Quadruped robot cooperative mooring unmanned aerial vehicle photovoltaic inspection method and device

Abstract

The invention provides a quadruped robot cooperative mooring unmanned aerial vehicle photovoltaic inspection method and device, and the method comprises the steps: obtaining a global image through a photovoltaic power station fixed monitoring network, and positioning a space coordinate set of a suspected fault region based on a visual analysis algorithm; decomposing the space coordinate set into an ordered inspection sequence by using a ground control station, and issuing path planning parameters to a quadruped robot; the quadruped robot is controlled to autonomously navigate to target coordinates according to the path planning parameters, and the mooring unmanned aerial vehicle is released after the posture is stabilized; executing three-dimensional scanning and multispectral imaging by using the mooring unmanned aerial vehicle, returning a fused sensing data packet in real time, and marking abnormal confidence; and after the diagnosis is completed, controlling the mooring unmanned aerial vehicle to autonomously land to the back of the quadruped robot, and dynamically updating an inspection sequence based on the abnormal confidence until the inspection is completed. And the tethered unmanned aerial vehicle is transported through the quadruped robot, so that the operation in a no-fly area is effectively solved, and the inspection efficiency of the unmanned aerial vehicle on the photovoltaic power station is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle photovoltaic inspection, and particularly relates to a quadruped robot cooperative tethered unmanned aerial vehicle photovoltaic inspection method and device. BACKGROUND

[0002] With the increasing demand for renewable energy worldwide, photovoltaic power stations, as an important part of clean energy, are rapidly expanding in scale and quantity. However, photovoltaic power stations are widely distributed, including some stations located in remote or special geographic areas such as no-fly zones, which poses great challenges to the maintenance and inspection of photovoltaic power stations.

[0003] Therefore, how to improve the inspection efficiency of unmanned aerial vehicles for photovoltaic power stations and the like has become a technical problem that technicians in the field need to solve urgently. SUMMARY

[0004] The present application provides a quadruped robot cooperative tethered unmanned aerial vehicle photovoltaic inspection method and device to solve the defect that the inspection efficiency of photovoltaic power stations in special areas such as no-fly zones is low by using unmanned aerial vehicles in the prior art.

[0005] In a first aspect, the present application provides a quadruped robot cooperative tethered unmanned aerial vehicle photovoltaic inspection method, comprising: acquiring global images through a photovoltaic power station fixed monitoring network, and positioning a spatial coordinate set of a suspected fault area based on a visual analysis algorithm; decomposing the spatial coordinate set into an ordered inspection sequence by using a ground control station, and issuing path planning parameters to a quadruped robot; controlling the quadruped robot to autonomously navigate to a target coordinate according to the path planning parameters, and releasing a tethered unmanned aerial vehicle after stabilizing the posture; performing three-dimensional scanning and multispectral imaging by using the tethered unmanned aerial vehicle, and real-time returning a fusion sensing data packet and marking an abnormal confidence; controlling the tethered unmanned aerial vehicle to autonomously land on the back of the quadruped robot after completing diagnosis, and dynamically updating an inspection sequence based on the abnormal confidence until the inspection is completed.

[0006] According to the quadruped robot cooperative tethered unmanned aerial vehicle photovoltaic inspection method provided by the present application, the autonomous navigation comprises: extracting point cloud curvature features in real time through a tightly coupled laser radar-inertial odometry system; fusing inertial measurement data and point cloud curvature features by using an iterative state estimation algorithm to generate positioning information resistant to motion interference; correcting mileage cumulative error by using context-aware loop detection, outputting a navigation instruction, and performing autonomous navigation.

[0007] According to the four-legged robot cooperative tethered unmanned aerial vehicle photovoltaic inspection method provided by the application, inertial measurement data and point cloud curvature characteristics are fused by using an iterative state estimation algorithm to generate positioning information resistant to motion interference, and the method comprises the following steps: The inertial measurement unit data is pre-integrated to construct a motion compensation model; Based on the motion compensation model, the matching residual of the point cloud and the submap is iteratively optimized by extended Kalman filtering; The optimized results are used to smooth high-frequency noise by using a low-pass filter, and the positioning information resistant to motion interference is output.

[0008] According to the four-legged robot cooperative tethered unmanned aerial vehicle photovoltaic inspection method provided by the application, the tethered unmanned aerial vehicle performs three-dimensional scanning and multispectral imaging, and real-time returns fusion sensing data packets and marks abnormal confidence, and the method comprises the following steps: The surface of the photovoltaic module is micro-deformed and three-dimensionally reconstructed by the laser radar of the tethered unmanned aerial vehicle, and the physical damage topological structure is identified; Synchronously, an infrared thermal imaging module is used to capture the temperature field distribution and locate the hot spot abnormal area; The physical damage topological structure and the hot spot abnormal area are spatially registered to generate a defect correlation graph and calculate a confidence score.

[0009] According to the four-legged robot cooperative tethered unmanned aerial vehicle photovoltaic inspection method provided by the application, the inspection sequence is dynamically updated based on the abnormal confidence, and the method comprises the following steps: When the abnormal confidence exceeds a first threshold value, an emergency diagnosis task is inserted at the first position of the inspection sequence; When an environmental sudden obstacle causes a path interruption, an air-ground cooperative obstacle avoidance mode is started to re-plan a route; Based on a historical fault distribution model, the priority of the coordinate set of the un-inspected area is optimized to complete the update of the inspection sequence.

[0010] According to the four-legged robot cooperative tethered unmanned aerial vehicle photovoltaic inspection method provided by the application, the method further comprises the following steps: The robot posture stability is monitored in real time by a pressure sensor in the charging station, and the ventilation system is triggered to adjust the internal temperature and humidity; The battery swelling deformation and the contact terminal state are captured by using a multispectral monitoring device, and the wireless charging power is dynamically adjusted; Based on the internal temperature and humidity and the wireless charging power, sensor zero-point calibration is performed after charging to compensate for the measurement drift caused by extreme environments.

