Robotic work methods, systems, devices, and media for high altitude steel structures

Abstract

This application provides a method, system, device, and medium for robotic operations on high-altitude steel structures. The method includes the following steps: calculating the current pose based on the matching of a 3D building information model and point cloud data of the steel structure to be operated, and generating the robot's movement, operation, and return trajectories; controlling multiple gripping units of the robot to cyclically execute release and fixation actions, moving to a preset work point; during movement and operation, calculating torque compensation values ​​based on real-time environmental interference data, and adjusting the output torque of the multi-joint servo motors; after mechanical locking at the work point, controlling the modular end effector to execute task commands in conjunction with force control servo parameters; after the task is completed, switching to the flaw detection end, acquiring and outputting quality inspection and work position data; and after receiving a qualified confirmation command, releasing the lock and controlling the robot to move along the return trajectory. This application achieves autonomous climbing, dynamic disturbance rejection, compliant operation, and a closed-loop quality inspection system under complex high-altitude working conditions.

Description

Technical Field

[0001] This application relates to the field of building construction technology, and in particular to a robot operation method, system, equipment and medium for high-altitude steel structures. Background Technology

[0002] With the rapid development of intelligent building construction and intelligent steel structure manufacturing, the workload of complex projects such as high-rise buildings and large-span spatial structures continues to rise, and the application scope of steel structures is constantly expanding. How to efficiently and autonomously execute and manage the multiple processes of welding, fastening, and testing of steel structures under high-altitude conditions has become a core requirement for ensuring personnel safety and improving construction quality and efficiency from the root.

[0003] In existing technologies, the construction of high-altitude steel structures is typically accomplished through manual climbing combined with fixed robotic arms or drone-assisted equipment. For example, large robotic arms are fixed to specific work positions for localized welding operations, or drones are used for non-contact visual inspections before and after the operation. However, in practical applications, existing technologies lack human-like bionic climbing mechanisms and dynamic center of gravity adjustment mechanisms. This poses risks for fully autonomous operations in complex high-altitude environments such as strong winds, structural vibrations, and narrow gaps between steel beams. Risks include equipment instability and falls, as well as operational interruptions or missed welds and inspections due to insufficient obstacle-crossing and avoidance capabilities. Summary of the Invention

[0004] In view of this, this application provides a robot operation method, system, equipment and medium for high-altitude steel structures to solve the above problems.

[0005] Firstly, a method for robotic operation on high-altitude steel structures is provided, the method comprising:

[0006] The robot acquires a 3D building information model of the steel structure to be operated, preset work points, and point cloud data collected by the robot. It then matches and calculates the current pose by matching the 3D building information model with the point cloud data, and generates the robot's movement trajectory, operation trajectory, and return trajectory based on the current pose and preset work points.

[0007] Multiple gripping units of the robot are controlled to perform release and fixation actions on the surface of the steel structure to be worked on in a cyclical manner until the robot moves to the preset work point along the movement trajectory. The fixation action includes mechanical locking action.

[0008] During the process of the robot moving along the movement trajectory and working along the operation trajectory, the environmental interference data collected by the robot in real time is acquired. The torque compensation value of each joint of the robot is calculated based on the environmental interference data, and the output torque of the multi-joint servo motor of the robot is adjusted according to the torque compensation value.

[0009] After controlling multiple gripping units to perform mechanical locking actions at preset work points, the robot's work units are controlled to switch to modular end effectors based on preset task instructions, and the modular end effectors are controlled to execute task instructions according to the work trajectory in combination with preset force control servo parameters.

[0010] After the task instruction is completed, the control unit switches the modular end effector to the flaw detector end, and obtains quality inspection data and corresponding operation position data by scanning through the flaw detector end.

[0011] It outputs quality inspection data and work position data. After receiving a qualified confirmation instruction based on the quality inspection data, it controls multiple gripping units to release the mechanical locking action and controls the robot to move according to the return trajectory.

[0012] The above technical solution, by acquiring a 3D building information model, preset work points and point clouds collected by the robot and matching the two to generate the current pose and three types of trajectories (movement trajectory, work trajectory and return trajectory), enables the robot to achieve precise positioning and path planning on complex high-altitude steel structures, ensuring that the robot reaches the work point along the correct spatial position and returns safely along the predetermined path after the task is completed, thereby reducing manual correction and reducing collisions or work deviations caused by errors.

[0013] Optionally, the current pose can be obtained by matching the 3D building information model with the point cloud data, specifically including:

[0014] Obtain structural topology nodes from a 3D building information model;

[0015] Feature segmentation is performed on the point cloud data to obtain the set of contour boundary points of the steel structure to be operated;

[0016] Calculate the spatial Euclidean distance between each corresponding point in the set of structural topology nodes and contour boundary points, and construct a matching error function based on the sum of squares of each spatial Euclidean distance;

[0017] The pose parameters are updated cyclically based on the matching error function and the preset reference coordinates to obtain the current pose.

[0018] The above technical solution extracts structural topology nodes from the 3D model, performs feature segmentation on the point cloud to obtain contour boundary points, and establishes a matching error function based on the sum of squared Euclidean distances in space to cyclically update the pose. This enables the robot to stably converge to an accurate pose relative to the component in a real point cloud environment with noise or local occlusion. This allows subsequent trajectory generation and gripping point selection to be based on real geometric relationships, improving positioning robustness and registration accuracy.

[0019] Optionally, multiple gripping units of the control robot cyclically perform release and fixation actions on the surface of the steel structure to be worked, specifically including:

[0020] The robot's current landing point is extracted based on its current pose, and the robot's target landing point sequence is analyzed based on its movement trajectory.

[0021] Control the first group of clamping units in a plurality of clamping units to maintain a fixed position at the current landing point, and control the second group of clamping units in a plurality of clamping units to perform a release movement.

[0022] After the second set of clamping units reaches the next target landing point in the target landing point sequence along the movement trajectory, the second set of clamping units is controlled to perform a fixing action, and the first set of clamping units is controlled to perform a release movement action.

[0023] The above technical solution, by dividing the gripping unit into two groups that work alternately and cyclically executing the hold-fix and release-movement actions according to the sequence of the current footing point and the target footing point, can realize a dynamic footing strategy for the robot to move gradually on the steel structure surface without continuous support. It maintains at least one group of gripping units to ensure the continuity of force and fall prevention during the operation, thereby achieving continuous movement without the need for external ladders or temporary supports.

[0024] Optionally, the second set of clamping units is controlled to perform a fixing action, specifically including:

[0025] The second set of clamping units is controlled to apply a preset clamping force to the surface of the steel structure to be worked, and to perform an anti-slip bonding action.

[0026] After the static friction force parameters between the second set of clamping units and the surface of the steel structure to be operated are obtained and reach the preset safety fall prevention threshold, the electromagnetic adsorption array of the second set of clamping units is controlled to perform magnetic adsorption action.

[0027] After detecting that the magnetic adsorption force has reached the preset adsorption threshold, the self-locking robotic arm of the second clamping unit is controlled to close, and a mechanical locking action is performed.

[0028] The above technical solution achieves anti-slip adhesion by first applying a preset clamping force during the fixed action, then triggering electromagnetic adsorption when the static friction safety threshold is reached, and finally closing the self-locking robotic arm to complete mechanical locking after the adsorption force reaches the threshold. This enables multiple redundant gripping and self-locking sequences (first increasing friction to prevent slippage, then generating adhesion, and finally mechanical locking), thereby significantly improving gripping reliability and reducing the risk of slippage or detachment caused by the failure of a single adhesion method.

[0029] Optionally, the torque compensation values ​​for each joint of the robot are calculated based on environmental interference data, and the output torque of the robot's multi-joint servo motors is adjusted according to the torque compensation values. Specifically, this includes:

[0030] Extract the current wind speed gradient vector and structural vibration frequency from environmental disturbance data;

[0031] Calculate the wind load disturbance vector based on the current wind speed gradient vector, and calculate the inertial disturbance vector based on the structural vibration frequency.

