Building robot intelligent selection and allocation scheduling method and system based on multi-machine linkage

Abstract

The invention provides a building robot intelligent selection and allocation scheduling method and system based on multi-machine linkage. The system comprises a plurality of robots and a cloud. Comprising the following steps: inputting an architectural drawing set to a cloud end, and updating a three-dimensional construction scene set and a total coordinate system to all robots of a construction site; the robot scans a fixed obstacle in an environment recognition space and sends the fixed obstacle to the cloud, and the updated three-dimensional construction scene is sent to other robots; the robot updates the process stage label through the process stage model, and the cloud selects the robot to go to the construction coordinates for construction tasks. By adopting the method, the three-dimensional construction scene set is constructed through the cloud, the unified total coordinate system is defined, and all robots are ensured to work under the same coordinate system. And the robots can share the position of the fixed obstacle in real time. The cloud can flexibly schedule the robots according to the current construction progress and requirements, and intelligently select the most suitable robot for operation.

Description

Technical Field

[0001] The present invention relates to the field of construction robots, and in particular to a method and system for intelligent selection and scheduling of construction robots based on multi-machine linkage. Background Art

[0002] As the construction industry evolves toward intelligent and digital development, the use of construction robots on construction sites is becoming increasingly widespread. Currently, construction robots are gradually being applied to various construction tasks, such as masonry, plastering, spray painting, and handling, improving construction efficiency and reducing manual labor intensity to a certain extent. However, many problems still exist in multi-robot collaborative operations. Due to the complex construction site environment and changing process flows, the lack of a unified task scheduling platform between different types of robots makes it difficult to achieve information exchange and efficient collaboration, resulting in waste of resources, task conflicts, and even frequent safety hazards.

[0003] In actual construction projects, task allocation for construction robots often relies on manual on-site scheduling based on experience. This is not only inefficient but also difficult to adapt to the ever-changing process stages and site conditions during construction. Some systems attempt to remotely manage robots through a centralized control platform. However, when multiple robots are working simultaneously, it is often impossible to establish a unified spatial coordinate system based on construction drawings and the on-site environment, resulting in frequent problems such as robot positioning deviations and path conflicts.

[0004] Furthermore, most current construction robotic systems only perform static path planning based on architectural blueprints before construction begins. They lack real-time recognition and feedback mechanisms for on-site obstacles, which can easily lead to obstacle avoidance failures or task interruptions during construction. Furthermore, existing systems lack effective automatic recognition and response capabilities for dynamic updates during the construction phase, such as process adjustments and changes to the construction sequence. These systems still require extensive manual intervention, limiting the intelligence and adaptability of construction robotic systems.

[0005] Therefore, how to achieve task collaboration and intelligent scheduling among construction robots in complex construction environments has become a key technical issue that needs to be urgently solved in the field of construction robots. Summary of the Invention

[0006] In order to overcome the existing technical problems, the present invention provides a method and system for intelligent selection and scheduling of construction robots based on multi-machine linkage.

[0007] The present invention adopts the following technical solutions.

[0008] A method for intelligent selection and scheduling of construction robots based on multi-machine linkage, comprising multiple robots with camera modules and a cloud connected to the signals of the multiple robots;

[0009] The following steps are involved:

[0010] S1. Input a set of architectural drawings to the cloud. The cloud constructs a 3D construction scene set based on the architectural drawings and defines a global coordinate system. The 3D construction scene set and the global coordinate system are then updated to all robots on the construction site.

[0011] S2: The robot scans the environment, identifies fixed obstacles in the space, and sends them to the cloud. The cloud updates and integrates them into the 3D construction scene set, and updates the 3D construction scene to other robots.

[0012] S3. The process stage labels are updated manually or the robot updates the process stage labels through the process stage model. The cloud selects the robot to go to the construction coordinates to perform the construction task based on the process stage labels and robot data.

[0013] As a further improvement of the present invention, each of the robots has a status tag synchronized with the cloud, and the status tag includes a construction status, a charging status, and an idle status;

[0014] If the robot has a construction task, it is marked as construction status; if it has no construction task or charging task, it is marked as idle status;

[0015] When the battery level of a robot marked as idle is lower than a preset charging threshold, a charging task is executed to switch the idle state to the charging state.

[0016] The robot is also preset with a limit power threshold that is lower than the charging threshold. When the power of the robot marked as the construction state is lower than the limit power threshold, the charging task is performed and the construction state is marked as the charging state.