[0011] According to the four-legged robot cooperative tethered unmanned aerial vehicle photovoltaic inspection method provided by the application, the method further comprises the following steps: The airspace control level to which the target coordinates belong is identified, and a restricted area take-off and landing permission instruction is generated; After the quadruped robot reaches the target coordinates, it deploys an electromagnetic shielding platform to ensure the communication security of the UAV. Upon completion of the mission, the drone's emergency landing protocol is automatically triggered, locking onto the robotic vehicle.

[0012] The photovoltaic inspection method for a quadruped robot-assisted tethered unmanned aerial vehicle provided by the present invention further includes: Real-time monitoring of wind speed, precipitation, and dust concentration to determine risk factors; When the risk coefficient exceeds the second threshold, the drone's hovering height is reduced and the stabilization time is extended, the quadruped robot's terrain attachment enhancement mode is activated simultaneously, and the low center of gravity gait is switched.

[0013] The photovoltaic inspection method for a quadruped robot-assisted tethered unmanned aerial vehicle provided by the present invention further includes: A quadruped robot scans the coordinates of ground obstacles using lidar, and a tethered drone captures the trajectory of aerial obstacles from a top-down perspective, transmitting the data bidirectionally to the ground control station. By integrating air and ground perception data at the control station, a three-dimensional dynamic obstacle map is constructed, and the collision risk coefficient is calculated in real time. When the collision risk coefficient is greater than 0.8, the emergency climb protocol of the tethered drone and the emergency stop protocol of the robot are triggered. When the collision risk coefficient ∈ (0.5, 0.8], the detour path is replanned and the drone hovering diagnosis time is shortened; When the collision risk coefficient is ≤0.5, the original inspection route is maintained and the sensor monitoring frequency is increased.

[0014] Secondly, this invention also protects a photovoltaic inspection device for a quadruped robot-assisted tethered unmanned aerial vehicle, comprising: The acquisition module is used to acquire global images through the fixed monitoring network of the photovoltaic power station and locate the spatial coordinate set of suspected fault areas based on visual analysis algorithms. The decomposition module is used to decompose the spatial coordinate set into an ordered inspection sequence using the ground control station, and then send path planning parameters to the quadruped robot. The transport module is used to control the quadruped robot to autonomously navigate to the target coordinates according to the path planning parameters, and release the tethered drone after stabilizing its posture; The diagnostic module is used to perform three-dimensional scanning and multispectral imaging using the tethered UAV, and to transmit fused sensor data packets back in real time and mark the confidence level of anomalies. The response module is used to control the tethered drone to autonomously land on the back of the quadruped robot after completing the diagnosis, and dynamically update the inspection sequence based on the anomaly confidence level until the inspection is completed.

[0015] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the photovoltaic inspection method of the quadruped robot cooperative tethered drone as described above.

[0016] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the photovoltaic inspection method of a quadruped robot cooperative tethered drone as described above.

[0017] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the photovoltaic inspection method for a quadruped robot cooperative tethered unmanned aerial vehicle as described above.

[0018] The present invention provides a method and apparatus for photovoltaic power plant inspection using a quadruped robot and a tethered drone. The method includes acquiring a global image through a fixed monitoring network of the photovoltaic power plant and locating a set of spatial coordinates for suspected fault areas based on a visual analysis algorithm; decomposing the spatial coordinate set into an ordered inspection sequence using a ground control station and issuing path planning parameters to the quadruped robot; controlling the quadruped robot to autonomously navigate to the target coordinates according to the path planning parameters, and releasing the tethered drone after stabilizing its attitude; using the tethered drone to perform 3D scanning and multispectral imaging, transmitting fused sensor data packets in real time and marking anomaly confidence levels; after completing the diagnosis, controlling the tethered drone to autonomously land on the back of the quadruped robot, and dynamically updating the inspection sequence based on the anomaly confidence levels until the inspection is completed. By using a quadruped robot to transport the tethered drone, the method effectively solves the problem of operation in no-fly zones and improves the efficiency of drone inspections of photovoltaic power plants. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the photovoltaic inspection method for a quadruped robot-assisted tethered drone provided in this embodiment; Figure 2 This is a schematic diagram illustrating the principle of the quadruped robot-assisted tethered drone photovoltaic inspection method provided in this embodiment; Figure 3 This is a schematic diagram of the structure of the quadruped robot collaborative tethered drone photovoltaic inspection device provided in this embodiment; Figure 4This is a schematic diagram of the structure of the electronic device provided in this embodiment. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0022] Figure 1 This is a flowchart illustrating the photovoltaic inspection method for a quadruped robot-assisted tethered drone provided in this embodiment.

[0023] like Figure 1 As shown in the figure, the photovoltaic inspection method of a quadruped robot collaborative tethered UAV provided by the embodiment of the present invention mainly includes the following steps: 101. Obtain global images through the fixed monitoring network of photovoltaic power stations, and locate the spatial coordinate set of suspected fault areas based on visual analysis algorithms.

[0024] In a typical implementation, a fixed monitoring network has been pre-deployed in the photovoltaic power station. This network consists of multiple monitoring cameras distributed in different locations. These cameras have wide-angle shooting capabilities, covering most areas of the photovoltaic power station to obtain a global image of the entire station. Through network transmission technologies such as 4G, WiFi, or Ethernet, the images captured in real time by these cameras are stably and quickly uploaded to the backend computing center (such as a ground control station).

[0025] At the backend computing center, the received global images are imported into a dedicated visual analysis algorithm module for processing. The visual analysis algorithm first preprocesses the images, including image enhancement and denoising, to improve image quality and make subsequent analysis more accurate. Next, the algorithm extracts and identifies features of the photovoltaic modules in the images based on deep learning or traditional image processing techniques. Deep learning algorithms learn from a large number of labeled images of normal and faulty photovoltaic modules to build a model capable of accurately identifying the different states of photovoltaic modules. Traditional image processing techniques utilize features such as grayscale, texture, and shape to distinguish between normal and abnormal regions.

