Masonry robot intelligent masonry analysis method

Abstract

The application discloses a masonry robot intelligent masonry analysis method and relates to the field of masonry robot data analysis.The application comprises the following steps: collecting point cloud data and BIM models of a construction site, generating a construction 3D model and a global coordinate system registration result through a point cloud registration algorithm, performing parameterized modeling on an arc-shaped wall body, processing input parameterized wall body models, standard brick sizes and construction constraints, outputting an optimal standard brick arrangement sequence, mortar joint parameters and angles of each brick, generating a mechanical arm collision-free motion trajectory and joint angle sequence based on an improved RRT* algorithm, combining brick and wall surface contact force data collected by a real-time force sensor and visual feedback, calculating an adjusted final placement pose based on an impedance control algorithm, and performing placement quality evaluation.The application realizes accurate force control placement of bricks through impedance control and ensures mortar joint quality.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of masonry robot data analysis, in particular to a masonry robot intelligent masonry analysis method. BACKGROUND

[0002] In the prior art, the masonry robot is mainly used for the masonry of straight walls, and there are the following technical difficulties for the masonry of arc-shaped walls: (1) accurate geometric modeling and brick arrangement calculation are required for arc-shaped walls; (2) accurate angle adjustment and mortar joint control are required for standard bricks in arc-shaped masonry; (3) motion planning and obstacle avoidance of the mechanical arm under complex arc-shaped trajectories; (4) contact force control and precision guarantee when placing the bricks. There is no technical solution that combines point cloud registration, parameterized modeling, multi-objective optimization arrangement, intelligent motion planning and force control placement organically. SUMMARY

[0003] The present application provides a masonry robot intelligent masonry analysis method, which solves the problems of the prior art. The present application generates a technical solution for masonry of arc-shaped walls based on standard bricks, covering the selection of multiple group masonry forms of wedge-shaped bricks, standard bricks, and a mixture of wedge-shaped bricks and standard bricks. The present application adopts a closed-loop architecture of visual perception-intelligent planning-force control execution to ensure the construction precision and quality of arc-shaped walls.

[0004] In a first aspect, the present application provides a masonry robot intelligent masonry analysis method, comprising the following steps:

[0005] Step one, collect point cloud data and BIM model of the construction site, generate a construction 3D model and a global coordinate system registration result through a point cloud registration algorithm;

[0006] Step two, parameterize the arc-shaped wall to obtain a parameterized wall model containing the geometric parameters of the arc-shaped wall and the pose information of the bricks; wherein the pose information of the bricks includes the position and orientation of the bricks, and the geometric parameters of the arc-shaped wall include the center coordinates, the radius, the central angle and the center line length;

[0007] Step three, based on a standard brick arrangement optimization algorithm, process the input parameterized wall model, standard brick size and construction constraints, and output the optimal standard brick arrangement sequence, mortar joint parameters and angle of each brick;

[0008] Step four, based on the improved RRT* algorithm, generate a collision-free motion trajectory of the mechanical arm and a sequence of joint angles through the optimal standard brick arrangement sequence, the current state of the mechanical arm and the environmental obstacle information;

[0009] Step five, obtain the target placement pose through the mortar parameters and angles of each brick, combine the real-time force sensor collected brick and wall contact force data and visual feedback, calculate the adjusted final placement pose based on impedance control algorithm, and perform placement quality evaluation.

[0010] Further, the point cloud data and BIM model of the construction site are collected, and a construction 3D model and a global coordinate system registration result are generated through a point cloud registration algorithm, including:

[0011] The construction site point cloud is scanned, and the theoretical point cloud in the BIM model is extracted.

[0012] Step A1, construct an initialization transformation matrix for point cloud registration, which is used to initially align the construction site point cloud and the theoretical point cloud;

[0013] Step A2, obtain the correspondence between the construction site point cloud points and the theoretical point cloud points through nearest neighbor search; calculate the rigid body transformation, apply the rigid body transformation to the construction site point cloud, and obtain the transformed point cloud, the rigid body transformation includes a rotation matrix R and a translation vector T;

[0014] Step A3, loop step A2 while checking convergence, and determine convergence when the deviation between two adjacent rigid body transformations is less than the convergence tolerance; after convergence, determine the current rigid body transformation as the optimal rigid body transformation and determine the correspondence between the construction site point cloud and the theoretical point cloud;

[0015] Wherein, the deviation between two adjacent rigid body transformations is the difference between the transformation matrices of two iterations, which is calculated by the difference norm of the transformation matrix parameters or the change of the average distance of the point cloud.

