high-precision ultrasonic sensor requires 5 seconds. The system obtains the robotic arm's average movement speed (e.g., 0.2 m / s) and the preset re-inspection task execution time (e.g., 15 seconds). Based on the shortest path length (1.5 meters), the movement time is calculated to be 1.5 meters / 0.2 meters / second = 7.5 seconds. Therefore, the task sequence for immediately executing the re-inspection request scheme is generated as follows: move to the uncertain area (7.5 seconds) -> switch sensors (5 seconds) -> execute the re-inspection task (15 seconds). This detailed task sequence will be used for subsequent total cost calculations.

[0073] Furthermore, in step S22, generating the optimized task sequence and path for the delayed execution of the re-examination task request scheme includes the following steps: S225: When the priority is a non-high priority re-inspection request task, or the battery power is less than or equal to the preset battery power threshold, select the scheme of delaying the execution of the re-inspection task request. S226: Obtain the region locations of all pending regular tasks; S227: Based on the regional location of the regular tasks to be executed and the regional location of the uncertain regions, perform spatial clustering operations to aggregate regular tasks with spatial spacing less than a preset spacing threshold and delayed re-examination tasks into one task; S228: Based on the current position of the robotic arm and the results of spatial clustering, generate a continuous task sequence and path that minimizes both the total moving distance and the number of sensor switching, as an optimized task sequence and path for delaying the execution of the re-inspection task request.

[0074] In step S225, non-high-priority re-inspection request tasks are classified as ordinary-priority tasks. Tasks that are not urgent or whose power is insufficient for immediate execution will be prioritized for delay to allow for more optimized planning. In step S226, routine tasks can be other inspection, cleaning, or maintenance tasks inside the blades, and their location is the specific spatial coordinates at which the robotic arm needs to travel to perform these tasks. In step S227, spatial clustering refers to grouping tasks with similar geographical locations or spatial coordinates. Specifically, if the spatial distance between routine tasks and delayed re-inspection tasks is less than a preset distance threshold, these tasks will be aggregated into one task. The purpose is to merge spatially adjacent tasks and reduce the number of times the robotic arm moves back and forth between different task points. The preset distance threshold can be set according to the actual application scenario, the robotic arm's movement accuracy, and task execution efficiency requirements. In step S228, after considering all clustered task points, the system will plan an optimal path so that the robotic arm starts from the current position and executes all tasks sequentially, minimizing not only the total travel distance but also the number of sensor switches. This optimal path and task sequence is thus determined as the optimized task sequence and path for delaying the execution of the re-inspection task request. Minimizing the number of sensor switching operations is particularly important for tasks that require sensor replacement, as sensor switching typically incurs additional time and energy consumption.

[0075] In some preferred embodiments, suppose that during the initial inspection of the blade's interior, the robotic arm discovers an "uncertain" area in region A, requiring re-inspection. This re-inspection task is determined by the system to be non-high priority, and the robotic arm's current battery level is below a preset threshold. Simultaneously, the robotic arm's task queue contains three routine inspection tasks to be executed, located in regions B, C, and D. Regions A, B, and C are spatially close to each other, while region D is relatively far away. In this case, the method of this application first obtains the regional locations of regions A, B, C, and D. Next, a spatial clustering operation is performed. For example, if the spatial distance between region A and region B is less than a preset distance threshold, and the spatial distance between region B and region C is also less than this threshold, then regions A, B, and C will be aggregated into a task group. Region D, due to its greater distance, may be treated as a separate task. Subsequently, based on the robotic arm's current position and the clustering results (e.g., executing clustering task group ABC first, then executing task D), the system plans a continuous task sequence and path that minimizes both the total travel distance and the number of sensor switches. For example, the robotic arm might be planned to first move to area A to complete a re-inspection task, then move along the same path to areas B and C to complete routine inspection tasks, and finally move to area D to complete the remaining routine tasks. During this process, if the tasks in areas A, B, and C require the same sensor type, unnecessary sensor switching can be avoided, further improving efficiency. In this way, the delayed re-inspection task is effectively integrated into the overall task planning of the robotic arm, achieving optimal resource allocation and maximizing task execution efficiency.