[0026] During the identification process, the algorithm compares the features of the photovoltaic modules in the current image with pre-set normal standard features. Once it finds that the features of certain areas deviate significantly from the normal features, these areas will be identified as suspected fault areas. For example, if it is found that the surface temperature of a photovoltaic module in a certain area is abnormally high, exhibiting different thermal radiation characteristics from the surrounding area in the infrared image, or showing abnormal color or texture in the visible light image, such as black spots or cracks, that area will be marked as a suspected fault area.

[0027] For the marked suspected fault areas, the algorithm further determines their coordinates in the actual photovoltaic power plant space through coordinate transformation and calculation. Since the position and shooting angle of the monitoring camera are known, the algorithm uses the camera's calibration parameters, combined with geometric methods such as triangulation, to convert the pixel coordinates in the image into three-dimensional coordinates in actual space. By calculating the coordinates of multiple suspected fault areas, a spatial coordinate set of the suspected fault areas is finally formed.

[0028] By acquiring global images through a fixed monitoring network, a comprehensive and macroscopic understanding of the overall condition of the photovoltaic power station can be obtained, without overlooking any potential fault areas. Suspected fault areas can be quickly filtered from a large amount of image data, and their spatial coordinates can be accurately located. This provides accurate target information for subsequent targeted inspections by quadruped robots and tethered drones, greatly improving the efficiency and accuracy of photovoltaic inspections, reducing the workload and false alarm rate of manual inspections, and enabling timely detection and handling of potential faults in photovoltaic power stations, ensuring their stable and efficient operation.

[0029] 102. Use the ground control station to decompose the spatial coordinate set into an ordered inspection sequence and send path planning parameters to the quadruped robot.

[0030] Specifically, the ground control station first receives the spatial coordinate set of suspected fault areas output by the visual analysis module. These coordinates precisely correspond to key locations in the photovoltaic power station that need to be investigated. Based on the digital twin map of the photovoltaic power station (including static obstacle information such as component array distribution, channel width, and equipment base), the control station initiates a sequence decomposition algorithm: first, it groups the spatial coordinates by array blocks through cluster analysis, allowing suspected fault points in the same area to be processed centrally, reducing the invalid travel of the robot across areas and initially improving inspection efficiency; then, combined with the motion parameters of the quadruped robot, it eliminates abnormal points in the coordinate set that exceed the robot's passage capability (such as misjudged coordinates located on top of high walls), avoiding resource waste caused by issuing invalid tasks.

[0031] After grouping, the control station uses an improved genetic algorithm to generate an ordered inspection sequence: a path network is constructed with the center point of each block as nodes, and the shortest traversal path is obtained through iterative calculation. For example, in a certain range of photovoltaic array area, the algorithm can automatically avoid narrow channels between component supports and prioritize wider maintenance channels as connecting paths, thus shortening the inspection path within a single block. At the same time, the sequence is sorted according to the principle of "easy first, difficult later", placing the coordinates of flat areas first and the coordinates of sloping areas last, matching the robot's power consumption characteristics and extending the single-cycle endurance.

[0032] During the path planning parameter generation phase, the control station calculates 3D path parameters for each coordinate point based on the inspection sequence: Laterally, a UTM coordinate system transformation is used to convert latitude and longitude coordinates into X / Y axis displacements in the robot's local coordinate system; longitudinally, combined with pre-built terrain elevation data from the LiDAR, a Z-axis height compensation value is generated (e.g., setting corresponding obstacle-crossing parameters for cable trenches at a certain height). The parameters also include a dynamic obstacle avoidance trigger threshold; when the real-time distance between the robot and a preset obstacle point is less than a set value, an alternative path segment is automatically activated. These parameters ensure a significant improvement in the robot's success rate in navigating complex terrain.

[0033] Finally, the ground control station packages and sends out the parameters via an industrial-grade 5G module. The data packet includes a sequence number, coordinate parameters, path priority, and energy consumption warning threshold. After receiving the data, the quadruped robot's motion controller automatically parses the parameters and drives the joint actuators to complete the inspection in sequence. The entire process, from receiving the coordinate set to sending the parameters, is short, enabling rapid response to the inspection task and allowing sufficient time for subsequent detailed inspection by tethered drones. Ultimately, this reduces the total time required for fault location in the photovoltaic power station and lowers the missed detection rate.

[0034] 103. Control the quadruped robot to autonomously navigate to the target coordinates according to the path planning parameters, and release the tethered drone after stabilizing its posture.

[0035] Specifically, after the quadruped robot initiates autonomous navigation, it first collects real-time environmental data of the photovoltaic power station through a tightly coupled lidar-inertial odometry system. The lidar scans the surrounding photovoltaic module array, supports, channels, and other scenes, extracting curvature features from the point cloud data. These features accurately reflect key terrain structures in the photovoltaic power station, such as module edges and support corners, providing a stable environmental reference benchmark for subsequent positioning and reducing feature loss caused by interference such as reflections from the photovoltaic panel surface.

[0036] Simultaneously, the inertial measurement unit (IMU) collects acceleration and angular velocity data during the robot's movement and performs pre-integration processing. The pre-integration results are combined with the robot's leg kinematics model to construct a motion compensation model. This model can correct measurement deviations caused by the robot's gait swaying in real time, ensuring that the point cloud data and the robot's motion state are kept in spatiotemporal synchronization, thus avoiding environmental perception distortion caused by body shaking.

[0037] Based on the motion compensation model described above, the system optimizes the matching between the LiDAR point cloud and the preset sub-map using an extended Kalman filter. During the filtering process, the matching residual between the point cloud and the sub-map is continuously calculated iteratively, and the robot pose estimation value is adjusted according to the residual. This effectively eliminates the positioning ambiguity caused by similar scenes within the photovoltaic power station (such as repeatedly arranged photovoltaic modules) and improves the stability of pose estimation.

[0038] The optimized pose data will be processed by a low-pass filter to filter out high-frequency noise caused by the impact of the robot's legs contacting the ground, and output smooth anti-motion interference positioning information to ensure that the positioning results are not affected by instantaneous vibration when the robot walks on uneven inspection channels or between components in photovoltaic power stations, and maintains the stability of the navigation path.