[0016] Further, the arc-shaped wall is parameterized modeled to obtain a parameterized wall model containing the geometric parameters of the arc-shaped wall and the brick pose information, specifically including:

[0017] Step B1, input the arc-shaped wall data in the BIM model and the standard brick size;

[0018] Step B2, calculate the center coordinates (x0, y0, z0) and the radius R of the arc by the least squares method to obtain the point set on the center line of the arc-shaped wall;

[0019] Step B3, calculate the central angle θ corresponding to the entire arc-shaped wall body by the center and the point set on the center line of the arc-shaped wall;

[0020] Step B4, calculate the total length L = R x θ of the center line of the entire arc-shaped wall by the central angle θ and the radius R;

[0021] Step B5, calculate the number of standard bricks N = ⌈L / (l+δ)⌉ by the standard brick length l, width w, height h and mortar parameter δ.

[0022] Step B6, generating normal direction field, containing unit normal vector at each standard brick placement point on the center line of the arc-shaped wall, i.e. the radial direction from the center to the placement point;

[0023] Step B7, calculating the yaw angle ψ of the standard brick through the tangent direction of the current point, and then calculating the three-dimensional coordinates of the standard brick through the arc-shaped wall thickness and the standard brick height;

[0024] Step B8, finally outputting the parameterized wall model containing the center of the arc-shaped wall, the radius, the unit normal vector at each brick placement point on the center line of the arc-shaped wall, and the calculated 6-DOF pose matrix of each brick.

[0025] Further, the standard brick arrangement optimization algorithm based on the input parameterized wall model, standard brick size and construction constraints, outputs the optimal standard brick arrangement sequence, the mortar joint parameters and angles of each brick, specifically including:

[0026] Optimizing the standard brick arrangement through the improved adaptive genetic algorithm, loading the multi-objective fitness function evaluation, the multi-objective fitness function including:

[0027] Mortar uniformity objective function: f1 = 1 - σ(δ) / δ avg , where σ(δ) is the standard deviation of mortar width, δ avg is the average mortar width; the closer the function value is to 1, the better the mortar uniformity.

[0028] Structural stability objective function: f2 = Σ(A contact ) / A total , where A contact is the effective contact area between bricks, A total is the theoretical maximum contact area; the specific calculation method is , where, is the effective contact area between the i-th brick and the i+1-th brick, and the contact area is calculated by the overlapping area between the bricks.

[0029] Construction efficiency objective function: f3 = 1 - t total / t max , where t total is the estimated total construction time, including robot movement time and brick placement time, t max is the maximum allowable construction time;

[0030] Arc fitting degree objective function: f4 = 1 - Σ|d i | / (N×ε), where d iε represents the deviation between the actual position of the i-th brick and the theoretical arc, i.e., the distance between the center point of the i-th brick and the center line of the theoretical curved wall, and ε is the maximum allowable deviation.

[0031] The weighted fitness F = w1f1 + w2f2 + w3f3 + w4f4, where w1+w2+w3+w4=1;

[0032] The optimization process of the adaptive genetic algorithm:

[0033] Initialize the population, setting the population size to 50-200;

[0034] Assess population fitness and calculate the weighted overall fitness score for each individual;

[0035] Determine if the termination condition is met. The termination condition is reaching the maximum number of iterations, which is 100-500.

[0036] If the conditions are met, the optimal individual is output as the standard brick layout scheme;

[0037] If the conditions are not met, a roulette wheel selection method is used to select the best individuals, which are defined as the top 20% of individuals in terms of fitness.

[0038] Perform a crossover operation with a crossover probability of 0.7-0.9 to exchange parent genes and generate new individuals;

[0039] Mutation operation, with a mutation probability of 0.01-0.1, randomly alters the genes of an individual;

[0040] Reassess population fitness, iterating repeatedly.

[0041] The individual represents the solution, encoded as an integer sequence, representing the arrangement and rotation state of the standard bricks.