[0017] As a further improvement of the present invention, each process stage label includes a plurality of construction steps, and the robots are preset with a construction step set including at least one construction step;

[0018] The step S3 further includes a step S4:

[0019] The robot has a display module, and the display module displays the current construction step of the robot;

[0020] When there are at least two robots performing the same construction step, each robot scans the environment through a camera module and performs posture correction to obtain an environment scan image. The environment scan image is then compared with the step scene image corresponding to the previous construction step in the 3D construction scene set for similarity calculation;

[0021] If the similarity is greater than the construction scene threshold, the robot is selected and all selected robots are integrated to form a pre-selection set;

[0022] If the similarities are all less than the construction scene threshold, each robot will move around a circle with itself as the origin and the preset election scene value as the radius. During the movement, it will stop multiple times to scan the environment and correct its posture to obtain a 3D adjacent environment scan map. The 3D adjacent environment scan map is then compared with the step scene map corresponding to the previous construction step for similarity calculation. The robot with the highest similarity value is selected as the master robot. The master robot will scan the environment in real time to construct a 3D adjacent environment scan map.

[0023] If there is a pre-selected election set, the collaborative election score W is calculated for each robot in it. i ,Select the robot with the highest collaborative election score as the master robot, and record the coordinates of the robot when scanning the environment as the supervision coordinates;

[0024]

[0025] Among them, Q i is the battery percentage of the i-th robot, α1 is the battery weight coefficient, is the sum of the distances between the i-th robot and the j robots in the same construction task, α2 is the weight coefficient of the average transmission distance, and E i is the construction efficiency of the i-th robot, α3 is the construction efficiency weight coefficient, H i is the average construction amount of the i-th robot, Η a is the average construction volume of robots of the same type as the i-th robot, W 极 is the attenuation of construction efficiency in extreme weather conditions, H ie is the average construction volume of the e-th extreme weather type;

[0026] If the current construction step requires at least two robots to execute together, the master robot sends an action synchronization instruction to the robots that execute together.

[0027] As a further improvement of the present invention, the specific steps of updating the process stage label by using the process stage model include:

[0028] The progress of the construction step is analyzed based on the environment scan map or the 3D adjacent environment scan map and the step scene map corresponding to the construction step in the 3D construction scene set. When the progress of the construction step exceeds a preset step threshold, the master robot broadcasts to the robot of the next construction step;

[0029] If the current construction step is the last construction step of the process stage, the comparison between the progress of the construction step and the step threshold is stopped. When the progress of the construction step is greater than the preset stage threshold, the master robot sends the process stage tag marked as ready to the cloud. The cloud selects the robot marked as idle based on the process stage tag marked as ready and the robot data, and switches the robot's idle state to ready.

[0030] When the battery level of a robot in the ready state drops below the preset ready power threshold, it remains in the ready state and is marked as charging to perform the charging task.

[0031] When the progress of the last construction step of a process stage is greater than a preset stage completion threshold, the master robot sends an environmental scan or a 3D adjacent environmental scan, as well as the next process stage label to the cloud. The cloud determines whether the progress of the step is greater than the stage completion threshold based on the environmental scan or the 3D adjacent environmental scan and the step scene diagram corresponding to the construction step in the 3D construction scene. If so, the process stage label is updated.

[0032] Based on the environmental scan or the 3D adjacent environmental scan, and the step scene diagram corresponding to the construction step in the 3D construction scene set, the specific steps for analyzing and deriving the progress of the construction step include:

[0033] The master robot goes to the supervisory coordinates according to the preset supervisory interval to scan the environment and obtains the environment scan map through posture correction, or the master robot performs real-time environment scanning to construct a 3D adjacent environment scan map. Both the environment scan map and the 3D adjacent environment scan map are point cloud data. The step scene map corresponding to the construction step has a construction component in the form of point cloud data. The progress of the construction step P ij ;

[0034] P m，graph =(X m ±X 误 , Y m ±Y 误 , Z m ±Z 误 ),

[0035] Among them, P ijg is the nth three-dimensional coordinate of the construction component in the jth construction step in the i-th process stage, P m，graph is the error range of the mth three-dimensional coordinate in the environmental scan or the three-dimensional adjacent environmental scan, in millimeters, num(P ijn ∈P m，graph ) is to calculate how many three-dimensional coordinates of the construction components fall within the error range of the three-dimensional coordinates of the environmental scan map or the three-dimensional adjacent environmental scan map, P 总ijIt is the total amount of point cloud data of the construction components in the jth construction step in the i-th process stage.

