Intelligent control method and system for intelligent installation robot of electromechanical module

Abstract

This application relates to the field of intelligent control technology and discloses an intelligent control method and system for an intelligent installation robot of electromechanical modules. The method includes: fusing multi-sensor data using a Kalman filter algorithm to generate a standardized spatial state vector; performing 3D path planning on a BIM model based on this vector to obtain a path instruction set; using a visual servoing algorithm to precisely position and correct the path instruction set to obtain corrected motion instructions; generating synchronous control instructions through multi-actuator synchronous coordination processing; and finally, combining LiDAR scanning data for safety obstacle avoidance assessment processing to obtain safe execution instructions, thereby achieving intelligent and precise installation of the electromechanical modules. This application solves the technical problems in intelligent installation of electromechanical modules, such as inaccurate multi-sensor data fusion, lack of 3D intelligence in path planning, insufficient accuracy of visual servoing, difficulty in multi-actuator synchronous coordination, and lack of predictive capability in safety obstacle avoidance.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology, and in particular to an intelligent control method and system for an intelligent installation robot for electromechanical modules. Background Technology

[0002] With the development of mechanization, digitalization, and intelligentization in the construction industry, electromechanical module installation technology is also constantly advancing. Currently, existing electromechanical module installation technologies mainly employ methods such as setting up lifting points and using hoists for lifting, spider cranes for assisted lifting, and hydraulic trucks for manual control and adjustment. These technologies reduce the intensity of manual labor to some extent, but still rely primarily on manual operation. Existing installation equipment is typically equipped with basic sensors and control systems, capable of performing simple position adjustments and force control. Some advanced equipment also integrates visual recognition and path planning functions to assist operators in performing precise installation operations.

[0003] However, existing technologies have many shortcomings: First, labor costs are high, as hoisting and installation operations require a large number of manual laborers for tasks such as processing and laying hoisting points, lifting and hauling hoists, moving trolleys back and forth to adjust positions, and manually adjusting the relative positions of pipes; second, installation efficiency is low, as supports block installation space, requiring pipe sections to be installed individually with short distances between supports, and each process requires manual adjustment and control, making the adjustment process very slow and complex; third, safety hazards are high, as excessive manual intervention in the hoisting process poses significant safety risks; and fourth, the technological content is low, as existing installation methods are not technologically advanced and have not formed an efficient and orderly automated transportation, positioning, lifting, and installation system.

[0004] Based on an in-depth analysis of the aforementioned technical shortcomings, existing technologies suffer from more fundamental technical problems in intelligent control methods: sensor data processing lacks an effective multi-dimensional fusion mechanism, resulting in the control system's inability to obtain accurate spatial state information; path planning algorithms cannot be effectively integrated with BIM models, lacking intelligent planning capabilities in three-dimensional space; visual servo control lacks precision, failing to meet millimeter-level precision positioning requirements; multi-actuator coordinated control lacks a synchronization mechanism, easily leading to inter-axis interference and control conflicts; and the safety obstacle avoidance system lacks predictive and layered protection capabilities, failing to effectively cope with complex safety threats in dynamic environments. These technical deficiencies at the control method level directly restrict the development of electromechanical module intelligent installation robots towards higher levels of automation and intelligence. Summary of the Invention

[0005] This application provides an intelligent control method and system for an intelligent installation robot of electromechanical modules, which solves the technical problems of inaccurate multi-sensor data fusion, lack of three-dimensional intelligence in path planning, insufficient visual servo accuracy, difficulty in synchronous coordination of multiple actuators, and lack of predictability in safety obstacle avoidance in intelligent installation of electromechanical modules, thereby improving the intelligence level and precision positioning capability of installation control.

[0006] Firstly, this application provides an intelligent control method for an intelligent robot for installing electromechanical modules, the intelligent control method for the intelligent robot for installing electromechanical modules comprising:

[0007] Step S1: Multidimensional fusion processing of displacement sensor data, angle sensor data, and visual recognition data is performed using the Kalman filter algorithm to obtain a standardized spatial state vector.

[0008] Step S2: Perform three-dimensional path planning processing on the installation location information in the BIM model based on the standardized spatial state vector to obtain a path instruction set;

[0009] Step S3: The path instruction set is precisely positioned and corrected using a visual servoing algorithm to obtain corrected motion instructions.

[0010] Step S4: Perform synchronous coordination processing on the motion parameters of the multiple actuators according to the modified motion command to obtain the synchronous control command;

[0011] Step S5: Combine the synchronous control command with the lidar scanning data to perform a safety obstacle avoidance assessment and obtain a safety execution command.

[0012] Secondly, this application provides an intelligent control system for an intelligent installation robot of electromechanical modules, the intelligent control system of the intelligent installation robot of electromechanical modules comprising:

[0013] The fusion module is used to perform multi-dimensional fusion processing on displacement sensor data, angle sensor data and visual recognition data through Kalman filtering algorithm to obtain a standardized spatial state vector.

[0014] The planning module is used to perform three-dimensional path planning processing on the installation location information in the BIM model based on the standardized spatial state vector to obtain a path instruction set;

[0015] The correction module is used to perform precise positioning correction processing on the path instruction set through a visual servoing algorithm to obtain corrected motion instructions;

[0016] The coordination module is used to perform synchronous coordination processing on the motion parameters of multiple actuators according to the modified motion command, so as to obtain the synchronous control command;

[0017] The evaluation module is used to combine the synchronous control command with the lidar scanning data to perform a safety obstacle avoidance evaluation and obtain a safety execution command.

[0018] Thirdly, an intelligent control device for an electromechanical module intelligent installation robot is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the intelligent control device for the electromechanical module intelligent installation robot to execute the aforementioned intelligent control method for the electromechanical module intelligent installation robot.

[0019] Fourthly, a computer-readable storage medium is provided, wherein instructions are stored therein, which, when executed on a computer, cause the computer to perform the aforementioned intelligent control method for an intelligent electromechanical module installation robot.

[0020] The technical solution provided in this application uses a Kalman filter algorithm to perform multi-dimensional fusion processing on displacement sensor data, angle sensor data, and visual recognition data. This solves the technical problems of inaccurate single-sensor data and inconsistent multi-sensor data in traditional installation methods. The state estimation and error correction characteristics of the Kalman filter algorithm can effectively eliminate sensor noise and time delay. The generated standardized spatial state vector provides a high-precision position and attitude information basis for subsequent control. Based on the standardized spatial state vector, three-dimensional path planning is performed on the installation position information in the BIM model, and then... The heuristic search of the algorithm and the obstacle recognition of the DBSCAN clustering algorithm have achieved a technological breakthrough from two-dimensional planar planning to three-dimensional spatial intelligent planning. The algorithm's global optimal search capability, combined with the DBSCAN algorithm's dynamic obstacle recognition capability, enables path planning to adapt to complex construction environments and avoid dynamic obstacles, significantly improving the intelligence level and environmental adaptability of path planning.