[0032] By mapping the wind load disturbance vector and the inertial disturbance vector to the robot's center of mass coordinate system, the robot's real-time center of mass offset coordinates are obtained.

[0033] Calculate the projection offset of the real-time center of gravity offset coordinates within the preset multi-point support polygon, and calculate the omnidirectional restoring torque based on the preset stiffness damping coefficient and the projection offset.

[0034] Based on the robot's current pose, a multi-joint Jacobian matrix is ​​constructed. The multi-joint Jacobian matrix is ​​then used to perform inverse dynamics on the omnidirectional restoring torque to obtain the torque compensation value of each joint.

[0035] The torque compensation values ​​of each joint are superimposed with the preset initial drive command to obtain the target drive command, and the output torque of the multi-joint servo motor is controlled according to the target drive command.

[0036] The above technical solution extracts wind speed gradient and structural vibration frequency from real-time environmental disturbance data, calculates wind load and inertial disturbance force and maps them to the robot's center of mass to obtain the center of gravity offset. Based on the projection of the offset in the multi-point support polygon and the stiffness damping coefficient, the recovery torque is calculated. The torque compensation of each joint is obtained by using the multi-joint Jacobian matrix to perform inverse dynamics and superimposed on the drive command. It can cancel the center of gravity offset and torque disturbance caused by disturbance in real time under strong wind or vibration conditions, reduce joint load fluctuation, maintain posture stability and avoid positioning errors or motor overload caused by external disturbances.

[0037] Optionally, the modular end effector can be controlled to execute task instructions according to the work trajectory by combining preset force control servo parameters, specifically including:

[0038] Extract the working contact force threshold and compliance resistance coefficient from the preset force control servo parameters;

[0039] Control the modular end effector to execute task instructions according to the work trajectory, and obtain the real-time contact reaction force of the modular end effector during the execution process;

[0040] The impedance deviation between the real-time contact reaction force and the working contact force threshold is calculated. Based on the impedance deviation and the compliance impedance coefficient, the working trajectory is dynamically corrected, and the modular end effector is controlled to continue executing the task command along the corrected working trajectory.

[0041] The above technical solution extracts the working contact force threshold and compliance impedance coefficient when the modular end effector performs the task, measures the contact reaction force in real time and calculates the impedance deviation to dynamically correct the working trajectory. This enables the end effector to keep the force within the safe threshold during the contact process and adaptively adjust the motion path according to the force feedback, thereby improving the contact quality of contact operations (such as welding, riveting or shot blasting), reducing workpiece damage and tool overload, and ensuring the consistency of working mechanical conditions.

[0042] Optionally, quality inspection data and corresponding work location data can be obtained through flaw detection scanning, specifically including:

[0043] Acquire internal defect echo signals collected by the ultrasonic array sensor integrated at the flaw detection end, and surface morphology images collected by the visual sensor integrated at the flaw detection end.

[0044] Multimodal data fusion is performed on the internal defect echo signal and the surface morphology image to generate quality inspection data;

[0045] The preset work points are mapped onto the digital twin mesh corresponding to the 3D building information model to obtain work location data.

[0046] The above technical solution uses an ultrasonic array at the flaw detection end to collect internal defect echoes and a visual sensor to capture surface morphology. It performs multimodal data fusion and maps the detection results to a digital twin mesh in a three-dimensional model to output quality inspection data and corresponding work locations. It can provide defect and surface quality records with spatial coordinates after the operation, realize traceable non-destructive testing results, facilitate the location of rework points, and serve as a basis for quality acceptance judgment.

[0047] Secondly, a robotic operation system for high-altitude steel structures is provided, the system comprising:

[0048] The pose planning module is configured to acquire the 3D building information model of the steel structure to be operated, the preset work points, and the point cloud data collected by the robot. It matches the 3D building information model with the point cloud data to calculate the current pose, and generates the robot's movement trajectory, operation trajectory, and return trajectory based on the current pose and the preset work points.

[0049] The climbing control module is configured to control multiple gripping units of the robot to perform release and fixation actions on the surface of the steel structure to be worked on, until the robot moves to the preset work point along the movement trajectory. The fixation action includes mechanical locking action.

[0050] The dynamic anti-disturbance module is configured to acquire real-time environmental interference data collected by the robot during the robot's movement along the movement trajectory and operation along the work trajectory, calculate the torque compensation value of each joint of the robot based on the environmental interference data, and adjust the output torque of the robot's multi-joint servo motors based on the torque compensation value.

[0051] The force control operation module is configured to control multiple gripping units to perform mechanical locking actions at preset operation points, and then control the robot's operation units to switch to modular end effectors based on preset task instructions. It also controls the modular end effectors to execute task instructions according to the operation trajectory by combining preset force control servo parameters.

[0052] The quality inspection module is configured to control the work unit to switch the modular end effector to the flaw detector after the task instruction is completed, and to obtain quality inspection data and corresponding work position data by scanning through the flaw detector.

[0053] The closed-loop return module is configured to output quality inspection data and work position data. After receiving a qualified confirmation instruction based on the quality inspection data, it controls multiple gripping units to release the mechanical lock and controls the robot to move according to the return trajectory.

[0054] Thirdly, an electronic device is provided, including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the above.

[0055] Fourthly, a computer-readable storage medium is provided, the computer-readable storage medium storing instructions that, when executed, perform the method as described in any of the preceding claims.

[0056] In summary, implementing one or more technical solutions provided in this application has at least the following technical effects or advantages:

[0057] This invention integrates functions such as 3D building information modeling of steel structures, on-site point cloud perception data, robot trajectory planning, climbing and gripping control, environmental disturbance compensation control, force control operation, and post-operation flaw detection. This enables the robot to complete a complete closed-loop operation process from positioning, movement, operation to inspection and return in a high-altitude steel structure environment, thereby achieving automation and continuity of high-altitude steel structure operations. At the same time, through modular end effectors and a quality inspection result feedback mechanism, a unified data link is formed between the operation process and quality inspection. After the operation is completed, the inspection results with spatial location markers can be directly output for qualification confirmation. This improves the operation efficiency and quality controllability of high-altitude steel structure construction or maintenance, reduces the need for manual high-altitude operations, and improves overall operation safety and management traceability. Attached Figure Description

[0058] Figure 1 This is a flowchart illustrating a robot operation method for high-altitude steel structures disclosed in this application;

[0059] Figure 2 This is a schematic diagram of a robot operation system for high-altitude steel structures disclosed in this application;

[0060] Figure 3 This is a schematic diagram of the structure of an electronic device disclosed in this application.

[0061] Explanation of reference numerals in the attached diagram: 201, Pose planning module; 202, Climbing control module; 203, Dynamic interference immunity module; 204, Force control operation module; 205, Quality inspection module; 206, Closed-loop return module; 301, Processor; 302, Communication bus; 303, User interface; 304, Network interface; 305, Memory. Detailed Implementation

[0062] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0063] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0064] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0065] It should be noted that the robot described in this application typically comprises multiple highly free-form bionic mechanical limbs (such as quadruped, hexapod, or bi-arm configurations). The core "gripping unit" employs a mechatronics composite architecture, and its end effector typically integrates: a multi-dimensional force sensor for sensing the adhesion force to the steel structure surface, a flexible electromagnetic adsorption array or permanent magnet array for generating strong normal adsorption force, and a self-locking mechanical arm (caliper) driven by a servo motor and exhibiting a "C" or "U" shaped envelope structure. When mechanical locking is performed, the self-locking mechanical arm can achieve irreversible mechanical self-locking through a worm gear or ratchet mechanism, completely locking the edge of the steel beam flange in a physical encircling manner. Furthermore, the robot's working unit end effector is equipped with a standardized quick-change flange interface, compatible with various modular end effectors, including ultrasonic probes, line laser scanners, grinders, and servo welding torches, thus providing underlying hardware support for the complex climbing, operation, and inspection actions described in this application.

[0066] Figure 1 This is a flowchart illustrating a robotic operation method for high-altitude steel structures according to an embodiment of this application. This method can be implemented using a computer program, a microcontroller, or run on a robotic operation system for high-altitude steel structures. The computer program can be integrated into an application or run as a standalone tool application. The specific steps of the robotic operation method for high-altitude steel structures are described in detail below.