[0039] Furthermore, the system further corrects accumulated mileage errors through context-aware loop closure detection. When the robot travels to a previously visited area (such as the corner of a component array), the loop closure detection algorithm compares the current scene features with historical data. If it determines that the same location has been reached, error correction is triggered to avoid path deviations caused by the accumulation of small errors after long-term navigation, ensuring that the robot can accurately reach each target coordinate.

[0040] Based on the navigation commands generated by the above processing, the quadruped robot drives its leg joints to adjust its step frequency and stride length, autonomously avoiding obstacles such as photovoltaic supports and cables, and moving towards the target coordinates according to the planned path. Once the robot reaches the target coordinates and its attitude sensors confirm stability, its onboard tethered drone release mechanism unlocks, slowly releasing the tethered drone. The entire autonomous navigation process is highly adapted to the complex environment of photovoltaic power plants, ensuring the robot efficiently navigates through dense arrays of components, accurately reaching suspected fault areas, and providing a stable take-off and landing platform for subsequent close-range precision inspection of the tethered drone, thus improving the continuity and reliability of the overall inspection task.

[0041] 104. Utilize tethered UAVs to perform 3D scanning and multispectral imaging, and transmit fused sensor data packets back in real time and mark the confidence level of anomalies.

[0042] Specifically, after the tethered drone arrives above the target coordinates and adjusts to a stable hovering state, its onboard LiDAR begins to work, performing a dense scan of the surface of the photovoltaic modules below. The laser beam emitted by the LiDAR covers the module surface point by point, and the distance between each point is calculated by recording the time difference of the laser echo, thereby generating massive point cloud data. The system stitches and models this point cloud data to complete the three-dimensional reconstruction of the micro-deformation of the photovoltaic module surface, clearly showing whether there are physical damages such as cracks, dents, and bulges in the module, and constructing the topological structure of the damaged area. This three-dimensional reconstruction method can accurately capture subtle physical defects and avoid misjudgments caused by planar images due to perspective deviations, providing three-dimensional and intuitive basic data for subsequent damage analysis.

[0043] While the lidar performs 3D scanning, the infrared thermal imaging module on the tethered drone is simultaneously activated to detect thermal radiation in the photovoltaic module area and capture the temperature field distribution on the module surface. Infrared thermal imaging senses the differences in infrared radiation intensity in different areas and presents temperature changes in the form of a thermal map, thereby quickly locating hot spots with abnormally high temperatures. Compared with traditional visible light imaging, infrared thermal imaging can penetrate interference such as dust and stains on the module surface, directly reflecting thermal anomalies caused by circuit faults or microcracks inside the module, thus improving the accuracy of hot spot detection.

[0044] The system then spatially registers the physical damage topology with the hot spot anomaly region. By calibrating the 3D coordinates acquired by the lidar with the pixel coordinates of the infrared thermography, the physical damage location and the hot spot location correspond in the same spatial coordinate system, generating a defect correlation map. The map clearly marks the overlapping areas and adjacency relationships between the physical damage and the hot spot. Simultaneously, by combining the confidence parameters of the two detection methods (such as the matching accuracy of the lidar point cloud and the stability of infrared thermography), an algorithm calculates a comprehensive confidence score for each anomaly region. Spatial registration eliminates spatial biases between different sensor data, making the correlation analysis between physical damage and hot spots more scientific. The confidence score helps the subsequent system quickly distinguish high-priority suspected faults, improving fault screening efficiency.

[0045] After completing the above processing, the tethered UAV will transmit the sensor data packet, which integrates 3D reconstruction data, infrared thermal imaging data, defect correlation map and confidence score, back to the ground control station in real time via wireless communication. This ensures the stability and real-time performance of data transmission, and ensures that the ground control station can obtain complete and accurate detection data in a timely manner, providing strong support for the final judgment and handling decision of the fault.

[0046] 105. After completing the diagnosis, control the tethered drone to autonomously land on the back of the quadruped robot, and dynamically update the inspection sequence based on the anomaly confidence level until the inspection is completed.

[0047] Specifically, after the tethered UAV completes 3D scanning and multispectral imaging diagnostics, the ground control station issues a return command. Based on the real-time positioning information transmitted back by the quadruped robot, and combining its own positioning module and visual navigation module, the UAV plans the optimal landing path. Upon approaching the quadruped robot's back, it identifies the positioning markers on the landing platform using onboard visual sensors, adjusts its attitude and altitude in real time, and finally lands precisely on the platform. The landing mechanism automatically locks, completing the recovery. This autonomous landing process achieves efficient collaborative recovery between the UAV and the robot, avoiding human intervention, ensuring the safe storage of equipment in the complex environment of the photovoltaic power station, and laying the foundation for the continuous execution of subsequent inspection tasks.

[0048] Meanwhile, the ground control station initiates a dynamic update mechanism for the inspection sequence based on the anomaly confidence data transmitted back by the tethered drone. When the anomaly confidence level of a certain area exceeds the first threshold, the system determines that there is a high-risk fault in that area and immediately inserts the corresponding emergency diagnostic task at the top of the current inspection sequence, prioritizing the dispatch of the quadruped robot and drone for verification. This priority adjustment ensures that potentially serious faults are quickly confirmed a second time, reducing the risk of fault escalation and improving the emergency response capability of the inspection.

[0049] If the quadruped robot encounters a sudden environmental obstacle (such as temporarily piled-up maintenance materials or sudden equipment tipping over) that interrupts its path, the ground control station immediately activates the air-ground cooperative obstacle avoidance mode. The quadruped robot uses lidar to detect the obstacle range in real time and uploads the data to the control station. Simultaneously, the tethered drone scans the obstacle area from the air, constructing a 3D model of the obstacle. The control station combines the data from both systems to replan the obstacle avoidance route, update the path parameters, and send them back to the quadruped robot. Air-ground cooperative obstacle avoidance fully leverages the robot's ground detail perception and the drone's global aerial vision, ensuring the continuity of the inspection path even in complex and unexpected scenarios and preventing task interruption.