[0042] Furthermore, the process of generating the collision-free motion trajectory of the robotic arm and the sequence of joint angles based on the improved RRT* algorithm, using the optimal standard brick arrangement sequence, the current state of the robotic arm, and environmental obstacle information, specifically includes:

[0043] Convert the optimal standard brick layout sequence into the target pose sequence Q. goal = {q 1 ,q 2 ,...,qN};where N is the number of joints in the robotic arm, and q1 to qN represent the angle values ​​of the 1st to Nth joints, respectively;

[0044] The current state of the robotic arm is represented as the initial pose q. start ;

[0045] Environmental obstacle information is represented as a set of obstacle point clouds O. obs ;

[0046] Trajectory planning based on improved RRT* algorithm:

[0047] Initialize RRT* tree, root node is the initial pose q start ;

[0048] Repeat the following steps until a path is found or the maximum number of nodes is reached:

[0049] Randomly sample q in configuration space rand ; for randomly generated points in configuration space, used to guide the search tree to expand to unexplored areas;

[0050] Find the node q in the tree closest to q rand ; near ;

[0051] Expand the step size Δq from q near to q rand direction, get new node q new ;

[0052] Collision detection, check if the path (q near , q new ) intersects O obs ;

[0053] If the collision detection passes, add q new to the RRT* tree and perform reconnection optimization;

[0054] Extract the path from q start to Q goal from the RRT* tree and perform B-spline curve smoothing;

[0055] Calculate the joint angle sequence corresponding to each path point through inverse kinematics of the robot arm;

[0056] Output the smoothed collision-free motion trajectory and joint angle sequence.

[0057] Further, the target placement pose is obtained through the mortar parameters and angles of each brick, combined with real-time force sensor collected brick and wall contact force data and visual feedback, based on impedance control algorithm to calculate the adjusted final placement pose, and perform placement quality assessment, specifically including:

[0058] Set impedance model parameters, including expected inertia M d , expected damping D d , expected stiffness K d ;

[0059] The impedance control model is expressed as:

[0060] ;

[0061] in:

[0062] M d ,D d ,K d These are 6×6 diagonal matrices representing the expected inertia, damping, and stiffness, respectively.

[0063] For actual pose, velocity, and acceleration; Given the desired pose, velocity, and acceleration;

[0064] F ext The external contact force measured by the sensor;

[0065] F d The desired contact force.

[0066] The pose adjustment is calculated by numerical integration: ΔX = K d -1 ΔF;

[0067] The expected vertical force F{d,z} is used for mortar joint compaction, and the expected horizontal forces F{d,x} and F{d,y} are used for contact detection.

[0068] Acquire the actual contact force F collected by the force sensor s ;

[0069] Calculation of force error ΔF = F d - F s ;

[0070] Calculate the pose adjustment amount based on the impedance control law;

[0071] The final placement pose X after adjustment is calculated based on the pose adjustment amount. final = X target + ΔX;X target For the target location;

[0072] The quality assessment indicators for placement include:

[0073] Mortar joint compaction degree: η = F{s,z} / F{d,z};

[0074] Position accuracy: ε p = ||X final - X target ||;

[0075] Attitude accuracy: ε o = 2×arccos(|qfinal ·q targe t|);

[0076] where q final and q target are the quaternion representation of the final pose and target pose respectively.

[0077] Further, when the mechanical arm is grabbing, the standard brick image obtained by the camera is input into the CNN brick identification model after preprocessing, and the brick size classification is carried out, which is used for sorting the complete standard bricks.

[0078] The training process of the CNN brick identification model is as follows: a brick image data set is constructed, which contains standard brick images of different sizes and angles; the brick image is input into the CNN for convolution operation, and the calculation formula is as follows: wherein: i is an index variable, representing the index of the input feature map of the current layer (the l layer), and is the output of the jth feature map of the lth layer, M j is the set of input feature maps connected to the jth output feature map, is the convolution kernel weight, is the bias, f is the ReLU activation function; down-sampling is carried out through the maximum pooling layer, and the feature map is divided into 2x2 regions, and the maximum value of each region is taken; the feature map after the convolution and pooling operation of multiple layers is flattened and input into the fully connected layer; the fully connected layer adopts the ReLU activation function, and the output layer adopts the softmax function, to obtain the brick size classification result; the on-site parameters of the brick stacking area and the brick size classification result are input into the scheme generation model to obtain the brick selection scheme, and the starting position of each standard brick trajectory is determined.

[0079] The technical innovation of the application is that a variety of advanced technologies are integrated into the arc-shaped wall building robot system, forming a complete technical closed loop. The specific technical contributions include:

[0080] A parameterized modeling method for arc-shaped walls is proposed, which can accurately calculate the pose of each brick;

[0081] A genetic algorithm of multi-objective fitness function is designed, which optimizes the uniformity of mortar joint, structural stability, construction efficiency and arc fitting degree;

[0082] The brick arrangement sequence is closely combined with the motion planning of the mechanical arm, and efficient collision-free trajectory generation is realized;

[0083] The precise force control placement of the bricks is realized through impedance control, and the quality of the mortar joint is ensured. Attached Figure Description

[0084] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and constitute a part of this invention, are not intended to limit the scope of the invention. In the drawings:

[0085] Figure 1 A flowchart of an intelligent masonry analysis method for a masonry robot, provided as an exemplary embodiment of the present invention.