[0076] Furthermore, step S3 includes: S31: When the simulation plan is to immediately execute the re-inspection task request plan, calculate the movement time of the robotic arm to reach the uncertain area based on the length of the generated path and the moving speed of the robotic arm. S32: Add the movement time, sensor switching time, and preset re-inspection task execution time to get the total time; S33: Obtain the load on the robotic arm and calculate the power consumption based on the load and total time; S34: Obtain the number of sensor switching times based on the task sequence in the immediate execution re-inspection task request scheme.

[0077] Step S31 aims to calculate the movement time of the robotic arm to reach the uncertain region. This movement time is obtained by dividing the length of the generated path by the robotic arm's movement speed. The length of the generated path refers to the actual length of the planned path from the current position to the uncertain region. The robotic arm's movement speed refers to the average movement speed of the robotic arm when performing the task, which can be preset or obtained in real time based on factors such as the robotic arm's model, load, and operating environment.

[0078] In step S32, the total time is calculated as the sum of the movement time, sensor switching time, and preset re-inspection task execution time. Sensor switching time refers to the time required to change the sensor when the robot arm's current sensor type does not match the sensor type required by the re-inspection task request. Preset re-inspection task execution time refers to the standard time required to complete the re-inspection task in the uncertain area, which can be preset according to the complexity of the task and the sensor type.

[0079] In step S33, the calculation of power consumption depends on the robot arm load and the total time. The robot arm load refers to the weight or resistance borne by the robot arm during task execution, represented by the robot arm's current operating power, which directly affects the robot arm's power consumption. By correlating the robot arm load with the total time, the required power consumption of the scheme can be accurately assessed. That is, power consumption = current operating power × total time.

[0080] In step S34, the number of sensor switching is obtained based on the task sequence in the immediate execution re-inspection task request scheme. The task sequence details the various operations that the robotic arm needs to complete during the re-inspection task, including movement, sensor switching, and detection. The number of sensor switching can be determined by analyzing this sequence.

[0081] Furthermore, step S3 also includes: S35: When the simulation scheme is a delayed execution of the re-inspection task request scheme, the movement time is calculated based on the generated path and the moving speed of the robotic arm, and the movement time is added to the sensor switching time to obtain the total time. S36: Calculate power consumption based on the robotic arm load and total time; S37: Based on the optimized task sequence in the delayed execution re-inspection task request scheme, obtain the number of sensor switching times.

[0082] Specifically, in step S35, when the deduced scheme is determined to be a delayed execution of the re-inspection task request scheme, it is based on the path generated in step S22 and the moving speed of the robotic arm. The generated path refers to the continuous task sequence and path generated in step S228, based on the current position of the robotic arm and the results of spatial clustering, which minimizes both the total moving distance and the number of sensor switching times. The moving speed of the robotic arm refers to its average or preset moving speed when performing the task. By dividing the length of the generated path by the moving speed of the robotic arm, the moving time required for the robotic arm to execute the delayed scheme can be obtained. The moving time is then added to the sensor switching time to obtain the total time. The total time for this delayed execution of the re-inspection task request scheme differs from the total time for the immediate execution of the re-inspection task request scheme. This is because, in the delayed execution of the re-inspection task request scheme, the robotic arm does not need to go back and repeat the execution. Instead, it directly combines the uncertain area requiring re-inspection with the remaining area to be inspected, replans a path, and the robotic arm starts from its current position and moves along the replanned path. The moving time along this path, plus the sensor switching time, is the total time.

[0083] In step S36, power consumption is calculated. This calculation is based on the robot arm load (represented by the current operating power) and the total time. The calculation formula is: Power consumption = Current operating power × Total time.

[0084] In step S37, the number of sensor switching operations is obtained. This is obtained based on the optimized task sequence in the delayed execution re-inspection task request scheme. The optimized task sequence is a continuous task sequence with the minimum total travel distance and the minimum number of sensor switching operations generated after spatial clustering operations in step S228 above. By analyzing the sensor types required for different tasks in this optimized task sequence, the total number of sensor switching operations required during the execution of this sequence can be calculated.