[0036] As a further improvement of the present invention, the specific steps of selecting a robot based on the process stage label and robot data in the cloud include:

[0037] If the robot's construction step set contains the same construction step as that in the process stage label, and the robot is not in the charging state, the robot is selected, and multiple pre-selected construction sets are formed based on each construction step in the process stage label.

[0038] Each construction step is preset with a standard number of construction machines. If there is a robot in the preparation state in this construction step, the robot is selected and the preparation state is switched to the construction state. If the robot is also marked as charging, it is switched to the construction state after the charging task is completed. The number of robots required for each construction step is N. z ;

[0039] N z =P 标 ·S 目 -N ready , z∈[1, 2, …, j],

[0040] Among them, P 标 is the number of robots required per thousand square meters, S 目 is the construction size of this step, N ready is the number of robots in the preparation state, j is the number of construction steps;

[0041] If the number of robots required for a construction step is greater than the number of robots in the corresponding pre-selected construction set, all robots in the pre-selected construction set are marked as in the construction state;

[0042] If the number of robots required for a construction step is less than the number of robots in the corresponding pre-selected construction set, calculate the construction score M of each robot in the pre-selected construction set. i ;

[0043]

[0044] Among them, L0 is the standard moving length, L i is the path length of the i-th robot to the construction area, Q i is the battery percentage of the i-th robot, N max is the total number of construction steps in this process stage, N i is the number of the i-th robot selected by the pre-selected construction set, γ1, γ2 and γ3 are the selection score weight factors;

[0045] According to the optional construction score M iSelect robots from largest to smallest and mark them as construction status.

[0046] As a further improvement of the present invention, the specific steps of the robot going to the construction coordinate to perform the construction task include: the three-dimensional construction scene set includes a step scene graph of each construction step in each process stage and a real-time three-dimensional scene graph, the step scene graph having a construction area;

[0047] Multiple coordinates in the step scene graph are randomly selected as construction coordinates, and the construction coordinates are sent to the robot one by one. The robot performs Monte Carlo path planning based on the real-time 3D scene graph and the construction coordinates. During the process of moving to the construction coordinates, the robot scans the environment in real time and performs Monte Carlo path planning to avoid obstacles.

[0048] If there are at least two robots in the same construction step, and all robots establish signal connections with each other when they reach the construction coordinates, each robot performs an environmental scan to identify whether there are construction components for that construction step in the environmental scan map. If so, the construction component area is set as the material area and synchronized to other robots in the same construction step. If not, all robots move randomly and re-scan the environment. When there are material areas and construction areas, each robot performs construction tasks according to the preset program.

[0049] As a further improvement of the present invention, the steps of constructing a 3D construction scene set based on a set of architectural drawings and defining a global coordinate system in the cloud specifically include:

[0050] Using NLP natural language processing technology, the names of the drawings in the architectural drawing collection are identified and classified to obtain the plot master plan, individual building design drawings, and building construction drawings. The drawings in the individual building design drawings and building construction drawings are arranged according to the drawing numbers and imported into the BIM software to generate a 3D construction scene set as point cloud data. The southwest corner of the plot master plan is selected as the origin to generate the overall coordinate system.

[0051] The 3D construction scene set includes step scene diagrams of each construction step in each process stage and real-time 3D scene diagrams.

[0052] As a further improvement of the present invention, the robot scans the environment to identify fixed obstacles in the space and sends them to the cloud, which then updates and integrates them into the 3D construction scene set. The specific steps of updating the 3D construction scene to other robots include:

[0053] The robot has a preset obstacle detection interval. When the robot scans the environment and identifies the existence of an obstacle in a certain space, it will continuously scan the environment. If the obstacle moves within the obstacle detection interval, it is considered to be a dynamic obstacle and the continuous scanning of the environment will be stopped. If the obstacle remains stationary for more than the obstacle detection interval, it is considered to be a fixed obstacle. The robot will take multiple views of the fixed obstacle and send them to the cloud. The cloud will generate 3D reconstructed point cloud data through COLMAP and integrate and update it into the 3D construction scene set.