[0086] Figure 2 This is a flowchart of point cloud registration for the present invention.

[0087] Figure 3 This is a flowchart illustrating the parametric modeling of an arc-shaped wall according to the present invention. Detailed Implementation

[0088] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention.

[0089] The intelligent masonry analysis method for masonry robots provided by this invention aims to solve the above-mentioned technical problems in the prior art.

[0090] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0091] The intelligent masonry analysis method for masonry robots provided by this invention aims to solve the above-mentioned technical problems in the prior art.

[0092] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0093] like Figure 1 As shown, the intelligent masonry analysis method for masonry robots includes the following steps:

[0094] Step 1: Collect point cloud data and BIM model from the construction site, and generate a 3D construction model and global coordinate system registration results using a point cloud registration algorithm; for example... Figure 2The scanning obtains a construction site point cloud, and a theoretical point cloud in a BIM model is extracted;

[0095] Step A1, constructing an initialization transformation matrix of point cloud registration, for initial alignment of the construction site point cloud and the theoretical point cloud;

[0096] Step A2, obtaining a correspondence between points of the construction site point cloud and points of the theoretical point cloud through nearest neighbor search; calculating a rigid body transformation, applying the rigid body transformation to the construction site point cloud to obtain a transformed point cloud, the rigid body transformation including a rotation matrix R and a translation vector T;

[0097] Step A3, checking convergence while looping step A2, and judging convergence by a deviation between two adjacent rigid body transformations being less than a convergence tolerance; after convergence, determining a current rigid body transformation as an optimal rigid body transformation and determining a correspondence between the construction site point cloud and the theoretical point cloud;

[0098] The deviation between the two adjacent rigid body transformations is a difference between transformation matrices of two iterations, and the difference is calculated by a difference norm of transformation matrix parameters or a change amount of an average distance of point clouds.

[0099] Step two, parameterized modeling of the arc-shaped wall, to obtain a parameterized wall model containing geometric parameters of the arc-shaped wall and brick pose information; wherein the brick pose information includes a position and an orientation of the brick, and the geometric parameters of the arc-shaped wall include a center coordinate, a radius, a central angle and a center line length; as shown in the figure, specifically including: Figure 3

[0100] Step B1, inputting arc-shaped wall data in a BIM model and a standard brick size;

[0101] Step B2, calculating a point set on a center line of the arc-shaped wall by a least square method to obtain a center coordinate (x0, y0, z0) of the arc and a radius R;

[0102] Step B3, calculating a central angle θ corresponding to the entire arc-shaped wall by the center and the point set on the center line of the arc-shaped wall;

[0103] Step B4, calculating a total length L of the center line of the entire arc-shaped wall by the central angle θ and the radius R, L = R x θ;

[0104] Step B5, calculating a number N of standard bricks by a standard brick length l, a width w, a height h and a mortar joint parameter δ, N = ⌈L / (l+δ)⌉;

[0105] Step B6, generating a normal direction field containing a unit normal vector at each standard brick placement point on the center line of the arc-shaped wall, i.e., a radial direction from the center to the placement point;

[0106] ​Step B7, calculate the yaw angle ψ of the standard brick through the tangent direction of the current point, and then calculate the three-dimensional coordinates of the standard brick through the arc wall thickness and the standard brick height;

[0107] Step B8, finally output the parameterized wall model containing the center of the arc wall, the radius, the unit normal vector at each brick placement point on the center line of the arc wall, and the calculated 6-DOF pose matrix of each brick.

[0108] Step three, based on the standard brick arrangement optimization algorithm, process the input parameterized wall model, standard brick size and construction constraints, and output the optimal standard brick arrangement sequence, mortar joint parameters and angles of each brick;

[0109] Step four, based on the improved RRT* algorithm, generate the collision-free motion trajectory of the mechanical arm and the joint angle sequence through the optimal standard brick arrangement sequence, the current state of the mechanical arm, and the environmental obstacle information;

[0110] Step five, obtain the target placement pose through the mortar joint parameters and angles of each brick, combine the brick and wall contact force data collected by the real-time force sensor and the visual feedback, calculate the adjusted final placement pose based on the impedance control algorithm, and perform placement quality evaluation.

[0111] In the above steps, the present application provides Example 1: to build an arc-shaped flower bed with a radius of 2 meters and a height of 1 meter, using standard bricks (240mm x 115mm x 53mm). The target mortar joint width is 10mm, with an allowable deviation of ±2mm.