[0021] The application of visual servoing algorithms in the precision installation of electromechanical modules, through the combination of normalized cross-correlation matching and PID control, improves installation accuracy from the traditional centimeter level to the millimeter level. The real-time feedback characteristics of the visual servoing algorithm and the dynamic adjustment capability of the PID controller work together to enable the robot to correct its motion trajectory in real time based on visual feedback information, solving the technical problem of insufficient accuracy in traditional open-loop control methods. Multi-actuator synchronization and coordination processing, through a master-slave control strategy and cross-coupling control algorithm, solves the inter-axis interference and synchronization error problems existing in traditional independent axis control. The interference compensation characteristics of the cross-coupling control algorithm can effectively suppress the mutual influence between actuators, and the predictive compensation of feedforward control further improves the system's response speed and control accuracy. Safety obstacle avoidance assessment processing adopts the Kalman prediction algorithm and a layered protection strategy, transforming passive obstacle avoidance into active predictive protection. The motion trajectory prediction capability of the Kalman prediction algorithm, combined with the gradient safety response mechanism of the layered protection area, can minimize interference with normal operations while ensuring safety, significantly reducing the safety hazards existing in traditional manual intervention methods, and improving the safety and continuity of intelligent installation of electromechanical modules. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of an embodiment of the intelligent control method for the intelligent installation robot of the electromechanical module in this application.

[0024] Figure 2 This is a schematic diagram of an embodiment of the intelligent control system of the electromechanical module intelligent installation robot in this application;

[0025] Figure 3 This is a schematic block diagram of the intelligent control device of the electromechanical module intelligent installation robot in this embodiment of the invention. Detailed Implementation

[0026] This application provides an intelligent control method and system for an intelligent installation robot of electromechanical modules. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data used can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0027] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the intelligent control method for the electromechanical module intelligent installation robot in this application includes:

[0028] Step S1: Multidimensional fusion processing of displacement sensor data, angle sensor data, and visual recognition data is performed using the Kalman filter algorithm to obtain a standardized spatial state vector.

[0029] Step S2: Perform three-dimensional path planning processing on the installation location information in the BIM model based on the standardized spatial state vector to obtain the path instruction set;

[0030] Step S3: The path instruction set is precisely positioned and corrected using a visual servoing algorithm to obtain the corrected motion instructions.

[0031] Step S4: Synchronize and coordinate the motion parameters of multiple actuators according to the modified motion command to obtain the synchronization control command;

[0032] Step S5 involves combining the synchronous control commands with lidar scanning data to perform a safety obstacle avoidance assessment, resulting in a safety execution command.

[0033] It is understood that the executing entity of this application can be the intelligent control system of the electromechanical module intelligent installation robot, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiment uses a server as an example for illustration.

[0034] Specifically, the multidimensional data fusion processing first performs timestamp correction on the X, Y, and Z axis coordinate data collected by the displacement sensor to eliminate sampling delay differences between different sensors. Timestamp correction ensures data temporal consistency by rearranging the data from each sensor according to a unified time reference. The pitch, yaw, and roll angle data collected by the angle sensor are denoised using a noise filter. The filter employs a low-pass filtering principle to remove high-frequency noise signals while retaining valid angle change information. Depth calculation is performed on the binocular stereo images acquired by the visual recognition system. The three-dimensional spatial depth information of the target object is calculated by the pixel difference between the left and right camera images. The depth value is equal to the baseline distance multiplied by the focal length and then divided by the pixel difference. The Kalman filter uses the time-aligned position data, filtered angle data, and target depth information as observations, and generates a standardized spatial state vector containing position, velocity, and acceleration through two stages: state prediction and state update.

[0035] The 3D path planning process transforms the current position coordinates in the standardized spatial state vector with the preset electromechanical module installation positions in the BIM model. This transformation includes translation and rotation, unifying the position information from different coordinate systems into the robot's body coordinate system. The algorithm performs path search on a 3D raster map. It calculates the actual cost from the starting point to the current node and the heuristic cost from the current node to the destination, selecting the path node with the minimum total cost for expansion. The DBSCAN clustering algorithm identifies obstacles in the environmental point cloud data acquired by LiDAR. By setting a neighborhood radius and a minimum point count threshold, the algorithm groups point cloud data with similar density into the same obstacle. The path optimization objective function comprehensively considers three factors: path length, smoothness, and time cost. It calculates the comprehensive evaluation value of the path through a weighted summation method and selects the path with the optimal evaluation value as the final path instruction set.

[0036] Visual servoing precision positioning correction uses a normalized cross-correlation algorithm to match the position coordinates in the path command set with the feature templates of the electromechanical module. The algorithm calculates the correlation coefficient between the template image and the real-time image; the position corresponding to the maximum correlation coefficient is the actual position of the target object. The difference between the visual positioning result and the expected position in the path command set constitutes the visual error vector, which includes position deviations in the X, Y, and Z directions. A PID controller performs proportional-integral-derivative (PID) control on the visual error vector. The proportional term reflects the current error magnitude, the integral term eliminates steady-state error, and the derivative term predicts the error change trend. The three terms are combined to output the position correction. A coordinate transformation matrix transforms the position correction from the image coordinate system to the robot coordinate system, and the corrected motion parameters form the corrected motion command.

[0037] The multi-actuator synchronization and coordination process sets the hydraulic lifting mechanism parameters in the corrected motion command as the master spindle control reference. The master spindle motion reference signal includes motion parameters such as position, velocity, and acceleration. The X-axis motor, Y-axis motor, and Z-axis fine-tuning mechanism act as slave axes, synchronizing with the master spindle motion reference signal based on the time reference. The synchronization process uses an interpolation algorithm to distribute the master spindle's motion trajectory to each slave axis. The multi-actuator dynamic coupling model describes the interaction between the actuators, calculating the coupling force and torque between each axis by establishing mass, damping, and stiffness matrices. The cross-coupling controller generates compensation control quantities based on the synchronization error between axes, eliminating mutual interference between actuators. The feedforward control term pre-calculates the required control input based on the desired trajectory, and superimposes it with the cross-coupling control quantity to form the synchronization control command.

[0038] The obstacle avoidance assessment process uses a Kalman prediction algorithm to fuse synchronous control commands and LiDAR scanning data. The prediction algorithm, based on the target's motion model, uses historical position and velocity information to predict future trajectories. Layered protection zones are defined, dividing the robot's surrounding space into warning zones, protection zones, and emergency zones based on distance. The boundaries of each zone are determined using Euclidean distance calculations. The risk assessment quantification model takes the obstacle's position, velocity, and direction of motion as inputs and calculates a safety risk coefficient through weighted averages. A higher risk coefficient indicates a higher collision risk. The safety decision-making logic selects appropriate protective actions based on the risk coefficient, including different levels of safety response measures such as deceleration, pausing, and emergency stop. Finally, it generates safety execution commands to control the robot's actual movement.