[0067] S101: Acquire the 3D building information model of the steel structure to be operated, the preset operation points, and the point cloud data collected by the robot. Match the 3D building information model with the point cloud data to obtain the current pose, and generate the robot's movement trajectory, operation trajectory, and return trajectory based on the current pose and the preset operation points.

[0068] For example, when a robot is initially hoisted or deployed within a vast steel framework network such as the dome of a large stadium, the frame of a super high-rise building, or a cross-sea bridge, in order to achieve global initial positioning and path planning for the robot without prior coordinate references, the system first retrieves the 3D digital model data of the building and performs macroscopic spatial coordinate system registration and transformation with the actual spatial distribution point cloud generated by the sensors on the robot's body in real time. Through this virtual-real alignment mechanism, the robot can break free from its dependence on external GPS and complete the absolute coordinate calculation of its initial pose. After determining the starting point and preset work points, the system generates a global navigation path covering the entire work cycle through a multi-level path planning algorithm. This includes not only the movement trajectory guiding the robot to avoid cable interference, but also the microscopic work trajectory of the end-effector conforming to the workpiece surface for processing, and the return trajectory to the safety reference point after task completion, thus providing a high-precision spatial guidance reference for subsequent unmanned high-altitude operations.

[0069] In this embodiment, the movement trajectory refers to the collision-free global navigation path by which the robot safely transfers from its current pose to a preset work point in space; the work trajectory refers to the local continuous geometric spatial path followed by the end effector for fine processing within the work area; and the return trajectory refers to the optimal evacuation path after completing the task, leaving the work area, and returning to a safe reference point. Specifically, when generating the robot's movement trajectory, work trajectory, and return trajectory, the system first converts the 3D building information model into a global navigation topology map containing obstacle bounding boxes. Based on the current pose as the starting point and the preset work point as the ending point, a sampling-based path planning algorithm (such as the RRT algorithm) or a heuristic graph search algorithm (such as the A algorithm) is used to calculate a set of global discrete waypoints with minimum energy consumption and no collisions, while avoiding interference objects such as cables and node plates. The movement trajectory is then generated by smoothing the path using spline curves. Secondly, after reaching the preset work point, the system extracts three... In the 3D building information model, the local surface normal vector and curvature features of the corresponding steel components (such as welds or surfaces to be ground) are combined with the workspace constraints and process requirements of the end effector (such as welding speed and spraying angle) to generate a three-dimensional continuous envelope line along the surface of the component, which maintains a fixed offset distance from the workpiece contour, as the operation trajectory. After the task is completed, the system takes the current operation end point as the starting point and the set supply station or safety anchor point as the ending point. After confirming that there are no new obstacles in the evacuation space by calling the environmental dynamic perception map, the system generates the return trajectory by reverse optimization or calling the safe mirror path of the original movement trajectory.

[0070] In one possible implementation, the current pose is obtained by matching the 3D building information model with point cloud data. Specifically, this includes: acquiring structural topology nodes in the 3D building information model; performing feature segmentation on the point cloud data to obtain a set of contour boundary points of the steel structure to be operated; calculating the spatial Euclidean distances between the structural topology nodes and corresponding points in the contour boundary point set, and constructing a matching error function based on the sum of the squares of the spatial Euclidean distances; and iteratively updating the pose parameters based on the matching error function and preset reference coordinates to obtain the current pose.

[0071] In this embodiment, the matching error function refers to a mathematical optimization model used to quantify the degree of spatial position deviation between the three-dimensional sampling points on the surface of a physical entity and the key nodes of the virtual design model. For example, it is a quadratic objective function constructed by calculating the sum of squared distances between the actual contour boundary points of the steel beam obtained by LiDAR scanning and the standard design nodes in the Building Information Modeling (BIM). The smaller the calculated value of this function, the higher the degree of geometric alignment between the steel structure in the real environment and the virtual digital model in three-dimensional space.

[0072] Specifically, the system first parses a 3D building information model containing global standard design geometric parameters, extracting the 3D coordinates of structural topology nodes representing the intersections, connection features, or key stress-bearing parts of the main frame. Simultaneously, it performs normal vector or curvature feature segmentation on massive point cloud data collected in real-time by external sensors, accurately filtering out background noise and non-structural objects such as scaffolding, thus obtaining a set of contour boundary points that accurately depict the true external geometry of the steel structure to be constructed. A spatial nearest neighbor matching relationship is established between the extracted structural topology nodes and the feature-segmented contour boundary point set. The spatial Euclidean distance between each corresponding point pair is calculated using the 3D coordinate difference, and these distance values ​​are squared and summed. This global sum of squared distances is used as the target constraint to construct a matching error function. Specifically, the mathematical expression of the matching error function is as follows:

[0073]

[0074] Where E(R, T) represents the matching error function value; N represents the total number of valid matching point pairs; p i q represents the three-dimensional spatial coordinate vector of the i-th contour boundary point extracted from the point cloud data; i In a 3D building information model, p represents... iThe corresponding three-dimensional spatial coordinate vector of the i-th structural topology node; R represents the three-dimensional spatial rotation matrix composed of the Euler angles to be solved; T represents the three-dimensional spatial translation vector to be solved. By solving for the optimal solutions R and T that minimize E(R, T) through gradient descent or singular value decomposition, the current pose can be accurately constructed.

[0075] Furthermore, the preset reference coordinates (i.e., the pre-calibrated prior position or initial global coordinates) are used as the starting reference anchor point for spatial iterative optimization. The function value of the matching error function is minimized as the optimization criterion. The coordinate system transformation is driven by the numerical optimization algorithm, and the pose parameters describing the spatial rotation and translation relationship are updated iteratively. This allows the point cloud dataset to continuously approach and dynamically fit to the three-dimensional building information model in three-dimensional space until the iterative calculation process meets the preset error convergence stopping condition. The optimal spatial transformation matrix locked at the end of the iteration is output to obtain the current pose.

[0076] S102: Control the multiple gripping units of the robot to perform release and fixation actions on the surface of the steel structure to be worked on in a cyclical manner until the robot moves to the preset work point along the movement trajectory. The fixation action includes mechanical locking action.

[0077] For example, in complex truss or I-beam networks without scaffolding at high altitudes, the robot can safely attach and continuously traverse complex spatial structures. To address the technical risks of high-altitude falls caused by traditional pure electromagnetic adsorption or negative pressure vacuum adsorption in situations such as sudden power outages, severe rusting of steel structures, localized water accumulation, or peeling of thick anti-rust paint, this step introduces a "mechanical locking action" with solid geometric envelope characteristics. During actual climbing and traversing, the robot's various climbing execution ends (e.g., biomimetic multi-legged mechanical feet or gripping hands with alternating double-arm configurations) adopt a multi-support alternating load-bearing and gait switching strategy: the system always maintains a portion of the gripping units locked to the steel beam flanges or chords through a rigid mechanical envelope structure as fixed anchor points to resist external loads; simultaneously, another portion of the gripping units releases the gripping state and traverses along the planned trajectory. Through this cyclical control logic that alternates between load-bearing points and movement points, the robot can perform multi-directional maneuvers in three-dimensional space. Whether it is climbing vertical columns, sliding across horizontal beams, or climbing upside down on the building roof, it can reliably navigate through the complex building framework until it finally arrives smoothly at the preset work point where construction intervention is required.

[0078] In one possible implementation, multiple gripping units of the robot are controlled to cyclically perform release and fixation actions on the surface of the steel structure to be worked. Specifically, this includes: extracting the current landing point of the robot based on the current pose, and parsing the target landing point sequence of the robot according to the movement trajectory; controlling the first group of gripping units to maintain a fixed action at the current landing point, and controlling the second group of gripping units to perform release and movement actions; after the second group of gripping units reaches the next target landing point in the target landing point sequence along the movement trajectory, controlling the second group of gripping units to perform a fixation action, and controlling the first group of gripping units to perform a release and movement action.