[0112] S1: Environment scanning

[0113] Input: Laser scanner acquires basic point cloud (about 100,000 points), BIM model provides theoretical flower bed base circle (point cloud).

[0114] ICP parameters: maximum number of iterations 50, tolerance 1e-6, distance threshold 0.05m. The tolerance measures the difference between the transformation matrix calculated in this iteration and the transformation matrix of the last iteration; avoid unnecessary iterations, stop once the result is stable, save time.

[0115] Output: Transformation matrix T, after registration, it is found that the base has a 5mm deviation, which is compensated in the subsequent steps.

[0116] Data transition: Transformation matrix T is passed to S2 to convert the theoretical model to the field coordinate system;

[0117] S2: Parameterized modeling

[0118] Input: BIM center line point set (100 points), brick size (0.24, 0.115, 0.053)m.

[0119] Curve fitting: Least squares fitting of a circular arc, resulting in a center (0,0,0) and a radius of 2.0 m.

[0120] Normal field generation: Calculate the normal vector for each point (pointing outwards from the center).

[0121] Output: List of theoretical poses for the bricks (50 bricks per skin, 19 skins in total), including position (X,Y,Z) and Euler angles (Roll,Pitch,Yaw) for each brick.

[0122] Data transition: The list of poses is passed to S3 for optimization.

[0123] S3: Brick arrangement optimization

[0124] Input: List of theoretical poses, brick dimensions, constraints (grout 5-15 mm, stagger 1 / 4 brick length).

[0125] Genetic algorithm parameters: Population size 100, iterations 200, crossover rate 0.8, mutation rate 0.1, elite retention 5. Crossover rate: The probability of two parent individuals being crossed to produce offspring, mutation rate: The probability of a gene (brick order or rotation state) being randomly changed in an individual, elite retention: Ensures that the algorithm does not forget the best solution found, an important guarantee of convergence. Usually set to 1%~5% of the population size.

[0126] Fitness weights: Grout uniformity 0.3, stability 0.3, efficiency 0.2, arc fitting 0.2.

[0127] Output: Optimal arrangement sequence, grout width outer arc 10.2 mm, inner arc 9.8 mm, final pose for each brick.

[0128] Data transition: The list of poses after optimization is passed to S4.

[0129] S4: Motion trajectory generation

[0130] Input: Current pose of the robot arm (Home position), target pose of the first brick, obstacles (flower bed base, already laid bricks).

[0131] RRT* parameters: Maximum nodes 1000, step size 0.1 m, target bias 0.2, collision threshold 0.05 m. Collision threshold: Safety distance threshold for collision detection.

[0132] Output: Collision-free trajectory (100 intermediate points), planning time about 2 seconds.

[0133] Data transition: The trajectory is passed to the robot arm controller for execution.

[0134] S5: Force-controlled placement

[0135] Inputs: Target pose (from S3), force sensor data (real-time), current pose.

[0136] Impedance control parameters:

[0137] Desired inertia M_d = diag([1,1,1,0.1,0.1,0.1]), desired damping D_d = diag([10,10,10,1,1,1]), desired stiffness K_d = diag([100,100,100,10,10,10]). Where, the desired inertia simulates the mass of the end-effector, defines the inertial response of the system to force error; the desired damping simulates the damping coefficient of the system, equivalent to "viscosity", determines how the system consumes energy, affects the smoothness and stability of motion. The desired stiffness simulates the spring stiffness of the system, defines the proportional relationship between pose error and restoring force, determines the "softness" of the robot. Where, diag([a,b,c, d, e, f]) is a 6-dimensional diagonal matrix, corresponding to the desired dynamic characteristics of the robot end-effector in six degrees of freedom, the first three values ([1,1,1]): corresponding to the translational freedom, i.e. movement along the X, Y, Z axes, the last three values ([0.1,0.1,0.1]): corresponding to the rotational freedom, i.e. rotation around the X, Y, Z axes. For the desired inertia M d = diag([1, 1, 1,0.1, 0.1, 0.1]), [1, 1, 1] (translation): means that the robot in the invention behaves like an object with a mass of 1 kg when moving in the X, Y, Z direction. [0.1,0.1, 0.1] (rotation): means that the robot end in the invention has very small inertia (0.1 kg·m²) when rotating around the axes, as setting a small inertia can make the rotation adjustment more sensitive. d = diag([10, 10, 10, 1, 1, 1]), [10, 10, 10] (translation): a relatively high damping is set. This can ensure that the robot arm moves very smoothly and gently when contacting the brick or encountering resistance, and does not produce shaking or rebound, [1,1, 1] (rotation): the damping in the rotation direction is set to be small, allowing the rotation pose to be adjusted more quickly. For the desired stiffness K d= diag([100, 100, 100, 10, 10, 10]), [100, 100, 100] (translation): higher stiffness (unit: N / m) is set. This means high position accuracy: if an external force tries to push the robot arm away from the target position, it will generate a large force to resist and try to maintain the original position, ensuring the accuracy of placement. "Rigid" contact: for masonry work that requires compacting mortar joints, a certain stiffness is necessary to ensure that the bricks can be compacted and not retreat at the slightest touch.