[0039] In one specific embodiment, step S1 includes:

[0040] The X, Y, and Z axis coordinate data collected by the displacement sensor are time-stamped to obtain time-aligned position data.

[0041] The pitch angle, yaw angle and roll angle data collected by the angle sensor are input into the noise filter for noise reduction processing to obtain the filtered angle data;

[0042] Depth calculation is performed on the binocular stereo images acquired by the visual recognition system to obtain the target depth information;

[0043] Time-aligned position data, filtered angle data, and target depth information are input into a Kalman filter for state estimation to obtain a standardized spatial state vector.

[0044] Specifically, when performing timestamp correction processing on the X, Y, and Z axis coordinate data collected by displacement sensors, the original data packets of each sensor are read. Each data packet contains position coordinate values ​​and corresponding timestamp information. Due to differences in sampling frequency and data transmission delay among different sensors, spatial position data at the same moment may have different timestamps. The timestamp correction algorithm establishes a unified time reference and rearranges all sensor data according to a standard time axis. The correction process includes time offset calculation and data interpolation compensation. When the timestamp of a certain sensor data deviates from the reference time, the accurate position value at that moment is calculated using a linear interpolation method. Finally, an X, Y, and Z axis coordinate sequence with a unified time reference is generated, forming time-aligned position data.

[0045] When pitch, yaw, and roll angle data collected by the angle sensor are input into a noise filter for denoising, the angle sensor is based on a combination of gyroscope and accelerometer measurement principles. The gyroscope measures the angular velocity signal, and the accelerometer measures the gravitational acceleration component. The three-axis attitude angles are obtained through integration and geometric calculation. However, the sensor signal contains interference components such as high-frequency noise, zero-point drift, and temperature drift. The noise filter uses a low-pass filtering principle, setting a cutoff frequency to distinguish between valid and noise signals. When the frequency of the angle signal is lower than the cutoff frequency, the signal passes through, while noise components above the cutoff frequency are attenuated. The filter smooths continuous angle sample values ​​using a weighted averaging algorithm, calculating the weighted average of the current angle value and historical angle values. The weighting coefficients are dynamically adjusted according to the rate of signal change; signals with slow changes receive a larger historical weight, while signals with rapid changes receive a smaller historical weight. After filtering, stable pitch, yaw, and roll angle values ​​are output, constituting the filtered angle data.

[0046] When performing depth calculations on stereo images acquired by a visual recognition system, the stereo camera consists of two imaging units: a left camera and a right camera. The horizontal distance between the two cameras is called the baseline distance. The horizontal position of the same target object in the images of the left and right cameras differs; this difference is called disparity. The depth calculation algorithm first performs feature point matching on the left and right images. Through corner detection and feature descriptor matching, it finds the corresponding pixel positions of the same spatial point in the two images. Then, it calculates the difference in pixel coordinates of the corresponding points to obtain the disparity value. The depth value is equal to the baseline distance multiplied by the camera focal length and then divided by the disparity value. The disparity is larger when the target object is closer to the camera and smaller when it is farther away. This geometric relationship allows for accurate calculation of the target object's depth distance in three-dimensional space. The depth calculation process also needs to consider camera intrinsic parameters and distortion correction, ultimately generating three-dimensional coordinate data containing the target object's spatial depth information, forming the target depth information.

[0047] When the Kalman filter performs state estimation on time-aligned position data, filtered angle data, and target depth information, it establishes a state-space model of the robot's motion. The state vector includes motion parameters such as position, velocity, acceleration, and attitude angle. The Kalman filter algorithm consists of a prediction step and an update step. The prediction step predicts the current state value based on the state estimate from the previous moment and the motion model, while simultaneously calculating the covariance matrix of the prediction error. The update step compares the sensor observations with the predicted values, calculates the observation residuals and the Kalman gain. The Kalman gain reflects the confidence weight between the observations and the predictions; it is larger when the sensor measurement accuracy is high and smaller when the prediction model accuracy is high. The predicted state is corrected using the Kalman gain to obtain the optimal state estimate. The filter also updates the error covariance matrix to reflect the uncertainty of the estimation. After multiple iterations, it outputs comprehensive state information including position coordinates, velocity, acceleration, and attitude angle, forming a standardized spatial state vector.

[0048] In one specific embodiment, step S2 includes:

[0049] The coordinate system is transformed between the position coordinates in the standardized spatial state vector and the installation position information of the electromechanical modules in the BIM model to obtain the target point set.

[0050] A method for analyzing a 3D raster map based on a set of target points. The algorithm performs path search processing to obtain an initial path sequence;

[0051] Based on the environmental point cloud data obtained by lidar scanning, the initial path sequence is processed by the DBSCAN clustering algorithm for obstacle identification to obtain dynamic obstacle boundary information;

[0052] The dynamic obstacle boundary information is input into the path optimization objective function, and the path length, smoothness and time cost are comprehensively optimized to obtain the path instruction set.

[0053] Specifically, when performing coordinate system transformation between the position coordinates in the standardized spatial state vector and the electromechanical module installation position information in the BIM model, the pre-set electromechanical module installation position information in the BIM model is first read. The BIM model uses a building coordinate system, establishing a three-dimensional coordinate system with a fixed point of the building as the origin, while the robot uses its own body coordinate system, establishing a coordinate system with the center of the robot chassis as the origin. The coordinate system transformation algorithm unifies the two coordinate systems through translation and rotation transformations. The translation transformation calculates the position difference between the origins of the two coordinate systems, translating the position information in the BIM coordinate system to the robot coordinate system. The rotation transformation calculates the angular difference between the two coordinate systems and uses a rotation matrix to align the coordinate axes. The transformation process includes recalculating the coordinate values ​​in the X, Y, and Z axes, ultimately generating the installation target position coordinates in the robot coordinate system. These coordinate points constitute a set of target points, each containing spatial position information and corresponding installation posture requirements.

[0054] 3D raster map A During the algorithm's path search processing, the raster map divides the three-dimensional space into regular cubic grids. Each grid cell represents a spatial location, and the grid status is divided into two types: passable and impassable. Grids occupied by obstacles are marked as impassable, while free spaces are marked as passable. A The algorithm is based on the principle of heuristic search. It maintains two lists: an open list to store nodes to be evaluated and a closed list to store nodes that have been evaluated. Starting from the starting node, the algorithm calculates the cost function value of each adjacent node. The cost function includes two parts: the actual distance from the starting point to the current node and the estimated distance from the current node to the destination. The actual distance is calculated by accumulating the path length, and the estimated distance is calculated using the Euclidean distance formula. The algorithm selects the node with the smallest cost function value as the next node to expand, and repeats the expansion process until the destination is found or the open list is empty. The search process generates a sequence of nodes from the starting point to the destination. Each node corresponds to a position coordinate in three-dimensional space, and connecting these coordinate points forms the initial path sequence.