[0050] Furthermore, the system prioritizes the coordinate set of uninspected areas based on a historical fault distribution model. By analyzing past fault types and high-incidence areas of the photovoltaic power station, the weights of currently uninspected suspected fault points are adjusted, and the coordinate points in historically frequent fault areas are given higher priority. This dynamic optimization based on historical data tilts inspection resources towards high-risk areas, improving the targeting and efficiency of fault detection and avoiding resource waste caused by indiscriminate inspections.

[0051] Through this mechanism, the ground control station continuously and dynamically updates the inspection sequence until all suspected fault areas have been inspected. The entire process enables adaptive adjustment of inspection tasks, allowing the system to maintain efficient operation even in the face of complex environments and emergencies, ultimately ensuring comprehensive coverage and accurate diagnosis of the photovoltaic power station and guaranteeing its stable operation.

[0052] In this embodiment, a quadruped robot carrying a tethered drone is used. First, a preliminary inspection of the photovoltaic power station is conducted using fixed monitoring cameras to identify potential fault points. Then, the quadruped robot carries the tethered drone to a designated location for further confirmation of the fault, transmitting images and video information back to the control room in real time. A key feature is that the quadruped robot not only possesses autonomous navigation and obstacle avoidance capabilities but also serves as a mobile platform providing stable support for the tethered drone. This enables the drone to perform precise inspection tasks at high altitudes or in complex terrain, significantly improving inspection efficiency and coverage, and ensuring the safe operation of the photovoltaic power station. This system design cleverly combines the flexibility of a ground robot with the wide field of vision of an aerial drone, providing an efficient solution for photovoltaic power station inspection.

[0053] Furthermore, based on the above embodiments, this embodiment also includes: real-time monitoring of the robot's posture stability by a pressure sensor inside the charging station, triggering the ventilation system to adjust the internal temperature and humidity; using multispectral monitoring equipment to capture battery expansion deformation and contact terminal status, dynamically adjusting the wireless charging power; and performing sensor zero-point calibration after charging is completed based on the internal temperature and humidity and the wireless charging power to compensate for measurement drift caused by extreme environments.

[0054] Specifically, when the quadruped robot returns to the charging station after completing its phased inspection task, the pressure sensors deployed within the station immediately activate, monitoring the pressure distribution on the contact surface between the robot and the charging platform in real time. By analyzing the uniformity and stability of the pressure data, the system can determine whether the robot's posture remains stable. If an abnormal pressure distribution is detected, it indicates that the robot's posture is tilted or deviated, at which point the charging station automatically triggers the ventilation system. The ventilation system adjusts the airflow and speed according to preset temperature and humidity thresholds within the station, quickly balancing the internal temperature and humidity to ensure the robot charges in a stable posture and a suitable environment. This avoids poor charging contact caused by unstable posture and prevents damage to the equipment due to abnormal temperature and humidity, ensuring the safety and reliability of the charging process.

[0055] During charging, the charging station's multispectral monitoring equipment continuously monitors the robot's battery and contact terminals. By capturing light signals in different wavelengths, the equipment can clearly identify whether the battery has expanded or deformed, as well as the oxidation and wear of the contact terminals. Based on the monitored degree of battery expansion and the condition of the contact terminals, the system dynamically adjusts the wireless charging power. For example, if slight battery expansion or poor contact terminal condition is detected, the charging power is appropriately reduced to decrease heat generation; if the condition is good, the normal charging power is maintained. This dynamic adjustment mechanism adapts to the real-time status of the battery and terminals, avoiding battery damage or low charging efficiency due to improper charging power, extending battery life, and ensuring high efficiency in the charging process.

[0056] After charging is complete, the system initiates a sensor zero-point calibration procedure based on the internal temperature and humidity data recorded during charging and the dynamically adjusted wireless charging power information. Due to the complex environment of photovoltaic power plants, extreme high and low temperatures, humidity variations, etc., can cause measurement drift in various sensors on the robot (such as LiDAR and inertial measurement units). The calibration procedure corrects the zero-point parameters of the sensors by calling a preset calibration model and considering the influence of temperature, humidity, and charging power on the sensors. This calibration process effectively compensates for measurement deviations caused by extreme environmental factors, ensuring that the sensors can provide accurate measurement data in subsequent inspection tasks. This provides reliable support for the robot's autonomous navigation and the drone's precise detection, improving the overall detection accuracy of the inspection system.

[0057] Through the above mechanism, the charging station not only provides a stable and safe charging environment for the quadruped robot, but also ensures the performance stability of the equipment in complex environments through dynamic adjustment and calibration, providing strong support for the continuous and efficient operation of the entire photovoltaic inspection system and further guaranteeing the smooth progress of photovoltaic power station inspection tasks.

[0058] Furthermore, based on the above embodiments, this embodiment also includes: identifying the airspace control level to which the target coordinates belong and generating a restricted area take-off and landing permit; deploying an electromagnetic shielding platform after the quadruped robot arrives at the target coordinates to ensure the communication safety of the UAV; and automatically triggering the UAV emergency landing protocol and locking onto the robot vehicle after the mission is completed.

[0059] Specifically, when generating the inspection sequence, the ground control station simultaneously accesses a pre-set airspace control database to identify the airspace control level of each target coordinate. The database stores airspace delineation information for the photovoltaic power station and its surrounding area, including flight restrictions, time periods, and permit requirements for different areas. The system determines the corresponding airspace control level by comparing the target coordinates with the airspace boundary parameters in the database. If the target coordinates are in restricted airspace, the ground control station automatically initiates a takeoff and landing permit application to the management platform. After obtaining authorization, a restricted area takeoff and landing permit instruction is generated, containing restrictions such as permitted flight time periods and altitude ranges. This process ensures that the UAV conducts inspections within a compliant framework, avoiding safety risks caused by unauthorized entry into controlled airspace and guaranteeing the legality and safety of airspace use.