[0138] Expected force: 5N on Z-axis (compaction), 2N on X / Y-axis (contact).

[0139] Output: adjusted pose (Z-axis lowered by 0.5mm), placement successful, mortar joint compactness 95%.

[0140] Data transition: placement quality assessment record for subsequent brick adjustment.

[0141] Example 2: Build an arc-shaped exhibition wall with a radius of 3.5 meters and a height of 2.5 meters using standard bricks (240mm x 115mm x 53mm). The target width of the mortar joint is 8mm with an allowable deviation of ±1mm.

[0142] S1: Environment scanning and registration;

[0143] Input: 3D laser scanner to obtain point cloud (about 250,000 points), BIM model to provide theoretical wall model;

[0144] ICP registration parameters: maximum number of iterations 100, tolerance 1e-7, distance threshold 0.03m;

[0145] Output: registration accuracy reaches 2mm, meeting construction requirements;

[0146] S2: Parametric modeling;

[0147] The center of the circle is (1.2, -0.8, 0.3)m and the radius is 3.5m obtained by least squares fitting;

[0148] The central angle of the circle is 85° and the arc length of the center line is 5.19m;

[0149] Generate a pose sequence containing 185 bricks, with an accuracy of 0.1° in yaw angle for each brick;

[0150] S3: Brick arrangement optimization;

[0151] Genetic algorithm parameters: population size 150, iteration 300 times, crossover rate 0.85, mutation rate 0.05;

[0152] Fitness weight: joint uniformity 0.35, stability 0.35, efficiency 0.15, arc fitting 0.15;

[0153] Output: optimal arrangement scheme, joint width deviation ±0.5mm, arc fitting error <2mm;

[0154] S4: motion trajectory planning;

[0155] RRT* parameters: maximum nodes 5000, step size 0.08m, target bias 0.3;

[0156] Planning time: average trajectory planning time for each brick 1.5 seconds;

[0157] Output: collision-free smooth trajectory, each joint motion is smooth and has no abrupt change;

[0158] S5: force control placement;

[0159] Impedance control parameters:

[0160] Md = diag([1.2, 1.2, 1.2, 0.15, 0.15, 0.15]);

[0161] Dd = diag([12, 12, 12, 1.5, 1.5, 1.5]);

[0162] Kd = diag([120, 120, 120, 15, 15, 15]);

[0163] Desired contact force: 8N (compaction) in Z-axis, 3N (contact) in X / Y-axis;

[0164] Quality evaluation: joint compaction degree 98%, position accuracy ±0.3mm, attitude accuracy ±0.2°.

[0165] In addition, each functional module in the embodiments of the present application can be integrated in one processing module, or each module can exist physically alone, or two or more modules can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of hardware plus software functional module.

[0166] It is also to be noticed that the term "comprising", "including", and "having" or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises, includes or has a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a", "has... a", "includes... a", or "including a", does not, without more constraints, foreclose the existence of additional identical elements other than the listed one.

[0167] The present application is only an embodiment of the present application, and is not intended to limit the present application. The present application can have various changes and modifications for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the scope of claims of the present application.

[0168] Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed here. The present application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such steps, compositions, ingredients, products, and procedures within the scope of the application. The specification and examples are to be regarded as illustrative only, and the true scope and spirit of the application are indicated by the following claims.

[0169] It should be understood that the application is not limited to the precise construction here described and illustrated and that various modifications and changes can be made without departing from the scope thereof. The only scope of the application is defined by the appended claims.