[0055] When using the DBSCAN clustering algorithm to identify obstacles in the initial path sequence, the environmental point cloud data acquired by LiDAR scanning obtains 3D point cloud data of the surrounding environment by emitting a laser beam and receiving reflected signals. The point cloud data contains a large number of spatial coordinate points, each representing the intersection of the laser beam and the object's surface. The DBSCAN clustering algorithm is based on the principle of density clustering. The algorithm sets two key parameters: neighborhood radius and minimum point threshold. The neighborhood radius defines the maximum distance between points, and the minimum point threshold defines the minimum number of points required to form a cluster. The algorithm starts from any unvisited point and searches for all points within its neighborhood radius. If the number of points in the neighborhood is greater than or equal to the minimum point threshold, these points are grouped into the same cluster. Then, the cluster boundary is expanded, adding the neighborhoods of other points in the neighborhood to the cluster. This expansion process is repeated until the cluster can no longer grow. Each cluster represents an obstacle in the environment. The algorithm determines the geometry and size of the obstacle by calculating the spatial range of the cluster boundary points, generating dynamic obstacle boundary information containing the obstacle's location, shape, and size.

[0056] When the path optimization objective function comprehensively optimizes path length, smoothness, and time cost, it uses a weighted summation method to combine the three evaluation indicators into a single optimization objective. Path length is obtained by calculating and summing the distances between adjacent nodes on the path. Path smoothness is evaluated by calculating the angle change between adjacent line segments on the path; the smaller the angle change, the smoother the path. Time cost is calculated based on the robot's speed and acceleration constraints on different path segments. The algorithm assigns different weight coefficients to the three indicators, reflecting the importance of each indicator. The optimization algorithm minimizes the objective function value by adjusting the node positions on the path. The adjustment process includes fine-tuning node positions, replanning path segments, and deleting redundant nodes. When a conflict is found between the dynamic obstacle boundary information and the initial path, the algorithm recalculates the obstacle avoidance path, adds a safety boundary around the obstacle, and replans the detour path. The optimized path avoids obstacles while meeting the comprehensive requirements of length, smoothness, and time, ultimately generating a path instruction set containing detailed motion instructions. Each instruction in the instruction set includes parameters such as target position coordinates, motion speed, and motion direction.

[0057] In one specific embodiment, step S3 includes:

[0058] The location coordinates in the path instruction set are matched with the electromechanical module feature template library using a normalized cross-correlation algorithm to obtain the visual positioning result;

[0059] The positional deviation is calculated based on the visual positioning result and the expected position in the path instruction set to obtain the visual error vector;

[0060] The visual error vector is input into the PID controller for proportional-integral-derivative control processing to obtain the position correction amount.

[0061] Based on the position correction, the motion parameters in the path instruction set are adjusted by the coordinate transformation matrix to obtain the corrected motion instruction.

[0062] Specifically, when matching the location coordinates in the path instruction set with the electromechanical module feature template library using a normalized cross-correlation algorithm, the electromechanical module feature template library contains standard image templates of different types of electromechanical modules under various angles and lighting conditions. Each template image is preprocessed and normalized to a fixed-size pixel matrix. The normalized cross-correlation algorithm determines the target location by calculating the similarity between the template image and the real-time acquired image. The algorithm first divides the real-time image into sliding windows of the same size as the template, and then calculates the cross-correlation coefficient with the template for each window. The cross-correlation coefficient reflects the degree of matching between the two images; the closer the value is to 1, the higher the matching degree. The algorithm normalizes the template and the image to be matched to eliminate the influence of lighting changes and contrast differences. The normalization process includes subtracting the image mean and dividing by the standard deviation, so that images under different lighting conditions have the same statistical characteristics. During the matching process, the algorithm calculates the normalized cross-correlation value of each sliding window position, finds the position with the largest correlation value as the detection position of the target object, and converts the pixel coordinates corresponding to this position into actual spatial coordinates to form a visual localization result. The visual localization result contains the precise position coordinates and pose angle information of the target object in three-dimensional space.

[0063] When calculating the positional deviation between the visual positioning result and the desired position in the path instruction set, the desired position is derived from the theoretical reach position calculated by the path planning algorithm. This position is determined based on the installation position information and path optimization results in the BIM model. Positional deviation calculation compares the actual position coordinates in the visual positioning result with the desired position coordinates, calculating the coordinate differences in the X, Y, and Z axes, and simultaneously calculating the deviations in the pitch, yaw, and roll angles. The deviation calculation process must consider the consistency of the coordinate system, ensuring that the visual positioning result and the desired position use the same coordinate system reference. When they use different coordinate systems, coordinate transformation is required. The sign of the positional deviation indicates the direction of offset; a positive value indicates that the actual position exceeds the desired position, and a negative value indicates that the actual position has not reached the desired position. The absolute value of the deviation indicates the magnitude of the offset distance. The six calculated deviation values, including three positional deviations and three angular deviations, combine to form a six-dimensional visual error vector, which fully describes the difference between the robot's current state and the desired state.

[0064] When the visual error vector is input to a PID controller for proportional-integral-derivative (PID) control, the PID controller performs independent PID calculations for each component of the error vector. The proportional control term is directly proportional to the current error, and the proportional coefficient determines the controller's response strength to the current error. Proportional control can respond quickly to error changes but cannot eliminate steady-state errors. The integral control term accumulates and sums historical errors, and the integral coefficient controls the degree of influence of the accumulated error on the output. Integral control can eliminate steady-state errors but has a slower response speed and is prone to overshoot. The derivative control term controls based on the rate of change of the error, and the derivative coefficient determines the degree of prediction of the error change trend. Derivative control can predict the error development trend and adjust in advance, but it is sensitive to noise. The PID controller weighted sums the outputs of the three control terms to obtain the final control output. Each error component corresponds to a correction control quantity, and the six correction control quantities combine to form the position correction quantity, which includes position adjustment values ​​in the X, Y, and Z directions and adjustment values ​​for the three attitude angles.

