Installation simulation and control methods, devices, electronic equipment, and media applied to glass curtain walls
By partitioning and modeling the images of the glass curtain wall installation scene, and combining this with path planning, the installation was carried out using a construction robot. This solved the problem of low installation efficiency caused by the variety of sensors, and achieved efficient glass curtain wall installation.
Patent Information
- Application Number
- CN202610006075.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-05
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2046-01-05
AI Technical Summary
In the current glass curtain wall installation process, the use of sensors for on-site detection results in a wide variety of sensors, and the installation and removal of these sensors is time-consuming, leading to low installation efficiency.
By partitioning the pre-acquired image set of the glass curtain wall installation scene, a set of unit point cloud information is generated, which is then modeled and verified. Path planning is performed based on the glass curtain wall model information, and installation is carried out using a construction robot.
This reduces the types of sensors and installation time, thus improving the efficiency of glass curtain wall installation.
Smart Images

Figure CN121959910B_ABST
Abstract
Description
Technical Field
[0001] The embodiments of this application relate to the field of digital construction, specifically to installation simulation and control methods, apparatuses, electronic devices, and media applied to glass curtain walls. Background Technology
[0002] Glass curtain walls are a modern architectural exterior wall decoration structure with advantages such as aesthetics, light transmission, and heat insulation, and are widely used in high-rise buildings. Currently, the common method for simulating glass curtain wall installation is as follows: 3D modeling of the glass curtain wall is performed, an installation model of the glass curtain wall is designed, on-site detection using sensors determines whether it meets the construction conditions, and then building materials are transported by drone for assembly.
[0003] However, in practice, it has been found that when using the above methods for installation simulation and control of glass curtain walls, the following technical problems often arise: The method of on-site detection using sensors requires a variety of sensors, which are difficult to install in a timely manner. Furthermore, the installation and removal of sensors both take time, resulting in reduced efficiency in glass curtain wall installation. Summary of the Invention
[0004] The summary section of this application is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description section below. This summary section is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.
[0005] Some embodiments of this application propose installation simulation and control methods, apparatuses, computer equipment, and computer-readable storage media for glass curtain walls to solve one or more of the technical problems mentioned in the background section above.
[0006] In a first aspect, some embodiments of this application provide an installation simulation and control method for glass curtain walls. The method includes: partitioning each glass curtain wall installation scene image in a pre-acquired glass curtain wall installation scene image set to obtain a set of unit point cloud information; performing modeling processing on the unit point cloud information set to obtain installation scene model information; performing verification processing on the installation scene model information based on the pre-acquired glass curtain wall model information to obtain a verification result; in response to determining that the verification result meets preset verification conditions, performing path planning processing on the glass curtain wall model information based on a pre-acquired installation equipment information set to obtain installation control planning path information; and sending the installation control planning path information to a target building robot to perform glass curtain wall installation.
[0007] Secondly, some embodiments of this disclosure provide an installation simulation and control device for glass curtain walls. The device includes: a partitioning unit configured to partition each glass curtain wall installation scene image in a pre-acquired glass curtain wall installation scene image set to obtain a set of unit point cloud information; a modeling unit configured to model the unit point cloud information set to obtain installation scene model information; a verification unit configured to verify the installation scene model information based on the pre-acquired glass curtain wall model information to obtain a verification result; a path planning unit configured to, in response to determining that the verification result meets preset verification conditions, perform path planning on the glass curtain wall model information based on a pre-acquired installation equipment information set to obtain installation control planning path information; and a sending unit configured to send the installation control planning path information to a target building robot for glass curtain wall installation.
[0008] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.
[0009] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.
[0010] The above embodiments of this application have the following beneficial effects: the installation simulation and control method for glass curtain walls applied to some embodiments of this application can improve the installation efficiency of glass curtain walls. Specifically, the reason for the reduced installation efficiency of glass curtain walls is that the on-site detection method using sensors requires a variety of sensors, making timely installation difficult, and both sensor installation and removal require time. Based on this, the installation simulation and control method for glass curtain walls applied to some embodiments of this application firstly partitions each glass curtain wall installation scene image in a pre-acquired glass curtain wall installation scene image set to obtain a set of unit point cloud information. This allows the acquisition of feature information from the scene images. Secondly, the unit point cloud information set is modeled to obtain installation scene model information. This allows the glass curtain wall installation scene to be modeled based on the image feature information. Then, based on the pre-acquired glass curtain wall model information, the installation scene model information is verified to obtain a verification result. This determines that the current installation scene meets the construction requirements of the glass curtain wall. Next, in response to the determination that the above verification results meet the preset verification conditions, based on the pre-acquired installation equipment information set, path planning processing is performed on the above glass curtain wall model information to obtain installation control planning path information. This allows for intelligent planning of the construction robot's trajectory. Finally, the above installation control planning path information is sent to the target construction robot for glass curtain wall installation. This allows for control of the glass curtain wall installation. Therefore, some of the installation simulation and control methods applied to glass curtain walls in this application can acquire images of the glass curtain wall installation scene using a single camera, and model the installation scene based on the acquired images. This reduces the types of sensors and installation time, thereby reducing the time for site calibration and inspection before glass curtain wall installation, and ultimately improving the installation efficiency of the glass curtain wall. Attached Figure Description
[0011] The above and other features, advantages, and aspects of the embodiments of this application will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.