[0079] In the embodiments of this application, the target landing point sequence refers to an ordered set of expected grasping position coordinates generated by discretizing the robot's global travel path on the spatial structure. It is used to represent the safe stopping reference points that the robot must reach in sequence for each gait alternation when moving from the current position to the destination on a complex geometric component. For example, when traveling along the surface of an I-beam, a set of three-dimensional spatial coordinate points distributed sequentially and at intervals is calculated and arranged according to the robot's maximum stride distance and the geometric characteristics of the flange edge of the steel beam.

[0080] Specifically, the system first performs inverse kinematics calculations based on the current pose in three-dimensional space to extract the current landing points of each end effector of the robot in actual contact with the surface of the steel structure to be worked, thus accurately determining the physical coordinates of the current safe support base. Simultaneously, based on the globally planned continuous movement trajectory, combined with the robot's maximum stride and fall-prevention safety verification rules, the continuous path is spatially discretized and interpolated to resolve the robot's target landing point sequence. Employing a multi-legged or dual-arm biomimetic alternating movement logic, the robot's multiple gripping units are divided into two alternating force-bearing motion groups. The first group of gripping units is controlled to maintain a fixed action at the current landing point, forming a rigid connection with the surface of the steel structure to be worked through pre-tensioning or mechanical locking, thereby constructing an absolutely stable anchoring support base. Under this stable and anti-overturning state, multiple... The second set of gripping units in the gripping unit performs a release movement action, that is, it releases the original gripping state and lifts it up, allowing it to suspend in space and move forward. During this dynamic process, kinematic trajectory monitoring is continuously performed. After the second set of gripping units accurately reaches the next target landing point in the target landing point sequence along the movement trajectory, the second set of gripping units is immediately controlled to perform a fixing action, so that it firmly grips the surface of the steel structure to be worked on to establish a new front-end pull-out support point. After the feedback from the bottom sensor confirms that the second set of gripping units has established a safe static friction force, the load-bearing body is handed over, and the first set of gripping units, which was originally in the supporting state, is controlled to perform a release movement action, so that it follows and moves forward. This cycle of weight transfer and gripping and releasing logic is repeated, ultimately realizing the control of multiple gripping units of the robot to perform release movement and fixing actions on the surface of the steel structure to be worked on.

[0081] In one possible implementation, controlling the second set of clamping units to perform a fixing action specifically includes: controlling the second set of clamping units to apply a preset clamping force to the surface of the steel structure to be operated, performing an anti-slip adhesion action; after obtaining the static friction force parameter between the second set of clamping units and the surface of the steel structure to be operated, reaching a preset safety anti-fall threshold, controlling the electromagnetic adsorption array of the second set of clamping units to perform a magnetic adsorption action; after detecting that the adsorption force corresponding to the magnetic adsorption action reaches a preset adsorption threshold, controlling the self-locking mechanical arm of the second set of clamping units to close, performing a mechanical locking action.

[0082] In this embodiment, the preset safety fall prevention threshold refers to the minimum static friction threshold value that, in a high-altitude working environment, is calculated through dynamics to ensure that there is no relative sliding between the clamping end and the contact surface of the steel structure in order to resist environmental interference such as the robot's own weight and external wind load. It is used to represent the safety judgment benchmark for triggering the next level of deep fixation program. For example, when the contact surface is a rust-proof painted steel plate with a specific roughness and the wind speed at the working height is a set level, a lower limit value of friction resistance in Newtons is set in advance. Only when the friction force parameter collected in real time by the torque sensor reaches this lower limit value will the system determine that the current preliminary bonding state is safe and allow the electromagnetic adsorption to be activated.

[0083] Specifically, the system first uses a low-level servo drive component to control the second set of clamping units to apply a preset clamping force to the surface of the steel structure to be worked on. This drives the flexible friction pads or contact elements at their ends to tightly press against the steel beam surface to eliminate gaps, thereby performing an anti-slip contact action and establishing preliminary anti-slip physical constraints. The system then uses a built-in multi-dimensional force sensor to monitor the tangential force state of the contact surface in real time. Once the static friction force parameters between the second set of clamping units and the surface of the steel structure to be worked on reach a preset safety anti-fall threshold, confirming that the current force state has the anti-slip capability to resist foundation disturbances, the system quickly activates the power circuit to control the electromagnetic adsorption array of the second set of clamping units to perform a magnetic adsorption action, utilizing instantaneous... The strong magnetic field generated creates a huge normal adsorption force perpendicular to the steel structure surface, establishing a secondary anti-overturning constraint. The feedback current or magnetic flux of the magnetic circuit is continuously monitored. When the adsorption force corresponding to the magnetic adsorption action reaches the preset adsorption threshold, it proves that the magnetic adsorption state is completely firm and there is no magnetic leakage or peeling attenuation. At this time, the final mechanical envelope closed-loop program is immediately triggered to control the self-locking mechanical arm of the second set of clamping units to close, so that its caliper or ratchet mechanism can firmly clamp the edge flange of the steel structure in a physical embrace, and perform a mechanical locking action. Thus, through the three progressively redundant anti-fall mechanism of "friction pressing - magnetic adsorption - mechanical locking", the second set of clamping units is finally controlled to perform the fixing action.

[0084] S103: During the process of the robot moving along the movement trajectory and working along the operation trajectory, the robot acquires environmental interference data collected in real time, calculates the torque compensation value of each joint of the robot based on the environmental interference data, and adjusts the output torque of the robot's multi-joint servo motors based on the torque compensation value.

[0085] For example, when a robot performs step-by-step movement or hovering operations on an exposed steel frame hundreds of meters above the ground, it typically faces an extremely complex dynamic physical environment. On one hand, there is the enormous aerodynamic thrust brought about by sudden strong crosswinds or gusts of wind shear; on the other hand, there is the low-frequency swaying of the super-tall steel structure itself under wind load, and the high-frequency vibrations transmitted to the steel beams from surrounding large construction machinery (such as tower crane operation and pile driving vibrations). If the robot's joints only output conventional torques to maintain their original posture, external combined impacts can easily cause the gripping ends to detach due to the force exceeding the maximum static friction force of the contact surface or the structural yield limit. Therefore, this step introduces an active dynamic disturbance rejection control mechanism. The system acquires real-time data on external airflow impacts and base oscillations, calculates the feedforward compensation torque required to counteract external disturbances, and superimposes it onto the servo drives of each joint. This allows the robot to dynamically adjust the torque output of each joint, actively absorb impact energy, and always maintain a stable margin of the overall center of gravity within the multi-point support polygon.

[0086] In one possible implementation, the torque compensation values ​​of each joint of the robot are calculated based on environmental disturbance data, and the output torque of the robot's multi-joint servo motors is adjusted according to the torque compensation values. Specifically, this includes: extracting the current wind speed gradient vector and structural vibration frequency from the environmental disturbance data; calculating the wind load disturbance vector based on the current wind speed gradient vector, and calculating the inertial disturbance vector based on the structural vibration frequency; mapping the wind load disturbance vector and the inertial disturbance vector to the robot's center of mass coordinate system to obtain the robot's real-time center of mass offset coordinates; calculating the projection offset of the real-time center of mass offset coordinates within a preset multi-point support polygon, and calculating the omnidirectional restoring torque based on a preset stiffness damping coefficient and the projection offset; constructing a multi-joint Jacobian matrix based on the robot's current pose, and using the multi-joint Jacobian matrix to perform inverse dynamics on the omnidirectional restoring torque to obtain the torque compensation values ​​of each joint; superimposing the torque compensation values ​​of each joint with a preset initial drive command to obtain the target drive command, and controlling the output torque of the multi-joint servo motors according to the target drive command.

[0087] In this embodiment, the preset multi-point support polygon refers to a convex polygonal geometric region formed by connecting the actual physical contact points between the robot's multiple gripping units currently in a fixed and locked state and the surface of the steel structure to be worked on a horizontal reference plane perpendicular to the direction of gravity. It is used to represent the safety and stability margin boundary of the robot in maintaining static and dynamic balance under high-altitude strong wind and vibration environments. For example, when the robot adopts a multi-legged or multi-arm alternating gait and three gripping units are currently firmly gripping the steel beam, the geometric center line connecting these three contact points will form a triangular support area on the horizontal plane. As long as the overall center of gravity projection of the robot falls strictly within this triangular area, it can be determined that the robot has sufficient anti-overturning ability and will not fall from a height.