[0054] As a further improvement of the present invention, when the power level of the robot marked as construction state is lower than the limit power threshold, the construction state is switched to charging state and synchronized with the main robot;

[0055] The main robot calculates the power P required for the remaining tasks r ;

[0056]

[0057] Among them, P start is the average power of each robot of the same type as the robot switched to charging state at the beginning of the construction step, P out is the average power of each robot of the same type when the robot switches to charging state and exits the construction task, N 真 is the number of robots of the same type as the robot switched to the charging state at the beginning of this construction step;

[0058] Calculate the total available power P of the robot currently in the construction state and of the same type as the robot switched to the charging state 可 ;

[0059]

[0060] Among them, P t is the current power of the t-th robot of the same type, P 极 is the limit power threshold, P 余 Is the remaining power, if the total available power P 可 Less than the power P required for the remaining task r , then select a robot of the same type that is in charging state and not in ready state and send the construction coordinates. After the robot completes the charging task, it will go to the construction coordinates to perform the construction task.

[0061] The present invention also proposes an intelligent selection and scheduling system for construction robots based on multi-machine linkage, which uses the above-mentioned intelligent selection and scheduling method for construction robots based on multi-machine linkage, including the following modules:

[0062] Multiple robots, pre-programmed with construction programs, are used to scan the environment and perform construction tasks;

[0063] The cloud connects to multiple robots and receives architectural drawings, defines the overall coordinate system, and sends construction coordinates and tasks to the robots.

[0064] A 3D construction scene building module, located in the cloud, builds a 3D construction scene set based on a set of architectural drawings or updates the 3D construction scene set based on fixed obstacles identified by the robot;

[0065] The process stage label update module is used to update the process stage labels manually or through the process stage model, so that the cloud can select robots for construction based on the updated process stage labels;

[0066] The path planning module is installed in the robot and plans the path to the construction coordinates based on the robot's global coordinate system and the construction coordinates;

[0067] The task selection module is located in the cloud and selects the robot to perform the construction task based on the process stage label and robot data;

[0068] The status monitoring module is used to detect the operating status of the robot and change the status label of the robot;

[0069] The process stage feedback module is used to feed back the robot's construction results to the cloud and trigger the allocation of tasks for the next process stage.

[0070] The beneficial effects of the present invention are as follows: by constructing a 3D construction scene set and defining a unified global coordinate system through the cloud, all robots are ensured to operate within the same coordinate system, reducing construction deviations caused by human measurement and positioning errors. The positions of fixed obstacles can be shared between robots in real time, and integrated into the 3D construction scene set through cloud updates, allowing each robot to obtain the latest construction site conditions. Process stage labels can be updated manually or automatically through process stage models, allowing the cloud to flexibly dispatch robots based on the current construction progress and needs, intelligently selecting the most suitable robot to work at a specific construction coordinate, maximizing resource utilization and reducing idle time. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0072] Figure 1 It is a schematic flow diagram of the present invention;

[0073] Figure 2This is an example diagram of the robot of the present invention arriving at the construction coordinates. DETAILED DESCRIPTION

[0074] The accompanying drawings are for illustrative purposes only and should not be construed as limiting this patent. In order to better illustrate this embodiment, certain components of the accompanying drawings may be omitted, enlarged, or reduced in size, and do not represent the actual size of the product.

[0075] It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings. The technical solution of the present invention will be further described below in conjunction with the drawings and embodiments.

[0076] Reference Figure 1 and Figure 2 , an intelligent selection and scheduling method for construction robots based on multi-machine linkage, including multiple robots with camera modules and a cloud connected to the signals of the multiple robots;

[0077] The following steps are involved:

[0078] S1. Input a set of architectural drawings to the cloud. The cloud constructs a 3D construction scene set based on the architectural drawings and defines a global coordinate system. The 3D construction scene set and the global coordinate system are then updated to all robots on the construction site.

[0079] As a further improvement of the present invention, the steps of constructing a 3D construction scene set based on a set of architectural drawings and defining a global coordinate system in the cloud specifically include:

[0080] Using NLP natural language processing technology, the names of the drawings in the architectural drawing collection are identified and classified to obtain the plot master plan, individual building design drawings, and building construction drawings. The drawings in the individual building design drawings and building construction drawings are arranged according to the drawing numbers and imported into the BIM software to generate a 3D construction scene set as point cloud data. The southwest corner of the plot master plan is selected as the origin to generate the overall coordinate system.