Claims

1. An intelligent masonry analysis method for masonry robots, characterized in that, The construction of curved walls includes the following steps: Step 1: Collect point cloud data and BIM model from the construction site, and generate a 3D construction model and global coordinate system registration results using the point cloud registration algorithm; Step 2: Perform parametric modeling on the curved wall to obtain a parametric wall model containing the geometric parameters of the curved wall and the pose information of the bricks; among which, the brick pose information includes the position and orientation of the bricks, and the geometric parameters of the curved wall include the center coordinates, radius, central angle and centerline length; Step 3: Based on the standard brick layout optimization algorithm, process the input parametric wall model, standard brick dimensions, and construction constraints to output the optimal standard brick layout sequence, mortar joint parameters, and angles for each brick, including: The standard brick arrangement is optimized using an improved adaptive genetic algorithm, and then evaluated using a multi-objective fitness function, which includes: Objective function for mortar joint uniformity: f1 = 1 - σ(δ) / δ avg Where σ(δ) is the standard deviation of the mortar joint width, δ avg This represents the average width of the mortar joint. Structural stability objective function: f2 = Σ(A contact ) / A total A contact A represents the effective contact area between the bricks. total This represents the theoretical maximum contact area. Construction efficiency objective function: f3 = 1 - t total / t max , where t total To estimate the total construction time, t max Maximum allowable construction time; Objective function for arc fit: f4 = 1 - Σ|d i | / (N×ε), where d i Let ε be the deviation between the actual position of the i-th brick and the theoretical arc, and let ε be the maximum allowable deviation. The weighted fitness F = w1f1 + w2f2 + w3f3 + w4f4, where w1+w2+w3+w4=1; The optimization process of the adaptive genetic algorithm: Initialize the population, setting the population size to 50-200; Assess population fitness and calculate the weighted overall fitness score for each individual; Determine if the termination condition is met. The termination condition is reaching the maximum number of iterations, which is 100-500. If the conditions are met, the optimal individual is output as the standard brick layout scheme; If the conditions are not met, a roulette wheel selection method is used to select the best individuals, which are defined as the top 20% of individuals in terms of fitness. Perform a crossover operation with a crossover probability of 0.7-0.9 to exchange parent genes and generate new individuals; Mutation operation, with a mutation probability of 0.01-0.1, randomly alters the genes of an individual; Reassess population fitness, iterating repeatedly. The individual representation solution is encoded as an integer sequence, representing the arrangement order and rotation state of the standard bricks; Step 4: Based on the optimal standard brick arrangement sequence, the current state of the robotic arm, and environmental obstacle information, generate the collision-free motion trajectory of the robotic arm and the sequence of angles of each joint using the improved RRT* algorithm; Step 5: Obtain the target placement posture by using the mortar joint parameters and angles of each brick. Combine the contact force data between the brick and the wall collected by the real-time force sensor with visual feedback, calculate the final placement posture after adjustment based on the impedance control algorithm, and evaluate the placement quality.

2. The intelligent masonry analysis method for masonry robots according to claim 1, characterized in that, The process of collecting point cloud data and BIM models from the construction site, and generating a 3D construction model and global coordinate system registration results using point cloud registration algorithms, includes: The point cloud of the construction site is obtained by scanning, and the theoretical point cloud is extracted from the BIM model; Step A1: Construct the initialization transformation matrix for point cloud registration, which is used to initially align the construction site point cloud and the theoretical point cloud; Step A2: Obtain the correspondence between the construction site point cloud points and the theoretical point cloud points through nearest neighbor search; calculate rigid body transformation, apply rigid body transformation to the construction site point cloud, and obtain the transformed point cloud. The rigid body transformation includes rotation matrix R and translation vector T. Step A3: While repeating step A2, check for convergence. Convergence is determined when the deviation between two adjacent rigid body transformations is less than the convergence tolerance. After convergence, determine that the current rigid body transformation is the optimal rigid body transformation and determine the correspondence between the point cloud at the construction site and the theoretical point cloud. The deviation between two adjacent rigid body transformations is the difference between the transformation matrices of the two iterations, which is calculated by the difference norm of the transformation matrix parameters or the change in the average distance of the point cloud.

3. The intelligent masonry analysis method for masonry robots according to claim 2, characterized in that, The parametric modeling of the curved wall, obtaining a parametric wall model containing the geometric parameters of the curved wall and the pose information of the bricks, specifically includes: Step B1: Input the curved wall data and standard brick dimensions from the BIM model; Step B2: Calculate the point set on the center line of the curved wall using the least squares method to obtain the center coordinates (x0, y0, z0) and radius R of the arc. Step B3: Calculate the central angle θ corresponding to the entire arc wall using the set of points on the center line of the arc wall and the center line of the circle. Step B4: Calculate the total length L = R×θ of the centerline of the entire curved wall using the central angle θ and the radius R; Step B5: Calculate the number of standard bricks N = [L / (l+δ)] using the standard brick length l, width w, height h, and mortar joint parameter δ; Step B6: Generate the normal direction field, which contains the unit normal vector at each standard brick placement point on the center line of the curved wall, i.e., the radial direction from the center of the circle to the placement point; Step B7: Calculate the yaw angle ψ of the standard brick using the tangent direction at the current point, and then calculate the three-dimensional coordinates of the standard brick using the thickness of the curved wall and the height of the standard brick. Step B8: The final output parameterized wall model includes the center and radius of the curved wall, the unit normal vector at each brick placement point on the center line of the curved wall, and the 6-DOF pose matrix calculated for each brick.