[0088] Specifically, the system first uses an anemometer mounted on the outside of the robot and a built-in high-frequency inertial measurement unit (IMU) to monitor and extract changes in the external airflow and the state of the base support surface in real time. It extracts the current wind speed gradient vector, representing the intensity and three-dimensional direction changes of high-altitude gusts, from the environmental disturbance data, and extracts the periodic high-frequency micro-vibrations generated by the steel structure under construction or wind load, i.e., the structural vibration frequency. Based on a pre-set aerodynamic model and the geometric characteristics of the robot's windward surface, it calculates the equivalent external aerodynamic vector directly applied to the surface of the robot body, i.e., the wind load disturbance vector, based on the current wind speed gradient vector. Combining the robot's spatial mass distribution attributes, it calculates the additional mechanical impact force vector caused by foundation resonance, i.e., the inertial disturbance vector, based on the structural vibration frequency. Finally, through a spatial homogeneous transformation matrix, it distributes the wind load disturbances acting on different parts of the robot body. The force vector and the inertial disturbance force vector are uniformly mapped to the robot's center-of-mass coordinate system (i.e., the equivalent spatial coordinate system representing the concentrated mass of the robot's entire body) for vector synthesis. After dynamic calculation, the real-time center-of-gravity offset coordinates of the robot deviating from its ideal equilibrium position under the current combined external force disturbance are obtained. Then, these coordinates are projected vertically along the direction of gravity, and the projection offset of the real-time center-of-gravity offset coordinates within a preset multi-point support polygon is calculated to accurately quantify the degree of danger of the current center of gravity approaching the overturning edge. Based on a preset stiffness damping coefficient (a configuration parameter representing the robot's ability to resist external flexible deformation and absorb high-frequency vibration energy at each joint) and this projection offset, the global reverse correction torque required to pull the deviated center of gravity back to the absolute safe center of the support polygon, i.e., the omnidirectional restoring torque, is calculated using an impedance control law. Specifically, the omnidirectional restoring torque is calculated based on the following impedance control law:

[0089]

[0090] Among them, F rec ΔX represents the omnidirectional restoring torque (including generalized three-dimensional force and three-dimensional moment) mapped in the centroid coordinate system; ΔX represents the projected offset of the real-time centroid offset coordinates; ΔX1 represents the rate of change of the centroid offset; K p K is the preset stiffness coefficient matrix. d This is the preset damping coefficient matrix.

[0091] Based on this, the robot's current kinematic joint angle data is extracted. A mathematical matrix, the multi-joint Jacobian matrix, is constructed based on the robot's current pose to map the linear transformation relationship between the generalized force in Cartesian space at the end effector and the torque in joint space. This multi-joint Jacobian matrix is ​​then used to perform inverse kinematics allocation on the omnidirectional restoring torque acting on the center of mass, accurately calculating the disturbance rejection torque load that each motor axis needs to independently bear, thus obtaining the torque compensation value for each joint. The expression for inverse kinematics using the multi-joint Jacobian matrix is ​​as follows:

[0092]

[0093] Where, τ comp This represents the calculated column vector of joint torques containing the torque compensation values ​​for each joint; q is the current angle vector of each joint of the robot; J T (q) is the transpose of the multi-joint Jacobian matrix calculated based on the current robot configuration. This matrix establishes an accurate linear mapping from the generalized forces in the robot's operating space to the driving torques in the joint space.

[0094] Furthermore, the torque compensation values ​​of each joint used to resist environmental interference are used as feedforward compensation signals. These signals are superimposed and calculated with the baseline motion commands originally issued by the robot to perform climbing or maintenance actions, i.e., the preset initial drive commands. This process generates target drive commands that take into account both task execution and dynamic balance interference resistance. The output torque of the multi-joint servo motors is precisely controlled according to the target drive commands. Ultimately, the torque compensation values ​​of each joint of the robot are calculated based on environmental interference data, and the output torque of the robot's multi-joint servo motors is adjusted according to the torque compensation values.

[0095] S104: After controlling multiple gripping units to perform mechanical locking actions at preset work points, control the robot's work units to switch to modular end effectors based on preset task instructions, and control the modular end effectors to execute task instructions according to the work trajectory in combination with preset force control servo parameters.

[0096] For example, once the robot completes mechanical locking at a preset high-altitude work point, it establishes a rigid physical base resistant to wind and earthquakes for subsequent precision operations. At this point, the system analyzes the specific construction operation type (e.g., high-strength bolt tightening, full-penetration bevel welding, rust removal and grinding of steel components, or anti-corrosion coating spraying) and controls the quick-change flange of the main robotic arm to detach from the current mounting component, automatically connecting to a dedicated modular end effector (such as a servo electric screwdriver, automatic welding gun, or grinding disc) matched to the task. Considering that large steel structures commonly exhibit manufacturing tolerances, surface unevenness, or self-weight deflection deformation during actual high-altitude assembly, relying solely on traditional pure position rigid control can easily lead to engineering accidents such as tool jamming or over-grinding. Therefore, this step introduces a compliant impedance control mechanism by combining preset force control servo parameters, giving the robotic arm adaptive adjustment capabilities based on force feedback. This allows it to automatically adapt to the physical undulations of the steel beam surface, thereby ensuring that construction processes meet standards while avoiding irreversible rigid damage to the robot body and building substrate.

[0097] In one possible implementation, the modular end effector is controlled to execute task commands according to the work trajectory by combining preset force control servo parameters. Specifically, this includes: extracting the work contact force threshold and compliance impedance coefficient from the preset force control servo parameters; controlling the modular end effector to execute task commands according to the work trajectory, and acquiring the real-time contact reaction force of the modular end effector during execution; calculating the impedance deviation between the real-time contact reaction force and the work contact force threshold, dynamically correcting the work trajectory based on the impedance deviation and compliance impedance coefficient, and controlling the modular end effector to continue executing task commands along the corrected work trajectory.

[0098] In the embodiments of this application, the compliance resistance coefficient refers to the control parameter used to characterize the dynamic mechanical interaction relationship (i.e., equivalent stiffness, damping, and mass characteristics) between the end effector and the external physical environment when the robot performs contact operations. It is used to represent the sensitivity and buffering ability of the robot to actively yield or conform to the contour of the physical environment when subjected to unpredictable contact resistance. For example, when performing contact grinding or bolt tightening operations on the surface of high-altitude steel structures, a set of matrix parameters composed of spring stiffness constant and viscous damping constant is set in advance. The configuration of this coefficient enables the robot's end effector to maintain sufficient contact force for operation, and when encountering weld protrusions or processing deformations on the surface of steel beams, it can adaptively fine-tune the tool position based on the compliance control mechanism without causing destructive rigid hard collisions.

[0099] Specifically, the system first delves into the configuration space of the underlying controller, parsing the reference mechanical upper limit value representing a safe and ideal contact state from the preset force control servo parameters containing force interaction rules. This extracts the working contact force threshold from the preset force control servo parameters and simultaneously extracts the compliance impedance coefficient that determines the system's force / position coupling characteristics. It then sends down underlying drive position signals to control a modular end effector (such as a grinder or welding torch) equipped with a specific working tool to strictly follow a pre-planned three-dimensional spatial path (working trajectory) using point clouds to execute corresponding processing actions (i.e., execute task instructions). During this high-altitude dynamic physical contact execution, a six-dimensional force / torque sensor integrated at the end flange monitors and collects the actual three-dimensional contact force and torque feedback applied to the tool end by the external steel structure surface in real time at a millisecond-level sampling frequency, thus obtaining the real-time contact reaction force of the modular end effector. This is followed by comparison of underlying algorithms. In the device, the actual force data from sensor feedback is subtracted from and filtered by the ideal reference upper limit of force to accurately calculate the impedance deviation between the real-time contact reaction force and the working contact force threshold. This quantitatively assesses whether the end effector is in an under-pressure release state or an overload compression state. Then, this mechanical deviation value is input into a preset impedance control equation, and the admittance / impedance mapping relationship is used to convert it into a position and attitude offset compensation amount in Cartesian space. That is, the working trajectory is dynamically corrected based on the impedance deviation and the compliance impedance coefficient, so that the original pure position geometry path without dynamic compensation becomes an elastic adaptive path that can adapt to the actual surface error of the high-altitude steel structure. Finally, the corrected feedforward coordinates are sent to each joint motor, and the modular end effector is controlled to continue to execute the task command along the corrected working trajectory. In this way, the modular end effector is controlled to execute the task command according to the working trajectory by combining the preset force control servo parameters.