[0012] Figure 1 These are application scenario diagrams of some embodiments of the installation simulation and control method for glass curtain walls according to this application; Figure 2 This is a flowchart of some embodiments of the installation simulation and control method applied to glass curtain walls according to this application; Figure 3 These are schematic diagrams of some embodiments of the installation simulation and control device for glass curtain walls according to the present disclosure; Figure 4 This is a schematic diagram of the structure of a computer device suitable for implementing some embodiments of this application. Detailed Implementation
[0013] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.
[0014] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described herein can be combined with each other.
[0015] It should be noted that the concepts of "first" and "second" mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.
[0016] It should be noted that the terms "a" and "a plurality of" used in this application are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".
[0017] The names of the messages or information exchanged between multiple devices in the embodiments of this application are for illustrative purposes only and are not intended to limit the scope of these messages or information.
[0018] The present application will now be described in detail with reference to the accompanying drawings and embodiments.
[0019] Figure 1 This is a schematic diagram of an application scenario of the installation simulation and control method for glass curtain walls, which is one of the embodiments of this disclosure.
[0020] exist Figure 1In the application scenario, firstly, the computing device can partition the various glass curtain wall installation scene images in the pre-acquired glass curtain wall installation scene image set 102 to obtain a unit point cloud information group set 103. Optionally, the aforementioned glass curtain wall installation scene image set 102 can be acquired from the camera device 101. The glass curtain wall installation scene images in the aforementioned glass curtain wall installation scene image set 102 can be: glass curtain wall installation scene image A, glass curtain wall installation scene image B, and glass curtain wall installation scene image C. The unit point cloud information groups in the aforementioned unit point cloud information group set 103 can be: unit point cloud information group A, unit point cloud information group B, and unit point cloud information group C. Here, the glass curtain wall installation scene image A included in the aforementioned glass curtain wall installation scene image set 102 can correspond to the unit point cloud information group A included in the aforementioned unit point cloud information group set 103, the glass curtain wall installation scene image B included in the aforementioned glass curtain wall installation scene image set 102 can correspond to the unit point cloud information group B included in the aforementioned unit point cloud information group set 103, and the glass curtain wall installation scene image C included in the aforementioned glass curtain wall installation scene image set 102 can correspond to the unit point cloud information group C included in the aforementioned unit point cloud information group set 103. Then, the computing device can perform modeling processing on the aforementioned unit point cloud information group set 103 to obtain installation scene model information 104. Next, the computing device can perform verification processing on the aforementioned installation scene model information 104 based on the pre-acquired glass curtain wall model information 105 to obtain verification result 106. Subsequently, in response to determining that the above verification result 106 meets the preset verification conditions, the computing device can perform path planning processing on the above glass curtain wall model information 105 based on the pre-acquired installation equipment information set 107 to obtain installation control planning path information 108. Optionally, the above preset verification condition can be "verification successful". Optionally, the installation equipment information in the above installation equipment information set 107 can be: "elevator", "spraying equipment" or "crane". Finally, the computing device can send the above installation control planning path information 108 to the target construction robot 109 for glass curtain wall installation.
[0021] It should be noted that the aforementioned computing devices can be either hardware or software. When the computing device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or a single terminal device. When the computing device is software, it can be installed on the hardware devices listed above. It can be implemented as, for example, multiple software programs or software modules used to provide distributed services, or as a single software program or software module. No specific limitations are made here.
[0022] Next reference Figure 2 , Figure 2A flowchart 200 is shown, illustrating some embodiments of the installation simulation and control method for glass curtain walls according to this application. The installation simulation and control method for glass curtain walls includes the following steps: Step 201: Partition the glass curtain wall installation scene images in the pre-acquired glass curtain wall installation scene image set to obtain a set of unit point cloud information.
[0023] In some embodiments, the execution entity of the installation simulation and control method applied to glass curtain walls can partition each glass curtain wall installation scene image in a pre-acquired glass curtain wall installation scene image set to obtain a unit point cloud information set. In some embodiments, a point cloud extraction algorithm can be used to partition each glass curtain wall installation scene image in the pre-acquired glass curtain wall installation scene image set to obtain a unit point cloud information set. The point cloud extraction algorithm can be as follows: for each glass curtain wall installation scene image in the glass curtain wall installation scene image set, firstly, a depth image corresponding to the glass curtain wall installation scene image is extracted using a depth estimation model. Then, based on the pre-acquired camera parameter information corresponding to the glass curtain wall installation scene image, each depth image in the depth image set is converted into point cloud coordinates to generate a unit point cloud information set, thus obtaining a unit point cloud information set. The glass curtain wall installation scene image set can be acquired from at least one camera device via a wired or wireless connection. The pre-acquired camera parameter information corresponding to the glass curtain wall installation scene image can be camera parameter information acquired from the camera device used to capture the glass curtain wall installation scene image. The camera parameter information described above can characterize the parameter features of the corresponding camera equipment. The glass curtain wall installation scene images in the above glass curtain wall installation scene image set can be two-dimensional RGB (red-green-blue) images of the glass installation scene captured by camera equipment.
[0024] As an example, the aforementioned camera device could be a drone equipped with a camera.
[0025] It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other currently known or future wireless connection methods.