[0065] When adjusting the motion parameters in the path command set using coordinate transformation matrices, the position correction amount describes the transformation relationship from the current coordinate system to the target coordinate system. The matrix includes rotation and translation transformations. The rotation transformation matrix is ​​calculated based on the attitude angle correction amount, converting the angle correction amount into elements of the rotation matrix using Euler angles or quaternions. The rotation matrix is ​​used to adjust the robot's attitude orientation. The translation transformation vector directly uses the X, Y, and Z components of the position correction amount to adjust the robot's spatial position. The coordinate transformation process performs matrix operations on the original motion parameters and correction amounts in the path command set. These operations include matrix multiplication and vector addition. Attitude adjustment is achieved through matrix multiplication, and position adjustment is achieved through vector addition. The adjusted motion parameters include the corrected target position coordinates, velocity, acceleration, and attitude angle. These parameters are reorganized to form corrected motion commands. Each parameter in the corrected motion commands undergoes precise calibration based on visual feedback to ensure the robot accurately reaches the target installation position.

[0066] In one specific embodiment, step S4 includes:

[0067] The parameters of the hydraulic lifting mechanism in the corrected motion command are set as the main spindle control reference to obtain the main spindle motion reference signal;

[0068] Based on the spindle motion reference signal, the motion parameters of the X-axis motor, Y-axis motor and Z-axis fine-tuning mechanism are synchronized using a time reference to obtain the slave axis synchronization parameters.

[0069] Based on the shaft synchronization parameters, a multi-actuator dynamic coupling model is established to perform inter-axis interference compensation processing and obtain cross-coupled control quantities.

[0070] The cross-coupling control quantity and the feedforward control term are superimposed and processed to obtain the synchronization control command.

[0071] Specifically, when the parameters of the hydraulic lifting mechanism in the corrected motion command are set as the spindle control reference, the hydraulic lifting mechanism, as the robot's main vertical motion actuator, undertakes the task of precise positioning of the electromechanical module in the Z-axis direction. Its motion parameters include key information such as target height position, lifting speed, acceleration, and motion time. The process of establishing the spindle control reference first extracts the motion trajectory data of the hydraulic lifting mechanism from the corrected motion command. This trajectory data describes the complete motion process from the current position to the target position, including the acceleration motion in the start-up phase, the constant speed running phase, and the deceleration and stopping phase before reaching the target. The algorithm performs time discretization processing on the motion trajectory of the hydraulic lifting mechanism, dividing the continuous motion process into a series of discrete time points, each time point corresponding to a specific position, speed, and acceleration value. The spindle motion reference signal contains the motion state parameters of each sampling point in the time series. These parameters are standardized to form a unified time reference signal, which serves as a reference standard for the synchronous motion of other actuators, ensuring that all actuators perform corresponding motion actions at the same time node.

[0072] When the spindle motion reference signal is used for time reference synchronization of the motion parameters of the X-axis motor, Y-axis motor, and Z-axis fine-tuning mechanism, the time reference synchronization algorithm first analyzes the temporal characteristics of the spindle motion reference signal, including the motion period, sampling interval, and key time nodes. Then, it maps the motion parameters of the X-axis motor, Y-axis motor, and Z-axis fine-tuning mechanism onto the same time axis. The synchronization process includes two stages: time axis alignment and motion parameter interpolation. Time axis alignment adjusts the start and end times of each slave axis's motion to ensure that the motion time range of all actuators is consistent with the spindle. Motion parameter interpolation calculates the corresponding motion parameters of each slave axis at the spindle's time nodes using mathematical interpolation methods. The interpolation algorithm selects an appropriate interpolation method based on the motion characteristics of each slave axis; linear interpolation or spline interpolation is used for position parameters, while difference calculation or polynomial fitting methods are used for velocity and acceleration parameters. Synchronization processing also needs to consider the physical constraints of each actuator, including maximum motion speed, acceleration limits, and motion range constraints. When the synchronization parameters of a slave axis exceed the physical limits, the algorithm automatically adjusts its motion trajectory to meet the constraints, and finally generates slave axis synchronization parameters that are synchronized with the master axis time reference. These parameters ensure that each actuator maintains a precise time coordination relationship during motion.

[0073] When performing inter-axis interference compensation by establishing a multi-actuator dynamic coupling model based on axis synchronization parameters, the dynamic coupling model describes the interaction relationships and dynamic effects between each actuator. The model considers the robot's inertial characteristics, gravity effects, friction, and mechanical coupling effects between actuators. The coupling model establishment process first analyzes the robot's mechanical structure and mass distribution, calculates the inertial forces and torques generated by each actuator's motion, and then establishes a coupling matrix describing the mutual influence between actuators. Matrix elements reflect the magnitude of the interference force or torque generated by one actuator's motion on other actuators. The inter-axis interference compensation algorithm calculates the interference terms between each axis based on the coupling model. The interference calculation includes two parts: direct coupling interference and indirect coupling interference. Direct coupling interference originates from the direct mechanical connection between actuators, while indirect coupling interference originates from the deformation and vibration transmission of the shared support structure. The compensation process establishes an interference observer to estimate the interference force on each axis in real time. The observer infers the magnitude of the interference based on the difference between the actual motion state of the actuator and the theoretical motion model, and then calculates the corresponding compensation control quantity to counteract the interference effect. The cross-coupling control algorithm calculates the synchronization error between each axis and generates cross-coupling control quantities to reduce inter-axis interference. These control quantities include position corrections and torque compensations for each actuator, ensuring that the actuators can maintain precise synchronous motion even when they affect each other.

[0074] When cross-coupled control quantities and feedforward control terms are superimposed, the feedforward control term pre-calculates the required control input based on the desired motion trajectory. The feedforward control algorithm calculates the driving force or driving torque required to complete the specified motion under ideal conditions, based on the dynamic model of each actuator and the desired motion parameters. The calculation of the feedforward control term includes three components: the static force required to overcome gravity, the dynamic force required to generate the desired acceleration, and the compensation force required to overcome frictional resistance. The static force is calculated based on the robot's weight distribution and current posture, the dynamic force is calculated based on the desired acceleration and inertial parameters, and the frictional compensation force is calculated based on the motion speed and friction model. The superposition process groups the cross-coupled control quantities and feedforward control terms according to their corresponding actuators and adds them together. The addition operation considers the direction and magnitude of each control quantity to ensure that the superimposed control command can simultaneously meet the requirements of trajectory tracking and interference suppression. The superposition operation also includes saturation limiting processing of control quantities. When the total control quantity of an actuator exceeds its maximum output capacity, the algorithm reduces each control component proportionally, keeping the control direction unchanged while limiting the control amplitude within the actuator's capability range. Finally, it generates a synchronous control command containing position, speed, and torque commands. This command guides each actuator to coordinate and complete the precise installation of the electromechanical module.

[0075] In one specific embodiment, step S5 includes:

[0076] The synchronous control commands and lidar scanning data are processed by the Kalman prediction algorithm to obtain dynamic obstacle trajectory prediction information.