[0085] Furthermore, step S4 also includes: S41: Assign a preset first weight, a second weight, and a third weight to the total time, power consumption, and number of sensor switching, respectively, wherein the second weight is greater than the first weight, which is greater than the third weight; S42: The total time, power consumption, and number of sensor switching times are weighted and summed according to the first weight, the second weight, and the third weight to obtain a comprehensive score, which represents the total cost. S43: Select the simulation scheme with the lower total cost and execute it.

[0086] The first, second, and third weights are parameters used to quantify the relative importance of total time, power consumption, and sensor switching frequency in the total cost calculation. These weights can be preset or dynamically adjusted based on factors such as actual application requirements, the robotic arm's working environment, task priority, and energy management strategies. For example, when battery power is a key consideration, the second weight (corresponding to power consumption) can be set to a relatively high value. A second weight greater than the first weight greater than the third weight means that in the current application scenario, power consumption is considered the most important consideration, followed by total time, while the importance of sensor switching frequency is relatively low. This weight setting aims to guide the system to prioritize solutions with lower power consumption while also considering task completion time and operational efficiency.

[0087] Specifically, weighted summation involves multiplying the calculated total time, power consumption, and sensor switching count by their respective first, second, and third weights, then summing the products to obtain a comprehensive score. This comprehensive score is defined as the total cost of the proposed solution. Through the above technical solution, this application enables more refined and intelligent cost evaluation of multi-task planning schemes for robotic arms.

[0088] Please refer to Figure 2 , Figure 3 A multi-task planning system for a robotic arm, used to implement any of the above methods, the system comprising: Acquisition Module 201: When the robotic arm performs the initial inspection of the inside of the blade, if there is an uncertain area with an "uncertain" inspection result, the module acquires the re-inspection task request for the uncertain area, the current position of the robotic arm, the current sensor type, and the battery level. Module 202: Based on the re-inspection task request, the current position of the robotic arm, the current sensor type, and the battery level, it simulates the options for immediately executing the re-inspection task request and delaying the execution of the re-inspection task request. Calculation module 203: For each simulation scheme, calculate the total time, power consumption, and number of sensor switching. Execution module 204: Calculates the total cost by combining the total time, power consumption, and number of sensor switching, and selects the simulation scheme with the lowest total cost to execute.

[0089] The acquisition module 201 is configured to, during the initial inspection of the blade's interior by the robotic arm, if an uncertain region is identified (resulting in an "uncertain" status), acquire the re-inspection request for the uncertain region, the robotic arm's current position, the current sensor type, and the battery level. Specifically, the acquisition module 201 is responsible for collecting all initial information related to task planning, including but not limited to detailed information about the uncertain region (such as region location, required sensor type, uncertainty cause label, generation time, etc.) and the robotic arm's real-time status information (such as current position, current sensor type, battery level, etc.). This information forms the basis for subsequent task simulation and cost calculation.

[0090] The deduction module 202 is configured to deduce two possible execution schemes based on the re-inspection task request, the current position of the robotic arm, the current sensor type, and the battery level: one for immediately executing the re-inspection task request, and the other for delaying its execution. Specifically, the deduction module 202 will generate two possible task execution schemes based on the information provided by the acquisition module and in conjunction with preset planning strategies (such as task priority, battery level threshold, sensor matching degree, etc.): one is to execute the re-inspection task immediately, and the other is to delay the execution of the re-inspection task and potentially merge it with other regular tasks for optimization. Each scheme will include corresponding path planning and task sequences.

[0091] The calculation module 203 is configured to calculate the total time, power consumption, and number of sensor switches for each simulation scenario. Specifically, the calculation module 203 receives two scenarios generated by the simulation module and performs a detailed cost assessment for each scenario. This includes calculating the travel time based on the path length and the robot arm's movement speed, combining the sensor switch time and the re-inspection task execution time to obtain the total time; calculating the power consumption based on the robot arm's load and the total time; and determining the number of sensor switches based on the task sequence. These calculation results are crucial for decision-making.

[0092] The execution module 204 is configured to calculate the total cost by considering the total time, power consumption, and number of sensor switching operations, and then select the least cost-effective solution to execute. Specifically, the execution module 204 receives various cost data from the calculation module and performs a weighted sum of these costs according to preset weights (e.g., power consumption may have a higher weight than total time, and total time may have a higher weight than the number of sensor switching operations) to obtain the total cost for each solution. Finally, the execution module 204 selects the solution with the lowest total cost and instructs the robotic arm to perform the task according to that solution.