[0026] In some optional implementations of certain embodiments, the execution entity performs partitioning processing on each glass curtain wall installation scene image in the pre-acquired glass curtain wall installation scene image set to obtain a set of unit point cloud information, which may include the following steps: The first step involves restoring the glass curtain wall installation scene images from a pre-acquired image set to obtain a sparse point cloud information set of the installation scene. This restoration process can be achieved by: first, extracting feature points from each glass curtain wall installation scene image in the image set to generate a feature point set; then, matching the feature points in the feature point set to obtain a matching feature point information set, where each matching feature point information set includes two feature points corresponding to two glass curtain wall installation scene images; next, for each matching feature point in the matching feature point information set, based on the two corresponding glass curtain wall installation scene images in the image set, determining the camera pose information corresponding to the glass curtain wall installation scene image using a function, and then determining the sparse point cloud information of the installation scene corresponding to the matching feature point information using triangulation. Finally, the generated sparse point cloud information of each installation scene is determined as the sparse point cloud information set of the installation scene. Therefore, by restoring the structure through motion, the three-dimensional information in the two-dimensional image can be extracted. Using the pixel information in the two-dimensional image, the positions of these pixels in three-dimensional space, i.e., the three-dimensional point cloud information, can be calculated in reverse.
[0027] The second step involves splitting the sparse point cloud information of each installation scenario in the aforementioned sparse point cloud information set to obtain a set of sparse point cloud information for each installation scenario and a set of installation scenario location information. Specifically, the splitting of the sparse point cloud information of each installation scenario in the aforementioned sparse point cloud information set to obtain the set of sparse point cloud information for each installation scenario can be performed as follows: First, the sparse point cloud information of each installation scenario in the above sparse point cloud information set is vectorized to obtain a point cloud vector set.
[0028] Next, a target number of point cloud vectors are randomly selected from the point cloud vector set as the initial cluster center vectors to obtain the initial cluster center vector set.
[0029] Next, for each point cloud vector in the point cloud vector set that was not selected, the cosine value of the point cloud vector and each initial cluster center vector in the initial cluster center vector set is determined as the similarity value, thus obtaining a similarity value group.
[0030] Next, the point cloud vector is added to the initial cluster vector group corresponding to the initial cluster center vector corresponding to the smallest similarity value in the similarity value group. The initial cluster vector group initially includes the initial cluster center vector.
[0031] Next, for each generated initial cluster vector group, the following update sub-step is performed: The first sub-step involves determining the mean of each initial cluster vector in the initial cluster vector group as the updated cluster center vector.
[0032] The second sub-step involves determining the cosine value of the updated cluster center vector and the corresponding initial cluster center vector as the cluster center similarity value.
[0033] The third sub-step, in response to determining that the cluster center similarity value is greater than a preset center similarity value, identifies the sparse point cloud information of the installation scene corresponding to each initial cluster vector in the initial cluster vector group as a cluster information group, and identifies the sparse point cloud information of the installation scene corresponding to the updated cluster center vector as cluster center information. For example, the preset center similarity value can be 0.9.
[0034] Subsequently, the identified cluster information groups are defined as cluster information group sets, and the identified cluster center information groups are defined as cluster center information sets.
[0035] Then, the generated cluster information set can be identified as the sparse point cloud information set of the installation scene, and the generated cluster center information set can be identified as the installation scene location information set. Here, the sparse point cloud information sets of the installation scene in the aforementioned sparse point cloud information set correspond one-to-one with the installation scene location information in the aforementioned installation scene location information set. The installation scene location information in the aforementioned installation scene location information can characterize the position of the glass curtain wall installation scene image corresponding to the corresponding sparse point cloud information set in the glass curtain wall installation scene.
[0036] As an example, the target number could be 5.
[0037] The third step involves expanding each sparse point cloud information group in the aforementioned installation scenario sparse point cloud information group set to generate unit point cloud information groups, thus obtaining a set of unit point cloud information groups. Specifically, expanding each sparse point cloud information group in the aforementioned installation scenario sparse point cloud information group set to generate unit point cloud information groups can be achieved by randomly selecting a predetermined number of sparse point cloud information from each adjacent sparse point cloud information group and adding them to the aforementioned sparse point cloud information group, resulting in an added sparse point cloud information group, which serves as a unit point cloud information group. Then, each generated unit point cloud information group is defined as a set of unit point cloud information groups.
[0038] As an example, the aforementioned preset quantity can be the product of the number of installation scene sparse point cloud information in each installation scene sparse point cloud information group adjacent to the aforementioned installation scene sparse point cloud information group and a preset ratio. The aforementioned preset ratio can be, but is not limited to, at least one of the following: 0.1, 0.05, or 0.08.
[0039] Step 202: Model the unit point cloud information set to obtain the installation scene model information.
[0040] In some embodiments, the aforementioned execution entity can perform modeling processing on the aforementioned unit point cloud information set to obtain installation scene model information. Specifically, the installation scene model information can be obtained by modeling the aforementioned unit point cloud information set through 3D reconstruction.
[0041] In some optional implementations of certain embodiments, the execution entity performs modeling processing on the aforementioned unit point cloud information set to obtain installation scene model information, which may include the following steps: The first step is to perform Gaussian transformation on each unit point cloud information in the aforementioned unit point cloud information set to generate unit point cloud Gaussian information, thus obtaining the unit point cloud Gaussian information set. This can be achieved through 3D Gaussian modeling.
[0042] The second step involves inputting each Gaussian information group from the aforementioned unit point cloud Gaussian information group set into a pre-trained Gaussian decoupling network model to generate scene partitioned rendering images, thus obtaining a scene partitioned rendering image set. The pre-trained Gaussian decoupling network model can be an untrained neural network model that takes unit point cloud Gaussian information groups as input and scene partitioned rendering images as output.
