A method for controlling dust during wall construction

Abstract

The present invention discloses a wall construction dust control method, which is used for adaptively controlling dust generated by construction, and includes the steps of: constructing a roughness detection model using a 2D image of a wall to be constructed and corresponding depth data, and detecting the wall roughness value using the roughness detection model; obtaining the dust suction negative pressure corresponding to each wall roughness value through multiple experiments; judging whether the dust suction gap is greater than a threshold value, and if greater than the threshold value, adjusting the inclination angle of the support mechanism so that the negative pressure vacuum cleaner fits the construction wall. The wall roughness value is detected by the detection model, and the dust suction negative pressure corresponding to each wall roughness value is obtained. The dust 3D point cloud data is obtained by the depth image algorithm, and the construction gap of the grinding module is obtained by using the point cloud data. The construction gap is adaptively adjusted by changing the elongation of the elongation mechanism, so as to realize adaptive control of construction dust and achieve the purpose of clean construction.

Description

Technical Field

[0001] The present invention relates to the technical field of indoor robot intelligent construction, and in particular to a dust control method and system for wall construction. Background Art

[0002] With the intelligentization of the construction industry, a variety of construction robots have emerged. Construction robots can replace manual labor to achieve safe and efficient construction. The interior wall treatment includes grinding the cement wall, removing burrs on the cement surface, and applying putty and putty layers on the cement wall to make the wall smoother. Grinding the putty layer wall and removing the joints of the putty layer make the putty layer smoother; spraying or brushing paint on the putty layer.

[0003] During the rough grinding of cement walls and the fine grinding of putty walls, dust will be produced. In order to control the dust, a negative pressure device is needed to collect the dust to avoid environmental pollution and to prevent the perception hardware of the wall processing robot from being contaminated by dust. The physical degree of cement walls and putty walls in different projects is different, and the amount of dust generated when grinding with the same roughness is different. How to enable construction robots to adaptively realize the grinding function on cement and putty walls, reduce dust in the construction environment, and achieve clean construction is an urgent problem to be solved in the current indoor construction process. Summary of the invention

[0004] Existing indoor construction robots generate a lot of dust when performing indoor construction, causing environmental pollution. However, current indoor construction robots are unable to perform adaptive dust collection to achieve clean construction.

[0005] In view of the above problems, a method for indoor construction dust control is proposed. The wall roughness value is detected by a detection model, and the dust suction negative pressure corresponding to each wall roughness value is obtained. The dust 3D point cloud data is obtained by a deep image algorithm. The point cloud data is used to obtain the construction gap of the grinding module. The construction gap is adaptively adjusted by changing the elongation of the elongation mechanism, thereby realizing adaptive control of construction dust and achieving the purpose of clean construction.

[0006] In a first aspect, a wall construction dust control method is provided, which is used to adaptively control dust generated during construction, comprising:

[0007] Step 100: construct a roughness detection model using the 2D image of the wall to be constructed and the corresponding depth data, and use the roughness detection model to detect the wall roughness value;

[0008] Step 200, obtaining the vacuum negative pressure corresponding to each wall roughness value through multiple experiments;

[0009] Step 300, determine whether the vacuum gap is greater than a threshold value, and if so, adjust the inclination angle of the support mechanism so that the negative pressure vacuum cleaner fits the construction wall surface.

[0010] In conjunction with the wall construction dust control method of the present invention, in a first possible implementation manner, step 100 includes:

[0011] Step 110: Acquire a minimum sample set using the 2D image, the depth data, and the coarse value corresponding to the depth data;

[0012] Step 120: Determine a training set and a test set respectively according to the minimum sample set;

[0013] Step 130: Use the training set to train a neural network deep learning model to obtain a roughness detection model;

[0014] Step 140: Detect the 2D image and depth data information of the construction wall surface and input them into the roughness detection model to obtain the roughness of the construction wall surface.