[0081] Selecting the southwest corner as the origin to generate the total coordinate system can ensure that the coordinates of the robot moving throughout the construction site are positive, reducing the burden of data processing and providing intuitive location information to the operator. Due to the strict mapping standards in China, for some larger construction projects, the construction site will have a plot map containing the design locations of all individual buildings, such as "Plot A-Master Plan". If the project is smaller, there may be no plot map, only a single building design drawing. If some construction projects have multiple individual building design drawings but no plot map, the construction application will not be approved. Therefore, if there is no plot map at this time, the individual building design drawings will be regarded as the plot map at the same time, and the total coordinate system will be generated based on the southwest corner of the individual building design drawing as the origin. A coordinate system for the entire construction site can be established.

[0082] Because each drawing follows a strict naming convention, such as "Building B-1st Floor Plan," "Building B-Standard Floor Plan," and "Building B-Structural Construction Drawing-Foundation Detail," NLP natural language processing technology can be used to extract the drawing names and automatically classify them. Drawings are also numbered, such as "Building B-1st Floor Plan" might be B01, and can be directly imported into BIM software based on the order of the numbers.

[0083] The 3D construction scene set generated as point cloud data in BIM software can be manually modeled in BIM software based on the drawing information. This modeling needs to be gradually built according to the process stages and construction steps, and finally the point cloud data is directly generated by the software. The point cloud data can facilitate subsequent Monte Carlo path planning, building component comparison, and local incremental updates of the real-time 3D scene map in the 3D construction scene set.

[0084] It's important to note that while manual modeling of each process stage and construction step requires sequential, step-by-step modeling, once completed, it can be used directly to guide subsequent automated robotic construction. Furthermore, the robots can operate 24 / 7, significantly improving construction efficiency and safety. This shifts physical labor in hot environments to mental work in the office, and modeling, as an initial step, doesn't delay later construction progress. This significantly reduces construction time, making it highly practical. Since current process stages and construction steps have clear sequential specifications, manual modeling can be performed for each construction step within each process stage, generating a step scenario diagram and labeling the steps. These step scenarios are then automatically sorted according to the pre-set construction step sequence. Of course, since building construction, with the exception of the formwork removal step, is almost entirely incremental, manual modeling builds the next step based on the step scenario diagram of the previous step. Therefore, actual modeling follows the sequence of process stages and construction steps.

[0085] The 3D construction scene set includes step scene diagrams of each construction step in each process stage and real-time 3D scene diagrams.

[0086] S2: The robot scans the environment, identifies fixed obstacles in the space, and sends them to the cloud. The cloud updates and integrates them into the 3D construction scene set, and updates the 3D construction scene to other robots.

[0087] As a further improvement of the present invention, the robot scans the environment to identify fixed obstacles in the space and sends them to the cloud, which then updates and integrates them into the 3D construction scene set. The specific steps of updating the 3D construction scene to other robots include:

[0088] The robot has a preset obstacle detection interval. When the robot scans the environment and identifies the existence of an obstacle in a certain space, it will continuously scan the environment. If the obstacle moves within the obstacle detection interval, it is considered to be a dynamic obstacle and the continuous scanning of the environment will be stopped. If the obstacle remains stationary for more than the obstacle detection interval, it is considered to be a fixed obstacle. The robot will take multiple views of the fixed obstacle and send them to the cloud. The cloud will generate 3D reconstructed point cloud data through COLMAP and integrate and update it into the 3D construction scene set.

[0089] COLMAP is an open source computer vision library that is mainly used to generate three-dimensional reconstructed point clouds from multi-view images. Specifically, in order to facilitate the acquisition of depth information of obstacles and the subsequent acquisition of environmental scans or three-dimensional adjacent environmental scans, the robot's camera module uses an RGB-D depth camera to reduce the computational burden of generating point cloud data and extend the service life of the construction robot. By presetting the obstacle detection interval, it is possible to effectively determine whether the obstacle is temporary or always present. If it is temporary, it will not affect the path pre-planning of other robots and does not need to be uploaded to the cloud, such as when workers or other machines are just passing by. However, if it is always present, it may have a great impact on the actual path of the subsequent robot when it receives the construction coordinates for path pre-planning. For example, a pile of building components placed directly in the middle of the road may cause the robot to take a detour, affecting the optional construction score M. i To ensure the authenticity of the obstacle, the point cloud data of the fixed obstacle needs to be reported to the cloud and updated to other robots. Specifically, except for workers who are taking a break and will stay for a long time without obvious body movement, workers at other times will not remain motionless. Machines will move when in operation, or some structures will move up and down like the fork of a forklift. Therefore, setting the obstacle detection interval to twenty seconds can effectively detect fixed obstacles such as sleeping workers, arranged building components or mechanical devices. Of course, if a robot passes by and performs an environmental scan later and finds that the fixed obstacles in the area have disappeared, it can be updated to the cloud. Specifically, for multi-view shooting, it is possible to directly circle the obstacle and capture RGB images in at least four directions. Of course, due to the complexity of the construction scene, it is often impossible to circle the obstacle. In this case, a telescopic device can be added to the robot's camera module to enable it to capture the obstacle from multiple angles at different heights.