[0077] Based on the predicted motion trajectory information of dynamic obstacles, the robot's surrounding environment is divided into layered protection zones to obtain the safety zone boundaries of the warning zone, protection zone, and emergency zone.

[0078] Based on the safety zone boundary, a risk assessment quantification model is used to calculate the motion speed and acceleration parameters in the synchronization control command to obtain the safety risk coefficient.

[0079] The safety risk coefficient is input into the safety decision logic to select protection actions and obtain safety execution instructions.

[0080] Specifically, when synchronous control commands and lidar scanning data are processed using the Kalman prediction algorithm, the algorithm predicts future trajectories based on the dynamic obstacle's motion model. The prediction process first establishes a state-space model of the obstacle, with the state vector containing the obstacle's position coordinates, velocity, and acceleration information. LiDAR scanning data provides the obstacle's current spatial position information. The obstacle's velocity is calculated through comparative analysis of multiple consecutive frames of scanning data. Velocity calculation uses a difference method between adjacent position coordinates: subtracting the previous position from the current position and dividing by the time interval yields the velocity component. The Kalman prediction algorithm's prediction step estimates the next moment's position and velocity based on the obstacle's motion model and current state. The prediction model assumes the obstacle maintains uniform or uniformly accelerated motion over a short period, calculating the predicted state value using a state transition matrix. The algorithm's update step compares the real-time lidar observation data with the predicted value, calculates the observation residual and Kalman gain, and then corrects the predicted state to obtain the optimal estimate. The spatial relationship between the robot's motion trajectory in the synchronous control command and the predicted obstacle trajectory is analyzed. The collision risk is assessed by calculating the minimum distance between the two trajectories at future time. When the predicted distance is less than the safety threshold, it is marked as a potential risk area. Finally, dynamic obstacle motion trajectory prediction information containing the future position sequence of obstacles and risk assessment information is generated.

[0081] When using dynamic obstacle trajectory prediction information to divide the robot's surrounding environment into layered protection zones, the layered protection strategy establishes a multi-level safety protection system based on the distance attenuation principle. The algorithm first establishes a polar coordinate system with the robot's geometric center as the origin, and then divides the surrounding space into three concentric circular regions according to preset distance thresholds. The warning zone is a ring-shaped area two to five meters from the robot's center point. This area is mainly used for early warning and speed adjustment; when a dynamic obstacle enters the warning zone, a speed reduction command is triggered, but movement does not stop. The protection zone is a ring-shaped area one to two meters from the robot's center point. This area is used for motion mode switching and fine control; when an obstacle enters the protection zone, the robot switches to a slow motion mode and increases the sensor sampling frequency. The emergency zone is a circular area zero to one meter from the robot's center point. This area is the final safety barrier; any obstacle entering the emergency zone will trigger an immediate stop command. The region division algorithm dynamically adjusts the shape and boundaries of each region based on the dynamic obstacle trajectory prediction information. When the obstacle's movement speed is high, the range of each protection zone is expanded accordingly; when the obstacle's movement direction is away from the robot, the protection zone is appropriately reduced. The boundary calculation process takes into account the geometric dimensions and motion uncertainties of obstacles, adds a safety boundary margin around the obstacles, and finally forms safe area boundary data for warning zones, protection zones and emergency zones containing accurate boundary coordinates and regional attributes.

[0082] When calculating the motion speed and acceleration parameters in the synchronous control commands for the safety zone boundary using a risk assessment quantification model, the model comprehensively considers three key factors: collision probability, severity of collision consequences, and system response time to calculate the safety risk coefficient. Collision probability calculation is based on the spatial relationship analysis between obstacles and the robot's trajectory. The algorithm calculates the minimum approach distance between two moving objects at future moments, then establishes a probability density function based on distance distribution and motion uncertainty, and integrates to calculate the probability value of a collision event. The severity of collision consequences is assessed considering the robot's speed, mass, obstacle characteristics, and surrounding environmental factors. Collisions at high speeds are more severe, and obstacles involving people pose a higher risk level than obstacles involving stationary objects. System response time analysis includes sensor detection delay, data processing time, control command transmission time, and actuator response time. The total response time determines the time window within which the robot can take avoidance actions after detecting a hazard. The risk coefficient calculation uses a weighted summation method to combine the three factors into a single numerical index. The weighting coefficient is adjusted according to the safety requirements of different application scenarios. The calculation formula mathematically models the impact of collision probability, severity of consequences and response time delay to obtain a safety risk coefficient ranging from zero to one. The closer the value is to one, the higher the risk.

[0083] When the safety risk coefficient is input into the safety decision-making logic for selecting protective actions, the logic uses a multi-level threshold judgment mechanism to select the appropriate protective action. The logic establishes a mapping table between risk levels and protective actions, with different risk coefficient ranges corresponding to different safety response strategies. When the safety risk coefficient is in the low-risk range, the logic chooses to maintain the current motion state and execute the normal action, while increasing the sensor monitoring frequency to continuously track risk changes. When the risk coefficient enters the medium-risk range, the logic chooses to slow down the motion, extending reaction time and reducing collision consequences. When the risk coefficient reaches the high-risk range, the logic chooses to pause the current motion as a protective action; the robot stops its main movement and maintains its current position, while simultaneously initiating an active obstacle avoidance algorithm to replan a safe path. When the risk coefficient exceeds the extremely high-risk threshold, the logic chooses an emergency stop action; all movement immediately stops and the audible and visual alarm system is activated, while a safety alarm message is sent to the operator. The protection action selection process also considers the importance of the current task and the cost of interruption. A more conservative safety strategy is adopted for critical installation tasks, while a relatively flexible risk avoidance strategy is adopted for ordinary tasks that can be repeated. Finally, a safe execution instruction containing specific action instructions, execution parameters and safety constraints is generated.

[0084] In one specific embodiment, the process of dividing the robot's surrounding environment into layered protection zones based on the predicted trajectory information of dynamic obstacles can specifically include the following steps:

[0085] The position coordinates in the dynamic obstacle trajectory prediction information are processed by calculating the Euclidean distance from the robot's center point to obtain obstacle distance data;

[0086] Based on obstacle distance data, the space within a 5-meter radius around the robot is divided into three concentric circles to obtain the initial area range of a 2-5 meter warning zone, a 1-2 meter protection zone, and a 0-1 meter emergency zone.

[0087] Based on the initial area range and the temperature characteristics of people detected by thermal imaging sensors, a personnel identification algorithm is used to obtain personnel location distribution information.

[0088] The personnel location distribution information is input into the safety boundary dynamic adjustment algorithm for regional boundary correction processing to obtain the safety area boundaries of the warning zone, protection zone and emergency zone.