[0015] In combination with the first possible implementation manner of the present invention, in a second possible implementation manner, step 140 includes:

[0016] Preferably, the step 140 includes the sub-steps of:

[0017] Step 141: Process the 2D image using the first deep learning structure in the roughness detection model to obtain first depth data;

[0018] Step 142: Process the depth data information using the second deep learning structure in the roughness detection model to obtain second depth data;

[0019] Step 143: In the fully connected layer of the roughness detection model, the first depth data and the second depth data are merged and input into the fully connected layer again;

[0020] Step 144: classify and judge the roughness value of each point cloud of the construction wall surface image, and output the roughness value.

[0021] In combination with the second possible implementation manner of the present invention, in a third possible implementation manner, step 142 includes:

[0022] Step 1421, aligning the point cloud data set using the transformation matrix T-Net;

[0023] Step 1422, extracting the features of the point cloud data set through multiple MLPs and aligning them again using the transformation matrix T-Net;

[0024] Step 1423: perform a maxpooling operation on each dimension of the feature to obtain the final global feature;

[0025] Step 1424: Concatenate the global features with the local features of each point cloud and obtain the classification result of each data point through MLP.

[0026] In combination with the third possible implementation manner of the present invention, in a fourth possible implementation manner, step 200 includes:

[0027] Step 210: Obtain a discrete array of the wall surface roughness value and the vacuum negative pressure through experiments;

[0028] Step 220: Fit the discrete array to obtain the corresponding relationship function between the wall surface roughness value and the vacuum negative pressure:

[0029] P=R*A+B (1)

[0030] Among them, P is the vacuum pressure, R is the wall roughness value, A is the wall roughness parameter; and B is the basic vacuum pressure.

[0031] In combination with the fourth possible implementation manner of the present invention, in a fifth possible implementation manner, step 300 includes:

[0032] Step 310: Obtain construction wall dust point cloud information using an image depth algorithm;

[0033] Step 320: using the dust point cloud information to obtain a construction gap between the negative pressure vacuum cleaner and the wall;

[0034] Step 330: If the construction gap is greater than a threshold, the tilt angle of the negative pressure vacuum cleaner is adjusted.

[0035] In combination with the fifth possible implementation manner of the present invention, in a sixth possible implementation manner, step 310 includes:

[0036] Step 311, converting the depth image of the construction area into point cloud data information through coordinate transformation;

[0037] Step 312: partition the point cloud data information into corresponding partitions.

[0038] In combination with the sixth possible implementation manner of the present invention, in a seventh possible implementation manner, step 320 includes:

[0039] Step 321, obtaining 3D point cloud data obtained by shooting two frames of the same construction space;

[0040] Step 322: Calculate the number of 3D points where data points appear in each area using the 3D point cloud data;

[0041] Step 323: If the number of 3D points exceeds a threshold, the area is determined to be a construction gap.

[0042] In combination with the seventh possible implementation manner of the present invention, in an eighth possible implementation manner, step 321 includes:

[0043] Step 3211, using the time difference T between the two frames of image and the main lifting speed of the equipment, obtain the spatial height difference H corresponding to the two frames of image;

[0044] Step 3212, shift the next frame of image data downward by H as a whole;

[0045] Step 3213, obtaining the difference between the frame image data after the downward shift and the previous frame image data, and obtaining the difference image data;

[0046] Step 3214: Remove the image data in the top H and bottom H regions from the difference image data.

[0047] In combination with the eighth possible implementation manner of the present invention, in a ninth possible implementation manner, step 330 includes:

[0048] Step 331, extending the negative pressure vacuum cleaner grinding module in four directions: up, down, left, and right;

[0049] Step 332: Control the extension amount of the extension mechanism to obtain a specific tilt angle of the corresponding grinding module to fit the construction wall surface.