[0060] After the quadruped robot arrives at the target coordinates according to the path planning parameters, its onboard electromagnetic shielding platform automatically deploys. This platform activates its built-in shielding module to create a specific electromagnetic protection zone, blocking the influence of complex external electromagnetic environments (such as high-voltage equipment in photovoltaic power plants, radio interference sources, etc.) on the drone. Simultaneously, the platform's built-in signal enhancement unit strengthens the communication signal between the drone and the ground control station. The electromagnetic shielding platform creates a stable communication environment for the drone, preventing command transmission interruptions or data loss caused by external electromagnetic interference, ensuring that the drone can accurately perform inspection tasks even in areas with strong electromagnetic interference.

[0061] When the tethered drone completes its inspection mission, or when the ground control station detects an anomaly (such as low battery or signal interruption), the system automatically triggers the drone's emergency landing protocol. Upon activation, the drone immediately terminates its current mission and returns to the quadrupedal robot carrier along a pre-set landing path. The quadrupedal robot uses a visual positioning module to capture the drone's position in real time, guiding it to land precisely within the carrier's locking area. After landing, the carrier's mechanical locking mechanism automatically engages, securely locking the drone. This emergency landing protocol and locking mechanism enable rapid and safe recovery of the drone in emergencies, preventing equipment loss or damage and ensuring the integrity of the inspection equipment and the continuity of subsequent missions.

[0062] Through the above mechanism, the system realizes closed-loop control of airspace compliance management, communication security assurance and emergency equipment recovery, further improving the safety and reliability of photovoltaic inspection tasks in complex environments, and ensuring that the entire collaborative system operates efficiently under the premise of compliance with regulations.

[0063] Furthermore, based on the above embodiments, this embodiment also includes: real-time monitoring of wind speed, precipitation and dust concentration to determine the risk coefficient; when the risk coefficient exceeds the second threshold, reducing the hovering height of the drone and extending the stabilization time, simultaneously activating the quadruped robot terrain attachment enhancement mode, and switching to a low center of gravity gait.

[0064] Specifically, both the quadruped robot and the tethered drone are equipped with weather sensor modules to monitor meteorological parameters such as wind speed, precipitation, and dust concentration in the operating environment in real time. These parameters are continuously transmitted to the ground control station, which uses a pre-set risk assessment model to comprehensively analyze the data and calculate the current environmental risk coefficient. This model converts each meteorological parameter into a corresponding risk value based on its weight in relation to the equipment's operation, and then accumulates these values ​​to form a quantifiable risk coefficient. Real-time meteorological monitoring and risk assessment provide data support for the system to cope with complex weather conditions, ensuring that potential threats to operations from environmental changes can be detected in a timely manner.

[0065] When the ground control station determines that the calculated risk coefficient exceeds the second threshold, it immediately issues a coordinated response command to the tethered drone and the quadruped robot. For the tethered drone, the control module automatically lowers its hovering altitude after responding to the command, reducing the impact of strong high-altitude winds on the drone's stability. Simultaneously, it extends the hovering stabilization time and continuously adjusts the rotor speed via attitude sensors to ensure the drone remains relatively stationary under airflow disturbances. This adjustment reduces the risk of the drone crashing or colliding with photovoltaic modules due to weather factors, ensuring the drone's operational safety under complex weather conditions.

[0066] Simultaneously, the quadruped robot activates its terrain adhesion enhancement mode, and its leg drive system switches to a low center of gravity gait according to instructions. By adjusting the flexion and extension angles of the leg joints, the overall center of gravity of the robot is lowered, while the contact area between the feet and the ground is increased, enhancing friction. During walking, the robot's force feedback sensors perceive the ground adhesion in real time and dynamically adjust the support force of each leg. The low center of gravity gait and terrain adhesion enhancement mode significantly improve the robot's walking stability on wet or dusty photovoltaic power station grounds, preventing equipment damage and task interruption due to slipping or tipping.

[0067] Through the aforementioned collaborative response mechanism, the system can quickly adjust the operating status of the equipment when weather conditions deteriorate, enabling the quadruped robot and tethered drone to maintain a certain level of operational capability in complex weather environments. This ensures the continuity of photovoltaic inspection tasks and the safety of the equipment, and enhances the system's adaptability to harsh environments.

[0068] Furthermore, based on the above embodiments, this embodiment also includes: using a quadruped robot to scan the coordinates of ground obstacles with LiDAR, using a tethered drone to capture the trajectory of aerial obstacles from a top-down perspective, and transmitting the data bidirectionally to the ground control station; using the control station to fuse air-ground perception data to construct a three-dimensional dynamic obstacle map and calculate the collision risk coefficient in real time; when the collision risk coefficient > 0.8, triggering the tethered drone's emergency climb and the robot's emergency stop protocol; when the collision risk coefficient ∈ (0.5, 0.8], replanning the detour path and shortening the drone's hovering diagnosis time; when the collision risk coefficient ≤ 0.5, maintaining the original inspection route and increasing the sensor monitoring frequency.

[0069] Specifically, during autonomous navigation, the quadruped robot's onboard LiDAR continuously scans the ground, capturing the coordinates of obstacles in real time. These obstacles may include maintenance tools within the photovoltaic power station, fallen component fragments, and so on. Simultaneously, the tethered drone, while performing inspection tasks, uses its onboard visual sensors to monitor aerial dynamics from a top-down perspective, capturing the movement trajectories of aerial obstacles such as birds and falling debris. Both the quadruped robot and the tethered drone transmit their respective obstacle information bidirectionally to the ground control station via a communication module. This collaborative air-ground obstacle perception method comprehensively covers potential obstacles on the ground and in the air, avoiding omissions due to a single perspective and providing a comprehensive data foundation for subsequent collision risk assessment.