[0093] This application's system, through modular design, concretizes the logical steps in the multi-task planning method for robotic arms into functional modules that can operate independently yet collaborate with each other, thereby achieving effective management and execution of the complex planning process. The acquisition module, as the information input end, ensures the integrity and real-time nature of all data required for decision-making; the deduction module, based on this data, generates multiple possible execution strategies in parallel, providing a foundation for subsequent optimization; the calculation module quantitatively evaluates each strategy, transforming abstract execution schemes into comparable cost data; finally, the execution module makes intelligent decisions based on these quantified costs, ensuring that the robotic arm always selects the optimal task execution path. This clearly defined division of labor and collaborative mechanism makes the entire planning process more efficient and accurate, and adaptable to dynamically changing operating environments.

[0094] The obstacle avoidance control system for robotic arms proposed in this application represents a significant advancement compared to existing technologies. Traditional obstacle avoidance systems typically employ a uniform strategy based on a fixed safety distance, failing to differentiate the physical properties of obstacles. This can lead to situations where, when operating in complex and confined spaces such as wind power equipment, the robotic arm may lose its working space due to excessive avoidance of flexible obstacles, or its mission may be interrupted due to misjudged collisions.

[0095] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0096] The above description is merely an embodiment of this application and is not intended to limit the scope of protection 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 protection of this application.

Claims

1. A multi-task planning method for a robotic arm, characterized in that, The method includes the following steps: S1: When the robotic arm performs the initial inspection of the inside of the blade, if there is an uncertain area with an "uncertain" result, then obtain the re-inspection task request for the uncertain area, the current position of the robotic arm, the current sensor type, and the battery level. S2: Based on the re-inspection task request, the current position of the robotic arm, the current sensor type, and the battery level, respectively deduce the scheme of immediately executing the re-inspection task request and the scheme of delaying the execution of the re-inspection task request; S3: For each simulation scenario, calculate the total time, power consumption, and number of sensor switching operations; S4: Calculate the total cost by combining the total time, power consumption, and number of sensor switching, and select the simulation scheme with the lowest total cost to execute.

2. The multi-task planning method for a robotic arm according to claim 1, characterized in that, Step S1 includes: S11: When the robotic arm performs the initial inspection of the inside of the blade, acquire the detection data of multiple types of sensors in the detection area; S12: Generate a detection result based on the detection data. If the detection result is "uncertain", then determine the corresponding detection area as an uncertain area. S13: Obtain the location of the uncertain region, the required sensor type, the uncertainty cause label, and the generation time, and encapsulate them into a re-inspection task request for the uncertain region; S14: Obtain the current position of the robotic arm, the current sensor type, and the battery level information.

3. The multi-task planning method for a robotic arm according to claim 2, characterized in that, Step S12 includes: S121: When the detection data is an ultrasonic signal, the signal-to-noise ratio and waveform of the ultrasonic signal are obtained; S122: Compare the signal-to-noise ratio with a preset signal-to-noise ratio threshold. When the signal-to-noise ratio is less than the preset signal-to-noise ratio threshold, the detection result is "uncertain". S123: Calculate the mean square error between the waveform and the preset standard waveform. When the mean square error is greater than the preset error threshold, the detection result is "uncertain". S124: When the detection data is image information, obtain the sharpness and grayscale standard deviation of the image information; S125: When the sharpness is less than a preset sharpness threshold, or the grayscale standard deviation is greater than a preset grayscale standard deviation threshold, the detection result is "uncertain".

4. The multi-task planning method for a robotic arm according to claim 1, characterized in that, Step S2 includes: S21: Obtain the priority of the re-inspection task request, the current position of the robotic arm, the location of the uncertain area, the matching degree between the required sensor and the current sensor type, and the battery power information; S22: Based on the priority, the current position of the robotic arm, the location of the uncertain area, the matching degree, and the battery level, and in conjunction with a preset battery level threshold, generate a path and task sequence for immediately executing the re-inspection task request scheme, and an optimized task sequence and path for delaying the execution of the re-inspection task request scheme.