[0043] Optionally, the aforementioned Gaussian decoupling network model can consist of an initial convolutional sub-model, a downsampling sub-model, and a target convolutional sub-model. The initial convolutional sub-model can be a CNN (Convolutional Neural Network) to perform preliminary feature extraction from the Gaussian information group of the input unit point cloud. The downsampling sub-model can be a downsampling block model to progressively downsample the extracted feature maps, reducing the dimensionality of the feature data. The target convolutional sub-model can be a KAN (Kolmogorov-Arnold Network) convolutional network, which improves the accuracy of the convolutional neural network by replacing fixed weight parameters with learnable functions, thereby increasing the accuracy of the extracted feature images.
[0044] The third step involves stitching together the scene partition rendering images from the aforementioned scene partition rendering image set to obtain the installation scene model information. This stitching process can be performed by combining the scene partition rendering images from the aforementioned scene partition rendering image set into the installation scene model information based on the installation scene location information corresponding to the scene partition rendering images.
[0045] Optionally, before inputting each Gaussian information group of the aforementioned unit point cloud Gaussian information group set into a pre-trained Gaussian decoupling network model to generate scene partitioned rendering images and obtain a scene partitioned rendering image set, the aforementioned execution entity may also perform the following steps: The first step is to obtain a set of sample scene images. This set includes a set of sample pixel information. The sample scene image set can be obtained from a scene image database. This scene image database can be a database used to store installation scene images. The sample pixel information in the sample pixel information set represents a pixel in the corresponding sample scene image.
[0046] The second step involves selecting target sample scene images from the aforementioned sample scene image set and performing the following training sub-steps: The first sub-step involves performing a Gaussian transformation on each sample pixel in the target sample scene image's pixel information set to generate initial Gaussian pixel information, thus obtaining an initial Gaussian pixel information set. Specifically, a sample scene image can be randomly selected from the aforementioned sample scene image set as the target sample scene image. Gaussian modeling can be used to perform a Gaussian transformation on each sample pixel in the target sample scene image's pixel information set to generate the initial Gaussian pixel information set.
[0047] The second sub-step involves inputting the initial Gaussian information set of pixels into the initial convolutional sub-model included in the initial Gaussian decoupling network model to obtain the initial convolutional image. The initial Gaussian decoupling network model further includes a downsampling sub-model and a target convolutional sub-model. Specifically, the aforementioned initial Gaussian decoupling network model can be an untrained neural network model that takes the initial Gaussian information set of pixels as input and the initial rendered image as output.
[0048] Optionally, the initial convolutional sub-model described above can be a CNN (Convolutional Neural Network) to perform preliminary feature extraction from the Gaussian information group of the input unit point cloud. The downsampling sub-model described above can be a downsampling block model to progressively downsample the extracted feature maps, reducing the dimensionality of the feature data. The target convolutional sub-model described above can be composed of a KAN (Kolmogorov-Arnold Network) convolutional network. By replacing the fixed weight parameters with learnable functions, the accuracy of the convolutional neural network is improved, thereby increasing the accuracy of the extracted feature images.
[0049] The third sub-step involves inputting the initial convolutional image into the downsampling sub-model included in the initial Gaussian decoupled network model to obtain the initial downsampled image.
[0050] The fourth sub-step involves inputting the initial downsampled image into the target convolutional sub-model included in the initial Gaussian decoupled network model to obtain the initial rendered image.
[0051] The fifth sub-step involves determining the decoupling loss value between the target sample scene image and the initial rendered image based on a preset decoupling loss function. This decoupling loss function can include a distortion loss function, a consistency loss function, and a raw loss function. Specifically, determining the decoupling loss value based on the preset decoupling loss function can be achieved by: first, determining the distortion loss value between the target sample scene image and the initial rendered image based on the distortion loss function included in the decoupling loss function; second, determining the consistency loss value between the target sample scene image and the initial rendered image based on the consistency loss function included in the decoupling loss function; third, determining the raw loss value between the target sample scene image and the initial rendered image based on the raw loss function included in the decoupling loss function; and finally, determining the decoupling loss value by summing the distortion loss value, the consistency loss value, and the raw loss value.
[0052] As an example, the distortion loss function described above could be a depth distortion loss function. The consistency loss function described above could be a normal consistency loss function. The original loss function described above could be a 3D Gaussian loss function.
[0053] The sixth sub-step is to determine the initial Gaussian decoupling network model as a Gaussian decoupling network model in response to the determination that the decoupling loss value is less than the target threshold.
[0054] As an example, the target threshold mentioned above could be 0.01.
[0055] Optionally, in response to determining that the decoupling loss value is greater than or equal to the target threshold, the aforementioned execution entity may also adjust the relevant parameters in the initial Gaussian decoupling network model, determine the adjusted initial Gaussian decoupling network model as the initial Gaussian decoupling network model, and select target sample scene images from the aforementioned sample scene image set, from the unselected sample scene images, for re-executing the aforementioned training steps. The relevant parameters in the initial Gaussian decoupling network model can be adjusted using a preset adjustment algorithm.
[0056] As an example, the preset adjustment algorithm mentioned above can be, but is not limited to, one of the following: backpropagation algorithm or stochastic gradient algorithm.