[0070] After receiving air-to-ground obstacle data, the ground control station initiates a data fusion algorithm to process the data. The algorithm unifies the static coordinates of ground obstacles with the dynamic trajectories of air obstacles into the same three-dimensional coordinate system, constructing a three-dimensional dynamic obstacle map containing information such as obstacle position, size, and motion state. Based on this map, the control station calculates the collision risk coefficient between the quadruped robot and tethered drone and the obstacles in real time using a pre-set collision risk assessment model. This coefficient comprehensively considers factors such as the distance between the obstacle and the equipment, and the relative speed of movement. The construction of the three-dimensional dynamic obstacle map enables accurate depiction of obstacles, while the real-time calculation of the collision risk coefficient provides a quantitative basis for the system to take targeted countermeasures.

[0071] When the ground control station determines that the collision risk coefficient is greater than 0.8, it indicates an extremely high risk of collision and immediately triggers the emergency climb protocol for the tethered drone and the emergency stop protocol for the quadruped robot. The tethered drone rapidly increases its flight altitude to avoid aerial obstacles, while the quadruped robot immediately stops moving and resumes operation only after the obstacle threat has been eliminated. This measure can quickly avoid collisions in emergency situations and minimize the risk of equipment damage.

[0072] When the collision risk coefficient is between 0.5 and 0.8, it indicates a certain risk of collision but not to an emergency level. The ground control station initiates a path replanning procedure. Combining a 3D dynamic obstacle map, new detour paths are planned for the quadruped robot and the tethered drone. Simultaneously, the hovering diagnosis time of the tethered drone at each inspection point is shortened. By replanning the path to avoid collision risks, and shortening the hovering time, the probability of encountering obstacles is reduced while ensuring the progress of the inspection task.

[0073] When the collision risk coefficient is less than or equal to 0.5, it indicates that the collision risk is low. The ground control station controls the quadruped robot and tethered drone to maintain the original inspection route, while increasing the monitoring frequency of the sensors and improving the sensitivity to the dynamic changes of obstacles. Under the premise of ensuring inspection efficiency, the status of obstacles can be grasped in a timely manner by strengthening monitoring to ensure that the risk will not escalate further.

[0074] Through the above process, the system achieves comprehensive perception, accurate assessment and dynamic response to obstacles, effectively ensuring the inspection safety of quadruped robots and tethered drones in the complex environment of photovoltaic power plants, and improving the anti-interference capability and mission reliability of the entire system.

[0075] Figure 2 This is a schematic diagram illustrating the principle of the quadruped robot-assisted tethered drone photovoltaic inspection method provided in this embodiment.

[0076] like Figure 2 As shown, the quadruped robot has a combined operating endurance of 4-6 hours, standing dimensions of approximately 1098mm × 450mm × 645mm, and a weight of approximately 60kg (including battery). It possesses autonomous navigation and obstacle avoidance capabilities, enabling flexible movement in complex terrain. The quadruped robot and the tethered drone maintain real-time communication via a link. When the drone detects targets or anomalies in the air, it immediately sends relevant information to the quadruped robot and the ground control station, allowing the quadruped robot to adjust its route or perform corresponding tasks based on the received information. The diagram shows a wired connection; in actual applications, both wired and wireless connections are possible. The tethered drone, mounted on the quadruped robot, can take off after reaching a specific area to perform inspection tasks and then safely land on the quadruped robot's back to continue to the next inspection point. The multi-rotor drone is designed to be 50cm long, 50cm wide, and 20cm high, capable of carrying a load of at least 3kg. It is equipped with a 12W LiDAR and a 4-megapixel camera, with an effective detection range of 25m.

[0077] Based on the same general inventive concept, this invention also protects a photovoltaic inspection device for a quadruped robot collaborative tethered drone. The photovoltaic inspection device for a quadruped robot collaborative tethered drone described below and the photovoltaic inspection method for a quadruped robot collaborative tethered drone described above can be referred to in correspondence with each other.

[0078] Figure 3 This is a schematic diagram of the structure of the quadruped robot collaborative tethered drone photovoltaic inspection device provided in this embodiment.

[0079] like Figure 3 As shown in the figure, this embodiment provides a photovoltaic inspection device for a quadruped robot-assisted tethered unmanned aerial vehicle, comprising: The acquisition module 301 is used to acquire global images through the fixed monitoring network of the photovoltaic power station and locate the spatial coordinate set of suspected fault areas based on visual analysis algorithms. The decomposition module 302 is used to decompose the spatial coordinate set into an ordered inspection sequence using the ground control station and send path planning parameters to the quadruped robot. The transport module 303 is used to control the quadruped robot to autonomously navigate to the target coordinates according to the path planning parameters, and release the tethered drone after stabilizing its posture; The diagnostic module 304 is used to perform three-dimensional scanning and multispectral imaging using the tethered UAV, and to transmit fused sensor data packets back in real time and mark the confidence level of anomalies. The response module 305 is used to control the tethered drone to autonomously land on the back of the quadruped robot after completing the diagnosis, and dynamically update the inspection sequence based on the anomaly confidence level until the inspection is completed.

[0080] Figure 4 This is a schematic diagram of the structure of the electronic device provided in this embodiment.

[0081] like Figure 4 As shown, the electronic device may include a processor 410, a communication interface 420, a memory 430, and a communication bus 440. The processor 410, communication interface 420, and memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions from the memory 430 to execute a quadruped robot collaborative tethered UAV photovoltaic inspection method.

[0082] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0083] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the quadruped robot cooperative tethered UAV photovoltaic inspection method provided by the above methods.

[0084] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is implemented to perform the quadruped robot cooperative tethered UAV photovoltaic inspection method provided by the above methods.

[0085] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0086] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for photovoltaic inspection using a quadruped robot-assisted tethered unmanned aerial vehicle (UAV), characterized in that, include: Global images are acquired through a fixed monitoring network of a photovoltaic power station, and the spatial coordinate set of suspected fault areas is located based on visual analysis algorithms. The spatial coordinate set is decomposed into an ordered inspection sequence using a ground control station, and path planning parameters are sent to the quadruped robot. Control the quadruped robot to autonomously navigate to the target coordinates according to the path planning parameters, and release the tethered drone after stabilizing its posture; The tethered UAV is used to perform 3D scanning and multispectral imaging, and the fused sensor data packets are transmitted back in real time and anomaly confidence levels are marked. After completing the diagnosis, the tethered drone is controlled to autonomously land on the back of the quadruped robot, and the inspection sequence is dynamically updated based on the anomaly confidence level until the inspection is completed.