5. The multi-task planning method for a robotic arm according to claim 4, characterized in that, In step S22, generating the path and task sequence for immediately executing the re-inspection task request scheme includes the following steps: S221: When the priority is a high-priority re-inspection request task and the battery power is greater than the preset battery power threshold, select the scheme to immediately execute the re-inspection task request. S222: Taking the current position of the robotic arm as the starting point and the location of the uncertain area as the target point, plan a collision-free shortest path in the preset blade internal CAD model; the shortest path is the path to immediately execute the re-inspection task request scheme; S223: When the matching degree is less than the preset matching degree threshold, obtain the sensor switching time; S224: Obtain the robotic arm's moving speed and the preset re-inspection task execution time, and combine them with the path length of the shortest path and the sensor switching time to generate a task sequence for immediately executing the re-inspection task request scheme.

6. The multi-task planning method for a robotic arm according to claim 4, characterized in that, In step S22, generating the optimized task sequence and path for the delayed execution of the re-inspection task request scheme includes the following steps: S225: When the priority is a non-high priority re-inspection request task, or the battery power is less than or equal to the preset battery power threshold, select the deductive delay execution scheme for the re-inspection task request. S226: Obtain the region locations of all pending regular tasks; S227: Based on the regional location of the regular task to be executed and the regional location of the uncertain region, perform a spatial clustering operation to aggregate regular tasks with a spatial distance less than a preset distance threshold and delayed re-examination tasks into one task; S228: Based on the current position of the robotic arm and the results of spatial clustering, generate a continuous task sequence and path that minimizes both the total moving distance and the number of sensor switching, as an optimized task sequence and path for delaying the execution of the re-inspection task request.

7. The multi-task planning method for a robotic arm according to claim 1, characterized in that, Step S3 includes: S31: When the deduction scheme is to immediately execute the re-inspection task request scheme, calculate the movement time of the robotic arm to reach the uncertain area based on the length of the generated path and the moving speed of the robotic arm. S32: Add the movement time, sensor switching time, and preset re-inspection task execution time to obtain the total time; S33: Obtain the load on the robotic arm and calculate the power consumption based on the load on the robotic arm and the total time; S34: Obtain the number of sensor switching based on the task sequence in the immediate execution of the re-inspection task request scheme.

8. The multi-task planning method for a robotic arm according to claim 7, characterized in that, Step S3 also includes: S35: When the deduction scheme is to delay the execution of the re-inspection task request scheme, the movement time is calculated based on the generated path and the moving speed of the robotic arm, and the movement time is added to the sensor switching time to obtain the total time. S36: Calculate the power consumption based on the robotic arm load and the total time; S37: Based on the optimized task sequence in the delayed execution of the re-inspection task request scheme, obtain the number of sensor switching times.

9. A multi-task planning method for a robotic arm according to claim 8, characterized in that, Step S4 also includes: S41: Assign a preset first weight, a second weight, and a third weight to the total time, the power consumption, and the number of sensor switching times, respectively, wherein the second weight is greater than the first weight and the third weight; S42: The total time, power consumption, and number of sensor switching times are weighted and summed according to the first weight, the second weight, and the third weight to obtain a comprehensive score, which represents the total cost. S43: Select the simulation scheme with the smaller total cost and execute it.

10. A multi-task planning system for a robotic arm, characterized in that, The system for implementing the method according to any one of claims 1-9 comprises: Acquisition Module: When the robotic arm performs the initial inspection of the inside of the blade, if there is an uncertain area with an "uncertain" result, the module acquires the re-inspection task request for the uncertain area, the current position of the robotic arm, the current sensor type, and the battery level. The deduction module: Based on the re-inspection task request, the current position of the robotic arm, the current sensor type, and the battery level, it deduces the scheme of immediately executing the re-inspection task request and the scheme of delaying the execution of the re-inspection task request. Calculation module: For each simulation scenario, calculate the total time, power consumption, and number of sensor switching operations; Execution module: Calculates the total cost by combining the total time, power consumption, and number of sensor switching, and selects the simulation scheme with the lowest total cost to execute.

Images