[0057] The content related to step 202, as an inventive point of this application, solves the technical problem of "reduced installation efficiency of glass curtain wall installation." The factors leading to reduced installation efficiency of glass curtain wall installation are often as follows: existing methods of scene modeling through image acquisition only use a single neural network model for image modeling, resulting in reduced accuracy of the scene model. This makes it difficult to determine in a timely manner whether the installation scene is suitable for construction, thus reducing the installation efficiency of the glass curtain wall. Solving these factors can improve the installation efficiency of glass curtain wall installation. To achieve this effect, this application can establish Gaussian metadata corresponding to each point cloud using a Gaussian sputtering model. Then, the generated Gaussian information can be decoupled and further rendered sequentially through a first convolutional neural network, a sampling network, and a second convolutional neural network, thereby improving the accuracy of the scene model, reducing the probability of errors in subsequent verification, and thus improving the installation efficiency of the glass curtain wall.
[0058] Step 203: Based on the pre-acquired glass curtain wall model information, the installation scene model information is verified to obtain the verification result.
[0059] In some embodiments, the aforementioned execution entity can perform verification processing on the installation scene model information based on pre-acquired glass curtain wall model information to obtain a verification result. Specifically, this verification process based on the pre-acquired glass curtain wall model information can involve: first, inputting the point cloud information included in the glass curtain wall model information and the point cloud information included in the installation scene model information into a one-way nearest neighbor formula to obtain a similarity value; second, in response to determining that the similarity value is greater than or equal to a preset similarity value, determining the first preset verification information as the verification result; and then, in response to determining that the similarity value is less than the preset similarity value, determining the second preset verification information as the verification result. The aforementioned pre-acquired glass curtain wall model information can characterize the features of the glass curtain wall to be installed.
[0060] As an example, the aforementioned preset similarity matching algorithm may include, but is not limited to, at least one of the following: graph matching algorithm, chamfer distance formula, and Hausdorff distance formula. The aforementioned preset similarity value may be 0.8. The aforementioned first preset verification information may be information indicating "verification passed". The aforementioned second preset verification information may be information indicating "verification failed".
[0061] Step 204: In response to the determination that the verification result meets the preset verification conditions, based on the pre-acquired set of installation equipment information, the glass curtain wall model information is processed for path planning to obtain installation control planning path information.
[0062] In some embodiments, the execution entity may, in response to determining that the verification result meets preset verification conditions, perform path planning processing on the glass curtain wall model information based on a pre-acquired set of installation equipment information to obtain installation control planning path information. The preset verification conditions may be information indicating "verification passed" for the verification result. The installation equipment information in the installation equipment information set may represent a mechanical device. The installation equipment information may include: installation task information. The installation task information may include: task route start point, task route end point, initial task path information, and initial task data volume. The installation task information may represent the installation task that the installation equipment wants to perform. For example, transporting the glass curtain wall from a storage warehouse to the glass curtain wall installation point. The task route start point may be the storage warehouse. The task route end point may be the glass curtain wall installation point. The initial task path information may represent a randomly generated transportation path from the task route start point to the task route end point. The initial task data volume may be the unoptimized data volume required for the installation equipment to run when performing the installation task according to the initial task path information. Optionally, the above glass curtain wall model information can be processed by path planning using a random generation method to obtain installation control planning path information.
[0063] As an example, the aforementioned mechanical equipment could be: unmanned transport aircraft, elevators, spraying equipment, or cranes.
[0064] In some optional implementations of certain embodiments, the execution entity performs path planning processing on the glass curtain wall model information based on a pre-acquired set of installation equipment information to obtain installation control planning path information, which may include the following steps: The first step is to extract target points from the aforementioned glass curtain wall model information to obtain a set of installation target point pairs. This extraction process can involve identifying the glass curtain wall installation points and material storage points included in the glass curtain wall model information as the installation target point pairs.
[0065] The second step is to perform obstacle extraction processing on the aforementioned glass curtain wall model information to obtain a scene obstacle information set. This obstacle extraction process can be achieved by identifying point cloud information with height values greater than a preset height value within the point cloud information included in the glass curtain wall model information as scene obstacles, thus obtaining the obstacle information set.
[0066] As an example, the preset height value can be, but is not limited to, at least one of the following: 10 cm, 15 cm, or 20 cm.
[0067] The third step involves modeling each piece of installation equipment information in the aforementioned installation equipment information set to generate installation equipment motion model information, thus obtaining an installation equipment motion model information set. This modeling process can be achieved by first capturing images of at least one of the aforementioned installation equipment information using a camera, resulting in an installation equipment image set. Next, point cloud extraction algorithms can be used to extract point clouds from each installation equipment image in the image set, generating installation equipment point cloud information, thus obtaining an installation equipment point cloud information set. Then, 3D reconstruction can be used to model each installation equipment point cloud in the installation equipment point cloud information set, resulting in installation equipment motion model information. Therefore, the installation equipment information in the aforementioned installation equipment information set can be modeled to obtain a 3D model of the installation equipment. The installation equipment motion model information includes: equipment power values. These equipment power values characterize the operating power of the corresponding installation equipment.
[0068] The fourth step involves processing the motion model information of each installation device in the above-mentioned installation device motion model information set to generate initial path planning information, thus obtaining the initial path planning information set.