[0100] Specifically, the preset impedance control equation is a second-order dynamic model established in Cartesian space, and its mathematical expression is:

[0101]

[0102] Where E represents the amount of position and attitude offset compensation (i.e., position error) that the end effector needs to correct; E1 and E2 represent the first derivative (velocity deviation) and second derivative (acceleration deviation) of the offset compensation amount, respectively; F ext The real-time contact reaction force is obtained through a six-dimensional force / torque sensor; F d The preset working contact force threshold (i.e., the desired contact force); F ext -F d This refers to the impedance deviation between the real-time contact reaction force and the working contact force threshold; M d B d Kd These represent the target mass matrix, compliance damping matrix, and compliance stiffness matrix (i.e., compliance impedance coefficient) in the preset force control servo parameters, respectively. By solving this ordinary differential equation, a smooth pose compensation amount E can be dynamically output to correct the working trajectory.

[0103] S105: After the task instruction is completed, the control unit switches the modular end effector to the flaw detector end, and obtains quality inspection data and corresponding operation position data by scanning through the flaw detector end.

[0104] For example, after the robot completes processing tasks such as high-strength bolt tightening, steel structure node welding, or anti-corrosion coating spraying, the system can automatically unload the previous processing tools through a quick-change mechanism (such as a quick-change flange or automatic gun changer) at the end of the robotic arm, and connect to a flaw detection end dedicated to non-destructive testing via an automatic locking mechanism. The flaw detection end will directly perform in-situ re-scan along the previous construction operation trajectory to check for engineering defects such as surface cracks, internal slag inclusions, incomplete coating spraying, or loose connections at the construction site. During this process, the system not only converts the various physical indicators collected by scanning into quantified quality inspection data, but also simultaneously records the global three-dimensional spatial coordinates of the robot or the unique identifier of the physical component when the flaw detection is triggered. This generates a structural quality inspection report with a precise location label for each high-altitude construction operation point, providing intuitive data support for subsequent automatic project acceptance, full life-cycle quality traceability, or precise location rework.

[0105] In one possible implementation, quality inspection data and corresponding work location data are obtained by scanning with a flaw detector end. Specifically, this includes: acquiring internal defect echo signals collected by an ultrasonic array sensor integrated with the flaw detector end, and surface morphology images collected by a visual sensor integrated with the flaw detector end; performing multimodal data fusion on the internal defect echo signals and surface morphology images to generate quality inspection data; and mapping preset work points to the digital twin mesh corresponding to the three-dimensional building information model to obtain work location data.

[0106] In this embodiment, a digital twin mesh refers to a discretized three-dimensional topological coordinate reference base pre-constructed in the virtual digital space of 3D BIM, which strictly corresponds one-to-one with the steel structure to be operated in the real physical environment in terms of geometric dimensions and spatial position. It is used to represent the precise mapping coordinate matrix of high-altitude construction operation nodes in the physical real world in the virtual digital world. For example, a spatial three-dimensional mesh surface with a precision of millimeters is pre-divided on the surface of the virtual steel component in the 3D building information model. When the physical robot completes the operation and flaw detection at a specific physical coordinate point of the real high-altitude steel beam, the system can use this mesh system to directly convert the real physical absolute position into virtual three-dimensional node labels on the digital model, thereby realizing the data homology and precise linkage between the construction quality of the physical entity and the virtual building asset model.

[0107] Specifically, the system first activates the non-destructive testing hardware loop of the end-effector, and synchronously acquires sensor signals from different physical media through the underlying high-speed data interface. This includes acquiring echo signals from the ultrasonic array sensor (such as a high-frequency phased array ultrasonic probe) integrated at the flaw detector end, which transmits ultrasonic waves into the steel structure and receives the reflected energy from the material interface. These echoes are used to deeply reveal hidden conditions such as porosity and slag inclusions within welds or the substrate. Simultaneously, the system also acquires visual sensors (such as industrial-grade high-resolution cameras or line laser profilometers) integrated at the flaw detector end, which perform optical scanning and 3D reconstruction of the steel structure's appearance. These visually represent surface smoothness, undercut, or weld defects. The system generates surface topography images of the remaining height dimensions. In the underlying data processing architecture, heterogeneous data alignment and joint feature extraction algorithms are introduced. One-dimensional or two-dimensional acoustic time-series signals reflecting the internal material health status are precisely registered and fused with two-dimensional or three-dimensional optical images reflecting external physical geometric features in a time stamp and spatial coordinate system. This multi-modal data fusion of internal defect echo signals and surface topography images eliminates the detection blind spots and misjudgment risks of single sensors under complex working conditions. Furthermore, by integrating internal and external physical features, a final quantitative result is generated to comprehensively and objectively assess whether the current construction work quality meets the engineering acceptance standards—that is, quality inspection data is generated. The specific fusion process includes: First, the system uses a hardware clock synchronization mechanism to assign a unified global timestamp to the one-dimensional time-series echo signal acquired by the ultrasonic array sensor and the optical image acquired by the visual sensor, achieving strict alignment in the time dimension. Simultaneously, through pre-performed "hand-eye calibration," the spatial pose transformation matrix between the center of the ultrasonic probe and the optical center of the camera is obtained, accurately mapping the spatial origin point of the one-dimensional ultrasonic signal to the pixel coordinates corresponding to the two-dimensional / three-dimensional surface morphology image, achieving coordinate registration in the spatial dimension. A feature-based fusion strategy is adopted to extract the peak amplitude and decay time features in the ultrasonic signal to characterize the depth and size of internal defects, and to extract the gray-level gradient or point cloud depth change in the optical image to characterize surface cracks and smoothness. The mapped and aligned internal defect features and surface visual features are input into a pre-set joint evaluation discrimination matrix or a multilayer perceptron (MLP) model to generate a three-dimensional quality state feature vector that comprehensively reflects the internal material defects and external geometric morphology of the coordinate point, i.e., generating the quality detection data, thereby effectively compensating for the detection blind spots of a single sensor under complex working conditions.

[0108] Furthermore, the three-dimensional coordinates of the global physical space where the robot is located during the aforementioned construction operations and non-destructive testing are extracted, i.e., the previously planned preset work points. Through a preset spatial coordinate system transformation matrix, these physical coordinates are inversely transformed and projected into the pre-loaded digital building system. That is, the preset work points are mapped to the digital twin mesh corresponding to the three-dimensional building information model, so that the result of each quality inspection is precisely spatially anchored to a specific mesh node of the virtual model. This results in virtual spatial location information with absolute three-dimensional spatial traceability attributes and a unique binding relationship with the aforementioned quality inspection data, i.e., the corresponding work location data. Finally, the closed loop is completed, realizing the acquisition of quality inspection data and corresponding work location data through scanning at the flaw detection end.

[0109] S106: Outputs quality inspection data and work position data. After receiving a qualified confirmation instruction based on the quality inspection data, it controls multiple gripping units to release the mechanical locking action and controls the robot to move according to the return trajectory.