[0050] Compared with the existing technology, it has the following technical effects:

[0051] The wall roughness value is detected by the detection model, and the vacuum negative pressure corresponding to each wall roughness value is obtained. The dust 3D point cloud data is obtained through the deep image algorithm, and the point cloud data is used to obtain the construction gap of the grinding module. By changing the elongation of the elongation mechanism, the construction gap is adaptively adjusted, realizing adaptive control of construction dust and achieving the purpose of clean construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0053] Figure 1 It is a schematic diagram of a first embodiment of a method for controlling dust during wall construction in the present invention;

[0054] Figure 2 It is a schematic diagram of a second embodiment of a wall construction dust control method of the present invention;

[0055] Figure 3 It is a schematic diagram of a third embodiment of a wall construction dust control method of the present invention;

[0056] Figure 4 It is a schematic diagram of a fourth embodiment of a method for controlling dust during wall construction according to the present invention;

[0057] Figure 5 It is a schematic diagram of a fifth embodiment of a method for controlling dust during wall construction according to the present invention;

[0058] Figure 6 It is a schematic diagram of a sixth embodiment of a wall construction dust control method of the present invention;

[0059] Figure 7 It is a schematic diagram of a seventh embodiment of a method for controlling dust during wall construction according to the present invention;

[0060] Figure 8 It is a schematic diagram of an eighth embodiment of a method for controlling dust during wall construction according to the present invention;

[0061] Fig. 9 It is a schematic diagram of a ninth embodiment of a wall construction dust control method of the present invention;

[0062] Fig.10 It is a schematic diagram of a tenth embodiment of a method for controlling dust during wall construction according to the present invention; DETAILED DESCRIPTION

[0063] The following will be combined with the accompanying drawings in the invention to clearly and completely describe the technical solutions in the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the scope of protection of the present invention.

[0064] Glossary:

[0065] MLP: Multi-layer Perceptron, a feed-forward artificial neural network model that maps multiple input datasets to a single output dataset.

[0066] RANSAC algorithm: RANSAC is the abbreviation of "RANdom SAmple Consensus". It can estimate the parameters of a mathematical model from a set of observation data containing "outliers" in an iterative manner. It is an uncertain algorithm - it has a certain probability of obtaining a reasonable result; in order to increase the probability, the number of iterations must be increased.

[0067] SLAM: SLAM (simultaneous localization and mapping), also known as CML (Concurrent Mapping and Localization), real-time positioning and map construction, or concurrent mapping and positioning.

[0068] KD tree algorithm: k-dimensional tree, often used for space partitioning and nearest neighbor search, kd tree is a binary tree where each node is a k-dimensional numerical point, and each node on it represents a hyperplane, which is perpendicular to the coordinate axis of the current partition dimension and divides the space into two parts on this dimension, one in its left subtree and the other in its right subtree. That is, if the partition dimension of the current node is d, the coordinate values ​​of all points on its left subtree in the d dimension are less than the current value, and the coordinate values ​​of all points on the right subtree in the d dimension are greater than or equal to the current value.

[0069] BBF algorithm: By adding the nodes that may be passed by in the backtracking to the queue, and sorting them according to the distance from the search point to the hyperplane determined by the node, the highest priority node (i.e. the shortest distance) is traversed first each time until the queue is empty. The algorithm ends. At the same time, the BBF algorithm also sets a time limit. If the algorithm running time exceeds the limit, it will stop running regardless of whether it is empty or not, and return the current nearest neighbor as the result.

[0070] Existing indoor construction robots generate a lot of dust when performing indoor construction, causing environmental pollution. However, current indoor construction robots are unable to perform adaptive dust collection to achieve clean construction.

[0071] In view of the above problems, a method for controlling indoor construction dust is proposed.

[0072] A wall construction dust control method is used to adaptively control the dust generated by the construction, such as Figure 1 , Figure 1It is a schematic diagram of the first embodiment of a wall construction dust control method in the present invention, including: step 100, using the 2D image of the wall to be constructed and the corresponding depth data to construct a roughness detection model and use the roughness detection model to detect the wall roughness value; step 200, through multiple experiments, obtain the vacuum suction negative pressure corresponding to each wall roughness value; step 300, determine whether the vacuum suction gap is greater than the threshold value, if it is greater than the threshold value, adjust the inclination angle of the support mechanism so that the negative pressure vacuum cleaner fits the construction wall. The wall roughness value is detected by the detection model, and the vacuum suction negative pressure corresponding to each wall roughness value is obtained. The dust 3D point cloud data is obtained by the deep image algorithm, and the construction gap of the grinding module is obtained by using the point cloud data. By changing the elongation of the elongation mechanism, the construction gap is adaptively adjusted, and the construction dust is adaptively controlled, achieving the purpose of clean construction.