[0069] The fifth step involves optimizing the initial path planning information set to obtain installation control planning path information. This optimization can be achieved by: for each initial path planning information in the set, determining that it has no intersection with any other initial path planning information in the set, thus defining the set as the installation control planning path information; and re-exercising the planning steps if an intersection is found between the initial path planning information and any other initial path planning information in the set.
[0070] In some optional implementations of certain embodiments, the execution entity performs planning processing on each installation equipment motion model information in the installation equipment motion model information set based on the aforementioned scene obstacle information set to generate initial path planning information, which may include the following steps: The first step involves performing task unloading processing on the aforementioned installation equipment motion model information based on the scene obstacle information set, thereby obtaining unloading energy consumption information. Specifically, this process can be achieved by first determining the initial task information using a preset task duration value and the initial task data volume corresponding to the installation equipment motion model information. Then, in response to determining that the initial task data volume is less than the target data volume, the first preset region information is determined as the unloading energy consumption information, and this information is added to the initial task information to obtain the added initial task information, which serves as the unloading energy consumption information. Similarly, in response to determining that the initial task data volume is greater than or equal to the target data volume, the second preset region information is determined as the unloading energy consumption information, and this information is added to the initial task information to obtain the added initial task information, which serves as the unloading energy consumption information. The first preset region information can represent information indicating "execution locally." The second preset region information can represent information indicating "execution on other nodes."
[0071] As an example, the target data volume mentioned above can be, but is not limited to, at least one of the following: 30MB (Megabit), 50MB, or 100MB.
[0072] The second step is to determine the latency determination model based on the aforementioned offloaded energy consumption information. Specifically, determining the latency determination model based on the aforementioned offloaded energy consumption information can be achieved by: when the task execution region information included in the aforementioned offloaded energy consumption information represents "execution on local machine," determining the first preset model as the latency determination model; and when the task execution region information included in the aforementioned offloaded energy consumption information represents "execution on other nodes," determining the second preset model as the latency determination model.
[0073] As an example, the first preset model mentioned above can be a local latency calculation model, and the second preset model mentioned above can be an edge node latency calculation model.
[0074] The third step is to determine the equipment energy consumption information based on the aforementioned time delay determination model. Specifically, determining the equipment energy consumption information based on the aforementioned time delay determination model and the aforementioned scene obstacle information set can be achieved by first inputting the aforementioned installation equipment motion model information into the aforementioned time delay determination model to obtain the task time delay value; then, multiplying the aforementioned task time delay value by the equipment power value included in the aforementioned installation equipment motion model information to determine the aforementioned equipment energy consumption information.
[0075] The fourth step involves optimizing the aforementioned equipment energy consumption information to obtain initial path planning information. This optimization can be achieved by first inputting the energy consumption information into a pre-trained decision model to obtain a decision result. This decision result can represent either "optimization complete" or "optimization incomplete." Then, in response to the determination that the decision result satisfies a preset optimization condition, the initial task path information is identified as the initial path planning information. Here, the preset optimization condition can be that the decision result represents "optimization complete." The pre-trained decision model can employ a Markov decision model, thereby improving the accuracy of the generated initial path planning information.
[0076] Optionally, in response to determining that the above decision result does not meet the preset optimization conditions, the updated initial task path information is determined as the initial task path information to execute the above planning steps again.
[0077] Therefore, a model can be used to plan the path for at least one glass curtain wall installation device. Then, by optimizing the edge nodes, the path of each glass curtain wall installation device can be optimized with the goal of minimizing latency and energy consumption. This can reduce the frequency of conflicts during the operation of the glass curtain wall installation devices, thereby improving the efficiency of glass curtain wall installation.
[0078] Step 205: Send the installation control planning path information to the target building robot for glass curtain wall installation.
[0079] In some embodiments, the execution entity may send the installation control planning path information to the target building robot for glass curtain wall installation. The target building robot may be a robot used for installing glass curtain walls.
[0080] The above embodiments of this application have the following beneficial effects: the installation simulation and control method for glass curtain walls applied to some embodiments of this application can improve the installation efficiency of glass curtain walls. Specifically, the reason for the reduced installation efficiency of glass curtain walls is that the on-site detection method using sensors requires a variety of sensors, making timely installation difficult, and both sensor installation and removal require time. Based on this, the installation simulation and control method for glass curtain walls applied to some embodiments of this application firstly partitions each glass curtain wall installation scene image in a pre-acquired glass curtain wall installation scene image set to obtain a set of unit point cloud information. This allows the acquisition of feature information from the scene images. Secondly, the unit point cloud information set is modeled to obtain installation scene model information. This allows the glass curtain wall installation scene to be modeled based on the image feature information. Then, based on the pre-acquired glass curtain wall model information, the installation scene model information is verified to obtain a verification result. This determines that the current installation scene meets the construction requirements of the glass curtain wall. Next, in response to the determination that the above verification results meet the preset verification conditions, based on the pre-acquired installation equipment information set, path planning processing is performed on the above glass curtain wall model information to obtain installation control planning path information. This allows for intelligent planning of the construction robot's trajectory. Finally, the above installation control planning path information is sent to the target construction robot for glass curtain wall installation. This allows for control of the glass curtain wall installation. Therefore, some of the installation simulation and control methods applied to glass curtain walls in this application can acquire images of the glass curtain wall installation scene using a single camera, and model the installation scene based on the acquired images. This reduces the types of sensors and installation time, thereby reducing the time for site calibration and inspection before glass curtain wall installation, and ultimately improving the installation efficiency of the glass curtain wall.