[0110] In this embodiment, the qualified confirmation instruction refers to the control signal issued by the remote control terminal or automated quality inspection and evaluation model to the lower-level device after verifying the compliance of the non-destructive testing multimodal results uploaded on site, allowing evacuation. It is used to indicate that the construction quality of the current high-altitude work node has fully met the preset engineering acceptance standards and that no on-site rework or repair is required. For example, when the background system determines that the ultrasonic echo signal does not show internal cracks and the visual image confirms that the weld size meets the national standard tolerance requirements, the digital twin management and control platform automatically generates and sends a digital instruction packet containing an allow release code to the robot's main control motherboard.

[0111] Specifically, the system first uses a wireless communication module or industrial IoT interface to simultaneously package and upload the multimodal fusion judgment results of internal defects and surface morphology obtained from the previous in-situ scanning—that is, output quality inspection data and operation location data—to the back-end control center, cloud acceptance platform, or handheld operation terminal for digital archiving and quality traceability. While the communication link's listening port is in a waiting-for-feedback state, once the back-end terminal or automated quality inspection algorithm compares and verifies the uploaded fusion data and confirms that the current structural component's processing quality meets the standards, i.e., upon receiving a qualified confirmation instruction based on the quality inspection data, it proves that the current node's construction operation and automated acceptance process have formed an information interaction loop. At this point, the underlying servo unloading and attitude recovery program is quickly triggered, controlling multiple clamping units to release the mechanical locks. The stopping action involves sequentially driving the self-locking robotic arm to open the clamping gripper, cutting off the power circuit of the electromagnetic adsorption array, and removing the preset clamping force on the contact surface. This causes the robot to detach from the rigid physical fixation state with the steel structure to be worked on, restoring it to a freely movable, suspended, or initially connected state. The global navigation path generated in the initial planning stage is retrieved again, the robot's movement drive unit is activated, and the robot is controlled to move along the return trajectory, allowing it to safely and smoothly return to the initial starting point, material supply station, or the next designated work transfer station. This completely ends the complete work process of the current work station, ultimately outputting quality inspection data and work position data. Upon receiving a qualified confirmation instruction based on the quality inspection data feedback, multiple gripping units are controlled to release the mechanical locking action, and the robot is controlled to move along the return trajectory.

[0112] Figure 2 This is a schematic diagram of a robot operation system for high-altitude steel structures, as described in an embodiment of this application. This system can be implemented through software, hardware, or a combination of both, forming all or part of a larger system. Figure 2 As shown, the system includes:

[0113] The pose planning module 201 is configured to acquire the three-dimensional building information model of the steel structure to be operated, the preset operation points, and the point cloud data collected by the robot. It matches and calculates the current pose by matching the three-dimensional building information model with the point cloud data, and generates the robot's movement trajectory, operation trajectory, and return trajectory based on the current pose and the preset operation points.

[0114] The climbing control module 202 is configured to control multiple gripping units of the robot to perform release and fixation actions on the surface of the steel structure to be worked until the robot moves to the preset work point along the movement trajectory. The fixation action includes mechanical locking action.

[0115] The dynamic anti-disturbance module 203 is configured to acquire real-time environmental interference data collected by the robot during the robot's movement along the movement trajectory and operation along the work trajectory, calculate the torque compensation value of each joint of the robot based on the environmental interference data, and adjust the output torque of the robot's multi-joint servo motors based on the torque compensation value.

[0116] The force control operation module 204 is configured to control multiple gripping units to perform mechanical locking actions at preset operation points, and then control the robot's operation units to switch to modular end effectors based on preset task instructions, and control the modular end effectors to execute task instructions according to the operation trajectory in combination with preset force control servo parameters.

[0117] The quality inspection module 205 is configured to control the work unit to switch the modular end effector to the flaw detector after the task instruction is completed, and to obtain quality inspection data and corresponding work position data by scanning through the flaw detector.

[0118] The closed-loop return module 206 is configured to output quality inspection data and work position data. After receiving a qualified confirmation instruction based on the quality inspection data, it controls multiple gripping units to release the mechanical locking action and controls the robot to move according to the return trajectory.

[0119] Based on the above embodiments, as an optional embodiment, the pose planning module 201 is specifically used for: acquiring structural topology nodes in the three-dimensional building information model; performing feature segmentation on the point cloud data to obtain the set of contour boundary points of the steel structure to be operated; calculating the spatial Euclidean distances between the structural topology nodes and the corresponding points in the set of contour boundary points, and constructing a matching error function based on the sum of squares of the spatial Euclidean distances; and cyclically updating the pose parameters based on the matching error function and the preset reference coordinates to obtain the current pose.

[0120] Based on the above embodiments, as an optional embodiment, the climbing control module 202 is specifically used to: extract the current landing point of the robot based on the current pose, and parse the target landing point sequence of the robot according to the movement trajectory; control the first group of gripping units among the multiple gripping units to maintain a fixed action at the current landing point, and control the second group of gripping units among the multiple gripping units to perform a release movement action; after the second group of gripping units reaches the next target landing point in the target landing point sequence along the movement trajectory, control the second group of gripping units to perform a fixed action, and control the first group of gripping units to perform a release movement action.

[0121] Based on the above embodiments, as an optional embodiment, the climbing control module 202 is specifically used to: control the second set of clamping units to apply a preset clamping force to the surface of the steel structure to be operated, and perform an anti-slip contact action; after obtaining the static friction force parameter between the second set of clamping units and the surface of the steel structure to be operated, it reaches a preset safety fall prevention threshold, and control the electromagnetic adsorption array of the second set of clamping units to perform a magnetic adsorption action; after detecting that the adsorption force corresponding to the magnetic adsorption action reaches a preset adsorption threshold, it controls the self-locking mechanical arm of the second set of clamping units to close, and performs a mechanical locking action.

[0122] Based on the above embodiments, as an optional embodiment, the dynamic disturbance rejection module 203 is specifically used for: extracting the current wind speed gradient vector and structural vibration frequency from environmental disturbance data; calculating the wind load disturbance force vector based on the current wind speed gradient vector, and calculating the inertial disturbance force vector based on the structural vibration frequency; mapping the wind load disturbance force vector and the inertial disturbance force vector to the robot's center of mass coordinate system to obtain the robot's real-time center of mass offset coordinates; calculating the projection offset of the real-time center of mass offset coordinates within a preset multi-point support polygon, and calculating the omnidirectional restoring torque based on a preset stiffness damping coefficient and the projection offset; constructing a multi-joint Jacobian matrix based on the robot's current pose, and using the multi-joint Jacobian matrix to perform inverse dynamics on the omnidirectional restoring torque to obtain the torque compensation value of each joint; superimposing the torque compensation value of each joint with a preset initial drive command to obtain the target drive command, and controlling the output torque of the multi-joint servo motor according to the target drive command.

[0123] Based on the above embodiments, as an optional embodiment, the force control operation module 204 is specifically used to: extract the operation contact force threshold and compliance impedance coefficient from the preset force control servo parameters; control the modular end effector to execute the task instruction according to the operation trajectory, and obtain the real-time contact reaction force of the modular end effector during the execution process; calculate the impedance deviation between the real-time contact reaction force and the operation contact force threshold, dynamically correct the operation trajectory based on the impedance deviation and compliance impedance coefficient, and control the modular end effector to continue executing the task instruction along the corrected operation trajectory.

[0124] Based on the above embodiments, as an optional embodiment, the quality inspection module 205 is specifically used to: acquire the internal defect echo signal collected by the ultrasonic array sensor integrated at the flaw detection end, and the surface morphology image collected by the visual sensor integrated at the flaw detection end; perform multimodal data fusion on the internal defect echo signal and the surface morphology image to generate quality inspection data; and map the preset work point to the digital twin mesh corresponding to the three-dimensional building information model to obtain work location data.

[0125] It should be noted that the system provided in the above embodiments is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0126] This embodiment also discloses an electronic device, as shown in the reference. Figure 3 The electronic device may include: at least one processor 301, at least one communication bus 302, user interface 303, network interface 304, and at least one memory 305.

[0127] The communication bus 302 is used to enable communication between these components.

[0128] The user interface 303 may include a display screen and a camera. Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0129] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0130] The processor 301 may include one or more processing cores. The processor 301 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 305, and by calling data stored in memory 305. Optionally, the processor 301 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 301 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 301 and may be implemented as a separate chip.