[0073] Preferably, if Figure 2 , Figure 2 It is a schematic diagram of a second embodiment of a wall construction dust control method in the present invention, and step 100 includes: step 110, using 2D images, depth data and roughness values ​​corresponding to the depth data to obtain a minimum sample set; step 120, determining a training set and a test set respectively according to the minimum sample set; step 130, using the training set to train a neural network deep learning model to obtain a roughness detection model; step 140, detecting the 2D image and depth data information of the construction wall and inputting them into the roughness detection model to obtain the roughness of the construction wall.

[0074] Preferably, if Figure 3 , Figure 3 It is a schematic diagram of the third embodiment of a wall construction dust control method in the present invention, and step 140 includes sub-steps: step 141, using the first deep learning structure in the roughness detection model to process the 2D image to obtain first depth data; step 142, using the second deep learning structure in the roughness detection model to process the depth data information to obtain second depth data; step 143, in the fully connected layer of the roughness detection model, the first depth data and the second depth data are merged and input into the fully connected layer again; step 144, the roughness value of each point cloud of the construction wall image is classified and judged, and the roughness value is output.

[0075] Different neural network learning structures are used for 2D image and depth data information respectively. The first deep learning structure preferably uses RESNET structure, and the second deep learning structure preferably uses PointNet structure. RESNET structure and PointNet structure perform convolution operations on 2D image and depth data information respectively, and merge them in the fully connected layer after the convolution layer. The n*1024 of the PointNet structure and the n*1024 of the maxpool layer of the RESNET structure are merged into n*2048, and then input into the fully connected layer after the merger, the roughness of the construction wall is classified, and finally the roughness value is output.

[0076] Preferably, if Figure 4 , Figure 4 It is a schematic diagram of the fourth embodiment of a wall construction dust control method in the present invention, and step 142 includes: step 1421, using the transformation matrix T-Net to align the point cloud data set; step 1422, extracting the features of the point cloud data set through multiple MLPs and aligning them again using the transformation matrix T-Net; step 1423, performing maxpooling pooling operations on each dimension of the features to obtain the final global features; step 1424, concatenating the global features with the local features of each point cloud and obtaining the classification results of each data point through MLP.

[0077] The collection of all point cloud data can be represented as an n×3 2d Tensor, where n is the number of point clouds, corresponding to XYZ coordinates. The transformation matrix T-Net ensures the invariance of the roughness detection model to specific spatial transformations. After multiple MLPs are used to extract features from each point cloud data, a transformation matrix T-Net is used for alignment.

[0078] Preferably, if Figure 5 , Figure 5 It is a schematic diagram of a fifth embodiment of a wall construction dust control method in the present invention, and step 200 includes: step 210, obtaining a discrete array of wall roughness values ​​and dust suction negative pressure through experiments; step 220, fitting the discrete array to obtain a corresponding relationship function between the wall roughness value and the dust suction negative pressure:

[0079] P=R*A+B (1)

[0080] Among them, P is the vacuum negative pressure, R is the wall roughness value, A is the wall roughness parameter, cement wall and putty wall correspond to different wall roughness parameters respectively, and B is the basic vacuum pressure.

[0081] Based on the detected roughness, through multiple tests, we can get the relationship between the roughness of cement walls and putty walls and the amount of dust generated by grinding, and finally get the relationship between the vacuum pressure size P of the negative pressure vacuum cleaner and the roughness of the wall. Through visual roughness detection, the pressure of negative pressure vacuuming is determined. Negative pressure suction is a global control quantity that can reduce the global dust amount.

[0082] Preferably, if Figure 6 , Figure 6 It is a schematic diagram of the sixth embodiment of a wall construction dust control method in the present invention, and step 300 includes: step 310, using an image depth algorithm to obtain dust point cloud information on the construction wall; step 320, using the dust point cloud information to obtain the construction gap between the negative pressure vacuum cleaner and the wall; step 330, if the construction gap is greater than a threshold, adjusting the inclination angle of the negative pressure vacuum cleaner.