[0081] Further reference Figure 3 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of an installation simulation and control device applied to glass curtain walls. These device embodiments are similar to... Figure 2 Corresponding to the method embodiments shown, the installation simulation and control device applied to glass curtain walls can be specifically applied to various electronic devices.
[0082] like Figure 3As shown, an installation simulation and control device 300 for glass curtain walls in some embodiments includes: a partitioning unit 301, a modeling unit 302, a verification unit 303, a path planning unit 304, and a sending unit 305. The partitioning unit 301 is configured to partition each glass curtain wall installation scene image in a pre-acquired glass curtain wall installation scene image set to obtain a set of unit point cloud information. The modeling unit 302 is configured to model the unit point cloud information set to obtain installation scene model information. The verification unit 303 is configured to verify the installation scene model information based on the pre-acquired glass curtain wall model information to obtain a verification result. The path planning unit 304 is configured to, in response to determining that the verification result meets preset verification conditions, perform path planning on the glass curtain wall model information based on a pre-acquired installation equipment information set to obtain installation control planning path information. The sending unit 305 is configured to send the installation control planning path information to a target building robot for glass curtain wall installation.
[0083] It is understandable that the units and references described in the installation simulation and control device 300 applied to glass curtain walls... Figure 2 The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method also apply to the installation simulation and control device 300 and its constituent units applied to glass curtain walls, and will not be repeated here.
[0084] This application also provides a computer device 400. For example... Figure 4 As shown, the computer device 400 includes a bus 401, a processor 402, a memory 403, and a communication interface 404. The processor 402, the memory 403, and the communication interface 404 communicate with each other via the bus 401. The computer device 400 can be a server or a terminal device. It should be understood that this application does not limit the number of processors and memories in the computer device 400.
[0085] Bus 401 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 4 The bus 401 may be represented by a single line, but this does not mean that there is only one bus or one type of bus. The bus 401 may include a path for transmitting information between various components of the computer device 400 (e.g., memory 403, processor 402, communication interface 404).
[0086] Processor 402 may include any one or more processors such as a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).
[0087] Memory 403 may include volatile memory, such as random access memory (RAM). Memory 403 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).
[0088] The memory 403 stores executable program code, and the processor 402 executes this executable program code to implement the functions of the aforementioned partitioning unit, modeling unit, verification unit, path planning unit, and sending unit, thereby realizing the aforementioned installation simulation and control method applied to glass curtain walls. That is, the memory 403 stores instructions for executing the aforementioned installation simulation and control method applied to glass curtain walls.
[0089] The communication interface 404 uses transceiver modules such as, but not limited to, network interface cards and transceivers to enable communication between the computer device 400 and other devices or communication networks.
[0090] This application also provides a chip, which includes a processor and a data interface. The processor reads instructions stored in the memory through the data interface to execute the above-described installation simulation and control method applied to glass curtain walls.
[0091] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium capable of being stored by a computing device, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct the computing device to execute the aforementioned installation simulation and control method applied to glass curtain walls.
[0092] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0093] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 protection scope of the technical solutions of the embodiments of this application.
Claims
1. An installation simulation and control method for glass curtain walls, comprising: The glass curtain wall installation scene images in the pre-acquired glass curtain wall installation scene image set are partitioned to obtain a set of unit point cloud information. The unit point cloud information set is modeled to obtain installation scene model information; Based on the pre-acquired glass curtain wall model information, the installation scene model information is verified to obtain the verification result; In response to determining that the verification result meets the preset verification conditions, based on the pre-acquired set of installation equipment information, the glass curtain wall model information is processed for path planning to obtain installation control planning path information, wherein the installation equipment information in the set of installation equipment information includes: initial task path information; The installation control planning path information is sent to the target building robot for glass curtain wall installation; The step of performing path planning processing on the glass curtain wall model information based on a pre-acquired set of installation equipment information to obtain installation control planning path information includes: The glass curtain wall model information is processed to extract target points, resulting in an information set of installation target point pairs. Obstacle extraction processing is performed on the glass curtain wall model information to obtain a scene obstacle information set; Each piece of installation equipment information in the installation equipment information set is modeled to generate installation equipment motion model information, thus obtaining the installation equipment motion model information set; Based on the scene obstacle information set, the motion model information of each installation equipment in the installation equipment motion model information set is processed to generate initial path planning information, thus obtaining the initial path planning information set. The initial path planning information set is optimized to obtain installation control planning path information. For each initial path planning information in the initial path planning information set, in response to determining that the initial path planning information has no intersection with any other initial path planning information in the initial path planning information set, the initial path planning information set is determined as installation control planning path information. The step of generating initial path planning information by planning each installation equipment motion model information in the installation equipment motion model information set based on the scene obstacle information set includes: Based on the scene obstacle information set, the motion model information of the installed equipment is processed for task unloading to obtain unloading energy consumption information. Based on the unloading energy consumption information, a delay determination model is determined; Based on the aforementioned delay determination model, the device energy consumption information is determined; The energy consumption information of the equipment is optimized to obtain initial path planning information.