4. The intelligent masonry analysis method for masonry robots according to claim 3, characterized in that, The process involves generating a collision-free motion trajectory and joint angle sequences for the robotic arm based on an improved RRT* algorithm, using the optimal standard brick arrangement sequence, the current state of the robotic arm, and environmental obstacle information. Specifically, this includes: Convert the optimal standard brick layout sequence into the target pose sequence Q. goal = {q 1 ,q 2 ,...,qN};where N is the number of joints in the robotic arm, and q1 to qN represent the angle values ​​of the 1st to Nth joints, respectively; The current state of the robotic arm is represented as the initial pose q. start ; Environmental obstacle information is represented as a set of obstacle point clouds O. obs ; Trajectory planning based on the improved RRT* algorithm: Initialize the RRT* tree, with the root node being the initial pose q. start ; Repeat the following steps until a path is found or the maximum number of nodes is reached: Randomly sample q in the configuration space rand ; Find the distance q in the tree rand The nearest node q near ; From q near To q rand Expand the direction by step size Δq to obtain the new node q. new ; Perform collision detection and check the path (q) near ,q new Is it related to O? obs intersect; If collision detection is passed, then q new Add to the RRT* tree and perform reconnection optimization; Extract from q in the RRT* tree start To Q goal The path is smoothed using B-spline curves. The joint angle sequence corresponding to each path point is calculated using the inverse kinematics of the robotic arm. Output the smoothed, collision-free motion trajectory and the sequence of joint angles.

5. The intelligent masonry analysis method for masonry robots according to claim 4, characterized in that, The process involves obtaining the target placement posture by analyzing the mortar joint parameters and angles of each brick, combining this with real-time force sensor data on the contact force between the brick and the wall, and visual feedback. Based on an impedance control algorithm, the final adjusted placement posture is calculated, and the placement quality is evaluated. Specifically, this includes: Set the impedance model parameters, including the desired inertia M. d Expected damping D d Desired stiffness K d ; The desired contact force F is set according to the mortar joint parameters. d The vertical expected force F{d,z} is used for mortar joint compaction, and the horizontal expected forces F{d,x} and F{d,y} are used for contact detection. Acquire the actual contact force F collected by the force sensor s ; Calculation of force error ΔF = F d - F s ; Calculate the pose adjustment ΔX = K based on the impedance control law. d ⁻¹ΔF; The final placement pose X after adjustment is calculated based on the pose adjustment amount. final = X target + ΔX, X target For the target location; The quality assessment indicators for placement include: Mortar joint compaction degree: η = F{s,z} / F{d,z}; Position accuracy: ε p = ||X final - X target ||; Attitude accuracy: ε o = 2×arccos(|q final ·q targe t|); Where q final and q target These are the quaternion representations of the final attitude and the target attitude, respectively.

6. The intelligent masonry analysis method for masonry robots according to claim 5, characterized in that, It also includes using standard brick images captured by a camera during robotic arm grasping, and then inputting the preprocessed image data into a CNN brick recognition model to classify brick sizes for sorting complete standard bricks; The training process of the CNN brick recognition model is as follows: Construct a brick image dataset containing standard brick images of different sizes and angles; Input the brick images into the CNN for convolution operations, using the following formula: ;in: i : The index of the input feature map This is the output of the j-th feature map in the l-th layer. M j Let j be the set of input feature maps connected to the j-th output feature map. For convolution kernel weights, For bias, f The ReLU activation function is used; the feature map is downsampled through a max pooling layer, dividing it into 2×2 regions, and the maximum value of each region is taken; the feature map after multiple convolution and pooling operations is flattened and fed into a fully connected layer; the fully connected layer uses the ReLU activation function, and the output layer uses the softmax function to obtain the brick size classification result; the field parameters of the brick stacking area are obtained, and the field parameters are input into the scheme generation model to obtain the brick selection scheme and determine the starting position of each standard brick trajectory.

Images