[0089] Specifically, when calculating the Euclidean distance from the robot's center point to the position coordinates in the dynamic obstacle trajectory prediction information, the Euclidean distance calculation is based on the formula for the straight-line distance between two points in three-dimensional space. The calculation process first determines the spatial coordinates of the robot's center point as the reference origin, which is usually located at the geometric center of the robot's chassis. The obstacle position coordinates are derived from the dynamic obstacle trajectory prediction information, including the obstacle's X-axis, Y-axis, and Z-axis coordinates in three-dimensional space. These coordinate values ​​are obtained through the fusion processing of LiDAR scanning and Kalman prediction algorithms. The distance calculation algorithm calculates the spatial distance between the position coordinates of each obstacle and the robot's center point. The calculation process includes two steps: solving for the coordinate difference and square root operation. First, the differences between the obstacle coordinates and the robot's center point coordinates in the X, Y, and Z directions are calculated. Then, the three differences are squared and summed. Finally, the square root of the sum is taken to obtain the straight-line distance. The algorithm calculates the distance at each time point on the predicted trajectory, generating a sequence of distance changes between the obstacle and the robot. This sequence describes the trend of the obstacle approaching or moving away from the robot. The distance data contains key information such as the current distance value, the minimum predicted distance, and the rate of distance change. These data constitute a complete set of obstacle distance data.

[0090] When dividing the space within a five-meter radius around the robot into three concentric circular regions based on obstacle distance data, the concentric circle region division algorithm establishes three circular regions with different radii centered on the robot's center point. The region division is based on distance threshold settings and safety response requirements to determine boundary parameters. The algorithm first sets the boundary of the outermost warning zone, with an outer boundary radius of five meters and an inner boundary radius of two meters, forming a ring-shaped warning zone. This zone is used for early warning and movement speed adjustment; when an obstacle enters the warning zone, it triggers movement parameter adjustments without affecting the main operational flow. The outer boundary of the middle layer protection zone coincides with the inner boundary of the warning zone, with a radius of two meters, and the inner boundary radius is set to one meter, forming a medium-distance protection zone. This zone is used for movement mode switching and safety strategy upgrades; when an obstacle enters the protection zone, the robot switches to a cautious operation mode and enhances safety monitoring. The innermost emergency zone is a circular area with a radius of one meter. This is the highest level of safety protection zone; any obstacle entering the emergency zone triggers immediate stop and emergency protection measures. The region segmentation algorithm determines the safe zone to which each obstacle belongs based on obstacle distance data, determines the threat level of the obstacle by distance comparison, and generates initial region range data containing region identifiers and threat levels. This data lays the foundation for subsequent security decisions and action selection.

[0091] When processing the initial region range using the temperature characteristics of people detected by the thermal imaging sensor for personnel recognition algorithms, the algorithm distinguishes people from other types of obstacles based on a dual discrimination mechanism of human body temperature and morphological features. The thermal imaging sensor acquires temperature distribution images of various objects in the environment by detecting infrared radiation. The normal human body temperature ranges from 36 to 37 degrees Celsius, showing a significant difference from the ambient temperature and the temperatures of other objects. The temperature feature extraction algorithm first performs temperature threshold segmentation on the thermal imaging image, setting upper and lower limits for human body temperature, and filtering out pixel regions whose temperature range matches human characteristics. Then, connectivity analysis and morphological processing are performed on these regions to remove noise points and small interference areas. The morphological feature recognition algorithm analyzes the geometric shape features of regions with acceptable temperatures, including parameters such as area size, aspect ratio, and contour features. In thermal imaging images, the human body typically presents an elliptical or rectangular outline, with an area size within a specific range and an aspect ratio consistent with human proportions. The algorithm determines whether the detection area is a human target by combining the discrimination results of temperature and morphological features. If the human target is successfully identified, its position coordinates in the thermal imaging image are recorded. Then, it is converted into actual position information in a three-dimensional spatial coordinate system to generate human position distribution information that includes the number of people, position coordinates and movement status.

[0092] When the personnel location distribution information is input into the dynamic adjustment algorithm for safety boundary correction, the algorithm adaptively adjusts the initial area range based on factors such as personnel position, movement direction, and movement speed. The adjustment principle is based on the uncertainty of personnel behavior and the forward-looking requirements of safety protection. The boundary correction algorithm first analyzes the spatial distribution characteristics in the personnel location distribution information. When a dense distribution of personnel is detected in a certain direction, the protection range of each safety zone in that direction is increased accordingly. The calculation of the expanded range is based on the product of personnel density and average movement speed. The movement direction analysis algorithm calculates the movement direction vector based on the historical position changes of personnel. When a personnel's movement direction points towards the robot, the boundary range of the safety zone is expanded in that direction. The expansion magnitude is proportional to the movement speed; the faster the movement speed, the greater the boundary expansion. The safety margin calculation considers the randomness of personnel behavior and the limitations of reaction time, adding a dynamic safety margin to the standard area boundary. The size of the margin is determined comprehensively based on factors such as the number of personnel, movement speed, and environmental complexity. The boundary correction process also considers the robot's current operating status and motion parameters. When the robot performs high-precision installation tasks, a more conservative boundary setting is adopted, while when performing general movement tasks, a relatively relaxed boundary parameter is adopted. Finally, dynamically adjusted safety zone boundaries for warning zones, protection zones, and emergency zones are generated. These boundary data reflect the current safety status and protection requirements of the environment in real time.

[0093] The intelligent control method of the electromechanical module intelligent installation robot in the embodiments of this application has been described above. The intelligent control system of the electromechanical module intelligent installation robot in the embodiments of this application is described below. Please refer to [link / reference]. Figure 2 One embodiment of the intelligent control system for the electromechanical module intelligent installation robot in this application includes:

[0094] The fusion module is used to perform multi-dimensional fusion processing on displacement sensor data, angle sensor data and visual recognition data through Kalman filtering algorithm to obtain a standardized spatial state vector.

[0095] The planning module is used to perform three-dimensional path planning processing on the installation location information in the BIM model based on the standardized spatial state vector to obtain a path instruction set;

[0096] The correction module is used to perform precise positioning correction processing on the path instruction set through a visual servoing algorithm to obtain corrected motion instructions;

[0097] The coordination module is used to perform synchronous coordination processing on the motion parameters of multiple actuators according to the modified motion command, so as to obtain the synchronous control command;

[0098] The evaluation module is used to combine the synchronous control command with the lidar scanning data to perform a safety obstacle avoidance evaluation and obtain a safety execution command.