[0131] The memory 305 may include random access memory (RAM) or read-only memory. Optionally, the memory 305 may include a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 305 may also be at least one storage device located remotely from the aforementioned processor 301. Figure 3 As shown, the memory 305, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a robot operation method for high-altitude steel structures.

[0132] exist Figure 3 In the electronic device shown, the user interface 303 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 301 can be used to call an application program stored in the memory 305 for a robot operation method for high-altitude steel structures. When executed by one or more processors 301, the electronic device performs one or more methods as described in the above embodiments.

[0133] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0134] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0135] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0136] 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0137] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0138] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 305 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 of the various embodiments of this application. The aforementioned memory 305 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0139] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the disclosure in this specification. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are considered exemplary only, and the scope of this application is defined by the claims.

Claims

1. A robot work method for a high-altitude steel structure, characterized by, The method includes: The robot acquires a 3D building information model of the steel structure to be operated, preset work points, and point cloud data collected by the robot. It then matches and calculates the current pose by matching the 3D building information model with the point cloud data, and generates the robot's movement trajectory, operation trajectory, and return trajectory based on the current pose and the preset work points. The robot's multiple gripping units are controlled to cyclically perform release and fixation actions on the surface of the steel structure to be worked on until the robot moves along the movement trajectory to the preset work point, wherein the fixation action includes a mechanical locking action; During the process of the robot moving along the movement trajectory and working along the operation trajectory, the environmental interference data collected by the robot in real time is acquired, the torque compensation value of each joint of the robot is calculated based on the environmental interference data, and the output torque of the multi-joint servo motor of the robot is adjusted based on the torque compensation value. After controlling the multiple clamping units to perform the mechanical locking action at the preset work point, the robot's work unit is controlled to switch to a modular end effector based on the preset task instructions, and the modular end effector is controlled to execute the task instructions according to the work trajectory in combination with the preset force control servo parameters; After the task instruction is completed, the control unit switches the modular end effector to a flaw detector and obtains quality inspection data and corresponding work position data by scanning through the flaw detector. The system outputs the quality inspection data and the work position data. Upon receiving a pass confirmation instruction based on the quality inspection data, it controls the multiple clamping units to release the mechanical locking action and controls the robot to move according to the return trajectory.

2. The method of claim 1, wherein, The step of matching and calculating the current pose by the 3D building information model and the point cloud data specifically includes: Obtain the structural topology nodes in the three-dimensional building information model; The point cloud data is segmented to obtain the set of contour boundary points of the steel structure to be operated; Calculate the spatial Euclidean distance between the structural topology node and each corresponding point in the contour boundary point set, and construct a matching error function based on the sum of the squares of each spatial Euclidean distance; The pose parameters are updated cyclically based on the matching error function and the preset reference coordinates to obtain the current pose.

3. The method of claim 1, wherein, The multiple gripping units controlling the robot cyclically perform release and fixation actions on the surface of the steel structure to be worked, specifically including: The current landing point of the robot is extracted based on the current pose, and the target landing point sequence of the robot is parsed according to the movement trajectory. The first group of clamping units among the plurality of clamping units is controlled to maintain the fixed action at the current landing point, and the second group of clamping units among the plurality of clamping units is controlled to perform the release movement action; After the second set of clamping units reaches the next target landing point in the target landing point sequence along the movement trajectory, the second set of clamping units is controlled to perform the fixing action, and the first set of clamping units is controlled to perform the releasing movement action.

4. The method according to claim 3, characterized in that, The control of the second set of clamping units to perform the fixing action specifically includes: The second set of clamping units is controlled to apply a preset clamping force to the surface of the steel structure to be worked, thereby performing an anti-slip bonding action; After the static friction force parameter between the second set of clamping units and the surface of the steel structure to be operated is obtained to reach the preset safety anti-fall threshold, the electromagnetic adsorption array of the second set of clamping units is controlled to perform magnetic adsorption action. After detecting that the adsorption force corresponding to the magnetic adsorption action reaches a preset adsorption threshold, the self-locking mechanical arm of the second set of clamping units is controlled to close, and the mechanical locking action is executed.

5. The method according to claim 1, characterized in that, The step of calculating the torque compensation value of each joint of the robot based on the environmental interference data, and adjusting the output torque of the multi-joint servo motor of the robot based on the torque compensation value, specifically includes: Extract the current wind speed gradient vector and structural vibration frequency from the environmental disturbance data; Calculate the wind load disturbance vector based on the current wind speed gradient vector, and calculate the inertial disturbance vector based on the structural vibration frequency; The wind load disturbance vector and the inertial disturbance vector are mapped to the robot's center of mass coordinate system to obtain the robot's real-time center of mass offset coordinates; Calculate the projection offset of the real-time center of gravity offset coordinates within a preset multi-point support polygon, and calculate the omnidirectional restoring torque based on the preset stiffness damping coefficient and the projection offset. Based on the robot's current pose, a multi-joint Jacobian matrix is ​​constructed. The multi-joint Jacobian matrix is ​​then used to perform inverse dynamics on the omnidirectional restoring torque to obtain the torque compensation value for each joint. The torque compensation value of each joint is superimposed with the preset initial drive command to obtain the target drive command, and the output torque of the multi-joint servo motor is controlled according to the target drive command.

6. The method according to claim 1, characterized in that, The process of controlling the modular end effector to execute the task instruction according to the work trajectory by combining preset force control servo parameters specifically includes: Extract the working contact force threshold and compliance resistance coefficient from the preset force control servo parameters; The modular end effector is controlled to execute the task instruction according to the work trajectory, and the real-time contact reaction force of the modular end effector is acquired during the execution process; The impedance deviation between the real-time contact reaction force and the working contact force threshold is calculated. Based on the impedance deviation and the compliance impedance coefficient, the working trajectory is dynamically corrected, and the modular end effector is controlled to continue executing the task command along the corrected working trajectory.

7. The method according to claim 1, characterized in that, The process of acquiring quality inspection data and corresponding work location data through the flaw detection end specifically includes: The internal defect echo signal collected by the ultrasonic array sensor integrated at the flaw detection end, and the surface morphology image collected by the visual sensor integrated at the flaw detection end are obtained. The internal defect echo signal and the surface morphology image are fused using multimodal data to generate the quality inspection data; The preset work points are mapped onto the digital twin mesh corresponding to the three-dimensional building information model to obtain the work location data.

8. A robotic operation system for high-altitude steel structures, characterized in that, The system includes: The pose planning module is configured to acquire a 3D building information model of the steel structure to be operated, preset work points, and point cloud data collected by the robot. It matches and calculates the current pose by matching the 3D building information model with the point cloud data, and generates the robot's movement trajectory, operation trajectory, and return trajectory based on the current pose and the preset work points. The climbing control module is configured to control multiple gripping units of the robot to cyclically perform release and fixation actions on the surface of the steel structure to be worked on, until the robot moves along the movement trajectory to the preset work point, wherein the fixation action includes a mechanical locking action; A dynamic anti-interference module is configured to acquire real-time environmental interference data collected by the robot during the robot's movement along the movement trajectory and operation along the work trajectory, calculate the torque compensation value of each joint of the robot based on the environmental interference data, and adjust the output torque of the robot's multi-joint servo motors based on the torque compensation value. The force control operation module is configured to control the multiple clamping units to perform the mechanical locking action at the preset operation point, and then control the robot's operation unit to switch to a modular end effector based on a preset task instruction, and control the modular end effector to execute the task instruction according to the operation trajectory in combination with preset force control servo parameters; The quality inspection module is configured to control the work unit to switch the modular end effector to a flaw detector after the task instruction is completed, and to obtain quality inspection data and corresponding work position data by scanning through the flaw detector. The closed-loop return module is configured to output the quality inspection data and the operation position data. After receiving a qualified confirmation instruction based on the quality inspection data, it controls the multiple clamping units to release the mechanical locking action and controls the robot to move according to the return trajectory.

9. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. The user interface and the network interface are both used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1-7.

Images