[0090] S3. The process stage labels are updated manually or the robot updates the process stage labels through the process stage model. The cloud selects the robot to go to the construction coordinates to perform the construction task based on the process stage labels and robot data.

[0091] The process stage labels include foundation engineering stage, main structure construction stage, enclosure structure construction stage, decoration and renovation stage, and equipment installation stage;

[0092] The foundation engineering phase includes excavation, piling, pouring the concrete cushion and foundation structure, backfilling, and compaction. The main structure construction phase involves erecting scaffolding and formwork, tying the steel framework, pouring concrete, and removing the formwork and scaffolding. The enclosure construction phase includes exterior wall construction or installation of prefabricated exterior wall panels, roof waterproofing, insulation, and window and door frame installation. The decoration and renovation phase includes interior and exterior wall plastering, painting, flooring, ceiling installation, and partition installation. The equipment installation phase involves installing drainage pipes, electrical wiring, and the HVAC system.

[0093] As a further improvement of the present invention, each of the robots has a status tag synchronized with the cloud, and the status tag includes a construction status, a charging status, and an idle status;

[0094] If the robot has a construction task, it is marked as construction status; if it has no construction task or charging task, it is marked as idle status;

[0095] When the battery level of a robot marked as idle is lower than a preset charging threshold, a charging task is executed to switch the idle state to the charging state.

[0096] The robot is also preset with a limit power threshold that is lower than the charging threshold. When the power of the robot marked as the construction state is lower than the limit power threshold, the charging task is performed and the construction state is marked as the charging state.

[0097] In one specific embodiment of the present invention, the maximum power threshold is between 5% and 15%. This threshold is controlled based on the distance between the robot's construction location and the charging station. If a robot reaches the maximum power threshold and cannot reach the charging station, the maximum power threshold is adjusted higher. The charging threshold is between 70% and 90%, selected based on the size of the construction project to ensure sufficient power for subsequent robot selection. It is important to note that idle robots should not be directly assigned to charging tasks for two reasons. First, since not every robot is suitable for every construction step, there will theoretically be many idle robots. If these robots were directly assigned to charging tasks, a large number of robots would need to be charged after each construction step, resulting in insufficient charging stations and significant congestion on the construction site. Frequent charging could also affect the robot's battery life. Second, idle robots can monitor their surroundings in real time, enabling safety monitoring. In the event of a serious safety incident, video recording can be provided for review, and the environment can be scanned to update the 3D construction scene set.

[0098] By synchronizing the status tags to the cloud, when a robot in the construction state performs a charging task, the cloud can evaluate whether the remaining robots can complete the current construction step.

[0099] As a further improvement of the present invention, each process stage label includes a plurality of construction steps, and the robots are preset with a construction step set including at least one construction step;

[0100] The step S3 further includes a step S4:

[0101] The robot has a display module, and the display module displays the current construction step of the robot;

[0102] When there are at least two robots performing the same construction step, each robot scans the environment through a camera module and performs posture correction to obtain an environment scan image. The environment scan image is then compared with the step scene image corresponding to the previous construction step in the 3D construction scene set for similarity calculation;

[0103] At this point, it is important to note that the step scene graph refers to the effect graph after the construction step is completed. Therefore, at the beginning of construction, it is necessary to calculate the similarity with the step scene graph of the previous construction step to determine whether the current robot is in the correct construction area. The environmental scan is performed using the robot's RGB-D depth camera. After scanning, the pixels of this depth camera can carry depth information, which facilitates the subsequent generation of point cloud data. The robot will mark the current coordinates in the global coordinate system during scanning, so the data after the environmental scan can almost be directly compared with the 3D construction scene set. Of course, the density of the point cloud data after the environmental scan is similar to that of the point cloud data in the 3D construction scene set.