[0083] Preferably, if Figure 7 , Figure 7 It is a schematic diagram of the seventh embodiment of a wall construction dust control method in the present invention, and step 310 includes: step 311, converting the depth image of the construction area into point cloud data information through coordinate transformation; step 312, dividing the point cloud data information into corresponding partitions.

[0084] The collected 3D data is processed as follows:

[0085] First, the 3D data is partitioned. The area on the left side of the data corresponds to the dust generated by the left side of the grinding module, the area on the right side of the 3D data corresponds to the dust generated by the right side of the grinding module, the area above the 3D data corresponds to the dust generated by the upper part of the grinding module, and the area below the 3D data corresponds to the dust generated by the lower grinding module.

[0086] Preferably, if Figure 8 , Figure 8 It is a schematic diagram of the eighth embodiment of a wall construction dust control method in the present invention, and step 320 includes: step 321, obtaining 3D point cloud data obtained by shooting two frames of the same construction space; step 322, using the 3D point cloud data, calculating the number of 3D points where data points appear in each area; step 323, if the number of 3D points exceeds a threshold, determining that the area is a construction gap.

[0087] Preferably, if Fig. 9 , Fig. 9It is a schematic diagram of the ninth embodiment of a wall construction dust control method in the present invention, and step 321 includes: step 3211, using the time difference T between the previous and next frames of images and the main lifting and lowering speed of the equipment to obtain the spatial height difference H corresponding to the two frames of images; step 3212, moving the next frame of image data downward by H as a whole; step 3213, obtaining the difference between the frame image data after moving downward and the previous frame image data, and obtaining difference image data; step 3214, removing the image data in the top H and bottom H area range from the difference image data.

[0088] Preferably, Fig.10 It is a schematic diagram of the tenth embodiment of a wall construction dust control method in the present invention; step 330 includes: step 331, extending the negative pressure vacuum cleaner grinding module in four directions of up, down, left and right; step 332, controlling the extension amount of the extension mechanism to obtain a specific inclination angle of the grinding module to fit the construction wall.

[0089] Based on this 3D point cloud, the number of points in each area is calculated. If the number exceeds the set threshold, it can be considered that a gap appears in the corresponding area. Based on the gap position, the corresponding elongation structure of the grinding module is adjusted.

[0090] When the coarse sanding module and the fine sanding module are sanded close to the wall, dust is generated evenly and is completely sucked away by the negative pressure suction module through the internal pipe. When the sanding module is not in close contact with the wall and there is a gap, dust will splash out from the gap. The gap problem can be solved by changing the elongation of the sanding module in four positions: up, down, left and right.

[0091] The coarse grinding module and the fine grinding module can be simply understood as a high-speed rotating disc pressed against the wall, with sandpaper at the contact point between the wall and the disc, and the grinding effect is achieved through the contact between the sandpaper and the wall. Because the wall is tilted horizontally and vertically, there may be a gap between the grinding module and the wall. Therefore, the grinding module has an extension mechanism in four directions: up, down, left, and right. The extension amount of the extension mechanism can be adjusted to make the grinding module tilt vertically and horizontally to a certain extent, so that the grinding module fits the wall completely.

[0092] When the bottom of the wall is convex and the top is concave, the polishing module needs the upper extension mechanism to increase the extension amount so that the polishing module tilts downward and fits the wall; when the top of the wall is convex and the bottom is concave, the polishing module needs the lower extension mechanism to increase the extension amount so that the polishing module tilts upward and fits the wall.

[0093] When the left part of the wall is convex and the right part is concave, the grinding module needs the right extension mechanism to increase the extension amount so that the grinding module tilts to the left and fits the wall; when the right part of the wall is convex and the left part is concave, the grinding module needs the left extension mechanism to increase the extension amount so that the grinding module tilts to the right and fits the wall.

[0094] Compared with the existing technology, it has the following technical effects:

[0095] The wall roughness value is detected by the detection model, and the vacuum negative pressure corresponding to each wall roughness value is obtained. The dust 3D point cloud data is obtained through the deep image algorithm, and the point cloud data is used to obtain the construction gap of the grinding module. By changing the elongation of the elongation mechanism, the construction gap is adaptively adjusted, realizing adaptive control of construction dust and achieving the purpose of clean construction.