2. The method for photovoltaic inspection using a quadruped robot-assisted tethered unmanned aerial vehicle according to claim 1, characterized in that, The autonomous navigation includes: Point cloud curvature features are extracted in real time using a tightly coupled lidar-inertial odometry system. By using an iterative state estimation algorithm to fuse inertial measurement data and point cloud curvature features, positioning information resistant to motion interference can be generated. Context-aware loop closure detection corrects cumulative mileage errors, outputs navigation commands, and enables autonomous navigation.

3. The method for photovoltaic inspection by a quadruped robot-assisted tethered unmanned aerial vehicle according to claim 2, characterized in that, The method of using an iterative state estimation algorithm to fuse inertial measurement data and point cloud curvature features to generate motion-resistant positioning information includes: Pre-integration processing is performed on the inertial measurement unit data to construct a motion compensation model; Based on the motion compensation model, the matching residuals between the point cloud and the sub-map are iteratively optimized by extended Kalman filtering; Using the optimization results, a low-pass filter is employed to smooth high-frequency noise and output positioning information that is resistant to motion interference.

4. The method for photovoltaic inspection by a quadruped robot-assisted tethered unmanned aerial vehicle according to claim 1, characterized in that, The process of using the tethered UAV to perform 3D scanning and multispectral imaging, and to transmit fused sensor data packets back in real time and mark anomaly confidence levels includes: The lidar of the tethered drone is used to perform micro-deformation three-dimensional reconstruction of the photovoltaic module surface and identify the physical damage topology. Simultaneously, an infrared thermal imaging module is used to capture the temperature field distribution and locate abnormal hot spot areas; Spatial registration is performed between the physical damage topology and the hot spot anomaly region to generate a defect correlation map and calculate the confidence score.

5. The method for photovoltaic inspection by a quadruped robot-assisted tethered unmanned aerial vehicle according to claim 1, characterized in that, The dynamic updating of the inspection sequence based on the anomaly confidence level includes: When the anomaly confidence level exceeds the first threshold, an emergency diagnostic task is inserted at the beginning of the inspection sequence. When a sudden environmental obstacle causes the path to be interrupted, the air-ground cooperative obstacle avoidance mode is activated to re-plan the route. Based on the historical fault distribution model, the priority of the coordinate set of uninspected areas is optimized to complete the inspection sequence update.

6. The method for photovoltaic inspection using a quadruped robot-assisted tethered unmanned aerial vehicle according to any one of claims 1-5, characterized in that, Also includes: The robot's posture stability is monitored in real time by pressure sensors inside the charging station, triggering the ventilation system to adjust the internal temperature and humidity. Multispectral monitoring equipment is used to capture battery expansion deformation and contact terminal status, and the wireless charging power is dynamically adjusted accordingly. Based on the internal temperature and humidity and the wireless charging power, sensor zero-point calibration is performed after charging is completed to compensate for measurement drift caused by extreme environments.

7. The method for photovoltaic inspection using a quadruped robot-assisted tethered unmanned aerial vehicle according to any one of claims 1-5, characterized in that, Also includes: Identify the airspace control level to which the target coordinates belong and generate restricted area take-off and landing clearance instructions; After the quadruped robot reaches the target coordinates, it deploys an electromagnetic shielding platform to ensure the communication security of the UAV. Upon completion of the mission, the drone's emergency landing protocol is automatically triggered, locking onto the robotic vehicle.

8. The method for photovoltaic inspection using a quadruped robot-assisted tethered unmanned aerial vehicle according to any one of claims 1-5, characterized in that, Also includes: Real-time monitoring of wind speed, precipitation, and dust concentration to determine risk factors; When the risk coefficient exceeds the second threshold, the drone's hovering height is reduced and the stabilization time is extended, the quadruped robot's terrain attachment enhancement mode is activated simultaneously, and the low center of gravity gait is switched.

9. The method for photovoltaic inspection by a quadruped robot-assisted tethered unmanned aerial vehicle according to any one of claims 1-5, characterized in that, Also includes: A quadruped robot scans the coordinates of ground obstacles using lidar, and a tethered drone captures the trajectory of aerial obstacles from a top-down perspective, transmitting the data bidirectionally to the ground control station. By integrating air and ground perception data at the control station, a three-dimensional dynamic obstacle map is constructed, and the collision risk coefficient is calculated in real time. When the collision risk coefficient is greater than 0.8, the emergency climb protocol of the tethered drone and the emergency stop protocol of the robot are triggered. When the collision risk coefficient ∈ (0.5, 0.8], the detour path is replanned and the drone hovering diagnosis time is shortened; When the collision risk coefficient is ≤0.5, the original inspection route is maintained and the sensor monitoring frequency is increased.

10. A photovoltaic inspection device for a quadruped robot-assisted tethered unmanned aerial vehicle, characterized in that, include: The acquisition module is used to acquire global images through the fixed monitoring network of the photovoltaic power station and locate the spatial coordinate set of suspected fault areas based on visual analysis algorithms. The decomposition module is used to decompose the spatial coordinate set into an ordered inspection sequence using the ground control station, and then send path planning parameters to the quadruped robot. The transport module is used to control the quadruped robot to autonomously navigate to the target coordinates according to the path planning parameters, and release the tethered drone after stabilizing its posture; The diagnostic module is used to perform three-dimensional scanning and multispectral imaging using the tethered UAV, and to transmit fused sensor data packets back in real time and mark the confidence level of anomalies. The response module is used to control the tethered drone to autonomously land on the back of the quadruped robot after completing the diagnosis, and dynamically update the inspection sequence based on the anomaly confidence level until the inspection is completed.

Images