2. The installation simulation and control method for glass curtain walls according to claim 1, characterized in that, The step involves partitioning each glass curtain wall installation scene image in the pre-acquired glass curtain wall installation scene image set to obtain a set of unit point cloud information, including: The glass curtain wall installation scene images in the pre-acquired glass curtain wall installation scene image set are restored to obtain a sparse point cloud information set of the installation scene. The sparse point cloud information of each installation scenario in the sparse point cloud information set of the installation scenario is split and processed to obtain the sparse point cloud information set of the installation scenario. Each sparse point cloud information group in the installation scenario sparse point cloud information group set is expanded to generate a unit point cloud information group, thus obtaining a unit point cloud information group set.
3. The installation simulation and control method for glass curtain walls according to claim 1, characterized in that, The process of modeling the unit point cloud information set to obtain installation scene model information includes: Gaussian transformation is performed on each unit point cloud information in the unit point cloud information set to generate unit point cloud Gaussian information, thus obtaining the unit point cloud Gaussian information set. Each Gaussian information group of the unit point cloud Gaussian information group set is input into a pre-trained Gaussian decoupling network model to generate scene partition rendering images, thus obtaining a scene partition rendering image set. The scene partition rendering maps in the scene partition rendering map set are stitched together to obtain the installation scene model information.
4. The installation simulation and control method for glass curtain walls according to claim 3, characterized in that, Before inputting each Gaussian information group of the unit point cloud Gaussian information group set into a pre-trained Gaussian decoupling network model to generate a scene partition rendering image and obtain a scene partition rendering image set, the method further includes: Obtain a sample scene image set, wherein the sample scene images in the sample scene image set include: a sample pixel information set; Select target sample scene images from the sample scene image set and perform the following training steps: Gaussian transformation is performed on each sample pixel information in the sample pixel information set included in the target sample scene image to generate initial pixel Gaussian information, thus obtaining the initial pixel Gaussian information set; The initial Gaussian information set of the initial pixels is input into the initial convolutional sub-model included in the initial Gaussian decoupling network model to obtain the initial convolutional image. The initial Gaussian decoupling network model also includes a downsampling sub-model and a target convolutional sub-model. The initial convolutional image is input into the downsampling sub-model included in the initial Gaussian decoupled network model to obtain the initial downsampling image; The initial downsampled image is input into the target convolutional sub-model included in the initial Gaussian decoupled network model to obtain the initial rendered image; Based on the preset decoupling loss function, the decoupling loss value corresponding to the target sample scene image and the initial rendered image is determined; In response to the determination that the decoupling loss value is less than the target threshold, the initial Gaussian decoupling network model is determined to be a Gaussian decoupling network model.
5. The installation simulation and control method for glass curtain walls according to claim 4, characterized in that, The method further includes: In response to determining that the decoupling loss value is greater than or equal to the target threshold, the relevant parameters in the initial Gaussian decoupling network model are adjusted, and the adjusted initial Gaussian decoupling network model is determined as the initial Gaussian decoupling network model. Target sample scene images are selected from the sample scene image set, from the unselected sample scene images, for the training step to be performed again.
6. An installation simulation and control device for glass curtain walls, characterized in that, include: The partitioning unit is configured to partition each glass curtain wall installation scene image in a pre-acquired glass curtain wall installation scene image set to obtain a set of unit point cloud information. The modeling unit is configured to perform modeling processing on the set of unit point cloud information to obtain installation scene model information; The verification unit is configured to perform verification processing on the installation scene model information based on the pre-acquired glass curtain wall model information, and obtain the verification result; The path planning unit is configured to, in response to determining that the verification result meets the preset verification conditions, perform path planning processing on the glass curtain wall model information based on the pre-acquired installation equipment information set to obtain installation control planning path information, wherein the installation equipment information in the installation equipment information set includes: initial task path information; The sending unit is configured to send the installation control planning path information to the target building robot for glass curtain wall installation; The step of performing path planning processing on the glass curtain wall model information based on a pre-acquired set of installation equipment information to obtain installation control planning path information includes: The glass curtain wall model information is processed to extract target points, resulting in an information set of installation target point pairs. Obstacle extraction processing is performed on the glass curtain wall model information to obtain a scene obstacle information set; Each piece of installation equipment information in the installation equipment information set is modeled to generate installation equipment motion model information, thus obtaining the installation equipment motion model information set; Based on the scene obstacle information set, the motion model information of each installation equipment in the installation equipment motion model information set is processed to generate initial path planning information, thus obtaining the initial path planning information set. The initial path planning information set is optimized to obtain installation control planning path information. For each initial path planning information in the initial path planning information set, in response to determining that the initial path planning information has no intersection with any other initial path planning information in the initial path planning information set, the initial path planning information set is determined as installation control planning path information. The step of generating initial path planning information by planning each installation equipment motion model information in the installation equipment motion model information set based on the scene obstacle information set includes: Based on the scene obstacle information set, the motion model information of the installed equipment is processed for task unloading to obtain unloading energy consumption information. Based on the unloading energy consumption information, a delay determination model is determined; Based on the aforementioned delay determination model, the device energy consumption information is determined; The energy consumption information of the equipment is optimized to obtain initial path planning information.
7. An electronic device, characterized in that, include: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 5.
8. A computer-readable medium, characterized in that, It stores a computer program thereon, characterized in that the computer program, when executed by a processor, implements the method as described in any one of claims 1 to 5.
Citation Information
Patent Citations
3D modeling reconstruction system, method and device based on point cloud information and Gaussian cloud cluster
CN118196306A
Curtain wall installation control system and method of mechanical arm based on path planning
CN119567279A
Steel structure curved surface curtain wall installation method based on reverse modeling technology
CN121030849A