[0096] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A wall construction dust control method for adaptively controlling dust generated during construction, characterized in that: include: Step 100: construct a roughness detection model using the 2D image of the wall to be constructed and the corresponding depth data, and use the roughness detection model to detect the wall roughness value; Step 200, obtaining the vacuum negative pressure corresponding to each wall roughness value through multiple experiments; Step 300, determining whether the vacuum gap is greater than a threshold value, and if so, adjusting the inclination angle of the support mechanism so that the negative pressure vacuum cleaner fits the construction wall surface; Wherein, the step 300 includes: Step 310: Obtain construction wall dust point cloud information using an image depth algorithm; Step 320: using the dust point cloud information to obtain the construction gap between the negative pressure vacuum cleaner and the wall; Step 330: if the construction gap is greater than a threshold, adjusting the tilt angle of the negative pressure vacuum cleaner; The step 320 includes: Step 321, obtaining 3D point cloud data obtained by shooting two frames of the same construction space; Step 322: Calculate the number of 3D points where data points appear in each area using the 3D point cloud data; Step 323: If the number of 3D points exceeds a threshold, the area is determined to be a construction gap.

2. The wall construction dust control method according to claim 1, characterized in that: The step 100 comprises: Step 110: Acquire a minimum sample set using the 2D image, the depth data, and the coarse value corresponding to the depth data; Step 120: Determine a training set and a test set respectively according to the minimum sample set; Step 130: Use the training set to train a neural network deep learning model to obtain a roughness detection model; Step 140: Detect the 2D image and depth data information of the construction wall surface and input them into the roughness detection model to obtain the roughness of the construction wall surface.

3. The wall construction dust control method according to claim 2, characterized in that: The step 140 includes: Step 141: Process the 2D image using the first deep learning structure in the roughness detection model to obtain first depth data; Step 142: Process the depth data information using the second deep learning structure in the roughness detection model to obtain second depth data; Step 143: In the fully connected layer of the roughness detection model, the first depth data and the second depth data are merged and input into the fully connected layer again; Step 144: classify and judge the roughness value of each point cloud of the 2D image of the construction wall surface, and output the roughness value.

4. The wall construction dust control method according to claim 3, characterized in that: The step 142 includes: Step 1421, aligning the point cloud data set using the transformation matrix T-Net; Step 1422, extracting the features of the point cloud data set through multiple MLPs and aligning them again using the transformation matrix T-Net; Step 1-423, perform maxpooling pooling operation on each dimension of the feature to obtain the final global feature; Step 1424: Concatenate the global features with the local features of each point cloud and obtain the classification result of each data point through MLP.

5. The wall construction dust control method according to claim 4, characterized in that: The step 200 comprises: Step 210: Obtain a discrete array of the wall surface roughness value and the vacuum negative pressure through experiments; Step 220: Fit the discrete array to obtain the corresponding relationship function between the wall surface roughness value and the vacuum negative pressure: P = R * A + B; Among them, P is the vacuum pressure, R is the wall roughness value, A is the wall roughness parameter; and B is the basic vacuum pressure.

6. The wall construction dust control method according to claim 5, characterized in that: The step 310 includes: Step 311, converting the depth image of the construction area into point cloud data information through coordinate transformation; Step 312: partition the point cloud data information into corresponding partitions.

7. The wall construction dust control method according to claim 1, characterized in that: The step 321 includes: Step 3211, using the time difference T between the two frames of image and the main lifting speed of the equipment, obtain the spatial height difference H corresponding to the two frames of image; Step 3212, shift the next frame of image data downward by H as a whole; Step 3213, obtaining the difference between the frame image data after the downward shift and the previous frame image data, and obtaining the difference image data; Step 3214: Remove the image data in the top H and bottom H regions from the difference image data.

8. The wall construction dust control method according to claim 7, characterized in that: The step 330 includes: Step 331, extending the negative pressure vacuum cleaner grinding module in four directions: up, down, left, and right; Step 332: Control the extension amount of the extension mechanism to obtain a specific tilt angle of the corresponding grinding module to fit the construction wall surface.

Images