[0104] Posture correction is performed by a first IMU inertial measurement system on the robot base and a second IMU inertial measurement system provided on the camera module. Since posture correction is a common technical means, as described in CN110321754B, it will not be described in detail.

[0105] If the similarity is greater than the construction scene threshold, the robot is selected and all selected robots are integrated to form a pre-selection set;

[0106] If the similarities are all less than the construction scene threshold, each robot will move around a circle with itself as the origin and the preset election scene value as the radius. During the movement, it will stop multiple times to scan the environment and correct its posture to obtain a 3D adjacent environment scan map. The 3D adjacent environment scan map is then compared with the step scene map corresponding to the previous construction step for similarity calculation. The robot with the highest similarity value is selected as the master robot. The master robot will scan the environment in real time to construct a 3D adjacent environment scan map.

[0107] If the similarity is less than the construction scene threshold, it means that all robots in the current construction scene have obstacles blocking the construction target. Figure 2 In the example, robots B and C are both obstructed. Common construction steps, such as floor paving, may have numerous building components (tiles) or pillars blocking their view. Therefore, by circling the robot, the robot's surroundings can be determined to determine the degree of obstruction. The robot with the least obstruction is elected as the master robot. If the master robot's battery level drops below the limit and switches to charging, the next most similar robot becomes the master.

[0108] Specifically, due to errors in similarity calculation and environment scanning, the construction scene threshold can be set to 92%.

[0109] The advantage of using this method to select the master robot is that different types of robots perform different construction tasks in the same construction step, such as a handling robot, a tiling robot, and a caulking robot. In addition, when sending construction coordinates, the cloud uses a random selection method to send the coordinates. Therefore, each robot is blocked by obstacles to a different extent. This method can select the robot with the lowest impact, which can facilitate the calculation of the progress of subsequent construction steps and process stages.

[0110] If there is a pre-selected election set, the collaborative election score W is calculated for each robot in it. i ,Select the robot with the highest collaborative election score as the master robot, and record the coordinates of the robot when scanning the environment as the supervision coordinates;

[0111]

[0112] Among them, Q i is the battery percentage of the i-th robot, α1 is the battery weight coefficient, is the sum of the distances between the i-th robot and the j robots in the same construction task, α2 is the weight coefficient of the average transmission distance, and E i is the construction efficiency of the i-th robot, α3 is the construction efficiency weight coefficient, H i is the average construction amount of the i-th robot, Η a is the average construction volume of robots of the same type as the i-th robot, W 极 is the attenuation of construction efficiency in extreme weather conditions, H ie is the average construction volume of the e-th extreme weather type;

[0113] Since the robots in the pre-selected election set can almost all observe the entire construction object without obstacles, the collaborative election score is mainly determined by the power level, the transmission distance with other robots, and the construction efficiency of the main robot. In particular, the main robot should try to complete the entire construction step. The unit of distance is meter. The extreme weather types include various levels of high temperature warning, various levels of rainstorm warning, various levels of strong wind warning, etc. Calculate the value of the robot's construction efficiency under each extreme weather condition. If there is no construction efficiency value for the extreme weather condition, then H ie Considered equivalent to H i , so that the difference between the two is 0. If there is no data on N extreme weather conditions, then Among them, since the operating steps of each robot can be simply regarded as four steps: grabbing, carrying, placing, and fixing, the construction volume can be calculated by how many times the robot has carried the building components.

[0114] If the current construction step requires at least two robots to perform it together, the master robot sends synchronization commands to the robots. For example, if two robots are required to carry a large building component, the master robot can send synchronization commands to ensure the two robots walk in sync. This prevents the building component from deviating and falling due to asynchronous walking, which could damage the building component, the robots, or even threaten the safety of nearby construction workers.