[0099] above Figure 2 The intelligent control system of the electromechanical module intelligent installation robot in this embodiment of the invention is described in detail from the perspective of modular functional entities. The intelligent control device of the electromechanical module intelligent installation robot in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0100] Reference Figure 3 This invention also provides an intelligent control device for an intelligent electromechanical module installation robot. This intelligent control device can be a server, and its internal structure can be as follows: Figure 3As shown, the intelligent control device of this electromechanical modular intelligent installation robot includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor, designed as a computer, provides computing and control capabilities. The memory of the intelligent control device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the intelligent control device stores the data corresponding to this embodiment. The network interface of the intelligent control device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0101] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the intelligent control device of the electromechanical module intelligent installation robot to which the present invention is applied.

[0102] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the intelligent control method for the electromechanical module intelligent installation robot.

[0103] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0104] 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 medium. Based on this understanding, the technical solution of the present invention, 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 storage medium and includes several instructions to cause an intelligent control device (which may be a personal computer, server, or network device, etc.) of an electromechanical module intelligent installation robot to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0105] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent control method of an electromechanical module intelligent installation robot, characterized by, The method comprises: S1 step, multi-dimensional fusion processing of displacement sensor data, angle sensor data and visual recognition data by Kalman filtering algorithm to obtain a standardized space state vector; S2 step, according to the standardized space state vector, the installation position information in the BIM model is processed by three-dimensional path planning to obtain a path instruction set; S3 step, the path instruction set is corrected by visual servo algorithm to obtain a corrected motion instruction, including: the position coordinates in the standardized space state vector and the electromechanical module installation position information in the BIM model are processed by coordinate system conversion to obtain a target point set; based on the target point set, the three-dimensional grid map is processed by A* algorithm path search to obtain an initial path sequence; according to the environmental point cloud data obtained by laser radar scanning, the initial path sequence is processed by DBSCAN clustering algorithm obstacle identification to obtain dynamic obstacle boundary information; the dynamic obstacle boundary information is input into the path optimization objective function for path length, smoothness and time cost comprehensive optimization to obtain the path instruction set; S4 step, according to the corrected motion instruction, the motion parameters of multiple actuators are synchronized and coordinated to obtain a synchronous control instruction, including: setting the hydraulic lifting mechanism parameter in the corrected motion instruction as the main shaft control reference to obtain the main shaft motion reference signal; according to the main shaft motion reference signal, the motion parameters of X-axis motor, Y-axis motor and Z-axis fine adjustment mechanism are processed by time reference synchronization to obtain slave shaft synchronization parameters; based on the slave shaft synchronization parameters, a multi-actuator dynamics coupling model is established to compensate the interference between axes to obtain cross-coupling control quantity; the cross-coupling control quantity is superimposed with the feedforward control term to obtain the synchronous control instruction; S5 step, the synchronous control instruction is combined with laser radar scanning data for safety obstacle avoidance evaluation to obtain a safe execution instruction. 2.The intelligent control method of the electro-mechanical module intelligent installation robot according to claim 1, characterized in that, The S1 step comprises: The X, Y and Z three-axis coordinate data collected by the displacement sensor are processed by timestamp correction to obtain time-aligned position data; The pitch angle, yaw angle and roll angle data collected by the angle sensor are input into the noise filter for denoising to obtain filtered angle data; Based on the binocular stereo image obtained by the visual recognition system, depth calculation processing is performed to obtain target depth information; The time-aligned position data, the filtered angle data and the target depth information are input into the Kalman filter for state estimation to obtain a standardized space state vector. 3.The intelligent control method of the electromechanical module intelligent installation robot according to claim 1, characterized in that, The S3 step comprises: The position coordinates in the path instruction set are matched with the electromechanical module feature template library by normalized cross-correlation algorithm to obtain a visual positioning result; According to the visual positioning result and the expected position in the path instruction set, a position deviation is calculated to obtain a visual error vector; The visual error vector is input into the PID controller for proportional, integral and differential control to obtain a position correction amount; Based on the position correction amount, the motion parameters in the path instruction set are adjusted by coordinate transformation matrix to obtain a corrected motion instruction. 4.The intelligent control method of the electromechanical module intelligent installation robot according to claim 1, wherein, The S5 step comprises: The synchronization control instruction is processed by Kalman prediction algorithm with laser radar scanning data to obtain dynamic obstacle motion trajectory prediction information; The dynamic obstacle motion trajectory prediction information is used to divide the surrounding environment of the robot into layered protection areas to obtain the safety area boundaries of the warning area, the protection area and the emergency area; Based on the safety area boundaries, the risk assessment quantification model is calculated for the motion speed and acceleration parameters in the synchronization control instruction to obtain a safety risk coefficient; The safety risk coefficient is input into the safety decision logic for protection action selection processing to obtain a safety execution instruction. 5.The intelligent control method of the electro-mechanical module intelligent mounting robot according to claim 4, characterized in that, The dynamic obstacle motion trajectory prediction information is used to divide the surrounding environment of the robot into layered protection areas to obtain the safety area boundaries of the warning area, the protection area and the emergency area, comprising: The position coordinates in the dynamic obstacle motion trajectory prediction information are processed by Euclidean distance calculation from the robot center point to obtain obstacle distance data; The space within 5 meters around the robot is divided into three concentric circle areas according to the obstacle distance data to obtain the initial area range of the 2-5 meter warning area, the 1-2 meter protection area and the 0-1 meter emergency area; Based on the initial area range, personnel recognition algorithm processing is performed on the personnel temperature features detected by the thermal imaging sensor to obtain personnel position distribution information; The personnel position distribution information is input into the safety boundary dynamic adjustment algorithm for region boundary correction processing to obtain the safety area boundaries of the warning area, the protection area and the emergency area.

6. An intelligent control system of an electromechanical module intelligent installation robot, characterized in that, An intelligent control method for an electromechanical module intelligent installation robot as claimed in any one of claims 1-5, the intelligent control system of the electromechanical module intelligent installation robot comprising: A fusion module for performing multi-dimensional fusion processing on displacement sensor data, angle sensor data and visual recognition data by Kalman filtering algorithm to obtain a standardized space state vector; A planning module for performing three-dimensional path planning processing on installation position information in the BIM model according to the standardized space state vector to obtain a path instruction set; A correction module for performing precise positioning correction processing on the path instruction set by visual servo algorithm to obtain a corrected motion instruction; A coordination module for performing synchronization coordination processing on multi-actuator motion parameters according to the corrected motion instruction to obtain a synchronization control instruction; An evaluation module for performing safety obstacle avoidance evaluation processing on the synchronization control instruction in combination with laser radar scanning data to obtain a safety execution instruction.

7. An intelligent control device of an electromechanical module intelligent installation robot, characterized in that, A computer program is stored in the memory and can be run on the processor, and the processor implements the intelligent control method of the electromechanical module intelligent installation robot of any one of claims 1-5 when executing the computer program.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program makes the processor execute the intelligent control method of the electromechanical module intelligent installation robot of any one of claims 1-5 when the processor runs the computer program.

Images