[0115] As a further improvement of the present invention, the specific steps of updating the process stage label by using the process stage model include:

[0116] The progress of the construction step is analyzed based on the environment scan map or the 3D adjacent environment scan map and the step scene map corresponding to the construction step in the 3D construction scene set. When the progress of the construction step exceeds a preset step threshold, the master robot broadcasts to the robot of the next construction step;

[0117] It should be noted at this time that, since a robot will be selected to go to the construction coordinates when entering a new process stage, as a specific embodiment of the present invention, after the robot goes to the construction coordinates and the current construction step elects a master robot, the master robot will establish a signal connection with all robots in the process stage, and will send standby coordinates and standby instructions to robots that are not in the current construction step, thereby increasing the range of activity of the construction step and saving power for subsequent construction steps. Specifically, the step threshold is 90%, and the step threshold can be modified for different construction steps. For example, the next construction step after window installation is door frame installation. The degree of mutual influence between the two construction steps is low, so the robot of the next construction step can be awakened to enter the construction coordinates in advance when the step threshold is low. For construction steps such as laying floor tiles, if the robot of the next construction step is awakened in advance, it is easy for the ground where the tiles have not yet been laid to be stepped on by the robot.

[0118] If the current construction step is the last construction step of the process stage, the comparison between the progress of the construction step and the step threshold is stopped. When the progress of the construction step is greater than the preset stage threshold, the master robot sends the process stage tag marked as ready to the cloud. The cloud selects the robot marked as idle based on the process stage tag marked as ready and the robot data, and switches the robot's idle state to ready.

[0119] When the battery level of a robot in the ready state drops below the preset ready power threshold, it remains in the ready state and is marked as charging to perform the charging task.

[0120] Since this step is to select the robot in advance and charge it, the value of the threshold in this stage can be determined based on the charging time and the construction time, that is, the robot in the ready state can start construction in the next process stage after being fully charged.

[0121] When the progress of the last construction step of a process stage is greater than a preset stage completion threshold, the master robot sends an environmental scan or a 3D adjacent environmental scan, as well as the next process stage label to the cloud. The cloud determines whether the progress of the step is greater than the stage completion threshold based on the environmental scan or the 3D adjacent environmental scan and the step scene diagram corresponding to the construction step in the 3D construction scene. If so, the process stage label is updated.

[0122] This step is used for secondary confirmation to avoid the main robot making a wrong judgment or someone attacking the cloud, sending the wrong process stage label instruction to the cloud, causing it to update to the next stage label early, which in turn causes the entire process to be disrupted.

[0123] Based on the environmental scan or the 3D adjacent environmental scan, and the step scene diagram corresponding to the construction step in the 3D construction scene set, the specific steps for analyzing and deriving the progress of the construction step include:

[0124] The master robot goes to the supervisory coordinates according to the preset supervisory interval to scan the environment and obtains the environment scan map through posture correction, or the master robot performs real-time environment scanning to construct a 3D adjacent environment scan map. Both the environment scan map and the 3D adjacent environment scan map are point cloud data. The step scene map corresponding to the construction step has a construction component in the form of point cloud data. The progress of the construction step P ij ;

[0125] P m，graph =(X m ±X 误 , Y m ±Y 误 , Z m ±Z 误 ),

[0126] Among them, P ijg is the nth three-dimensional coordinate of the construction component in the jth construction step in the i-th process stage, P m，graph is the error range of the mth three-dimensional coordinate in the environmental scan or the three-dimensional adjacent environmental scan, in millimeters, num(P ijn ∈P m，graph ) is to calculate how many three-dimensional coordinates of the construction components fall within the error range of the three-dimensional coordinates of the environmental scan map or the three-dimensional adjacent environmental scan map, P 总ij It is the total amount of point cloud data of the construction components in the jth construction step in the i-th process stage.

[0127] Since the supervision coordinates are actually random construction coordinates sent to the robot by the cloud, and when the robot performs the preset construction task, the construction task is also executed based on the construction coordinates, so when it performs the construction task, it will not be too far away from the supervision coordinates. Therefore, regularly going to the supervision coordinates for scanning will not cause the robot to detour too long. Figure 2 The reason why the cloud randomly sends construction coordinates to the robots is not only to avoid congestion after the robots arrive at the construction site, but also to ensure that the robots can work together around the entire construction object, rather than all robots working together in one place, which is prone to construction conflicts. Figure 2 When robots a and b in the figure perform construction tasks, their paths to the building components are different, they do not affect each other, and they can be located on both sides of the construction object when performing construction tasks.

[0128] Since the environment scan map and the 3D adjacent environment scan map are generated based on the scan results, errors are prone to occur. If the error range is not set for direct comparison, the comparison will easily fail. For example, in the construction step of laying floor tiles, the floor tiles are building components, and the comparison is to compare whether the height of the ground has changed, and the range of change is similar to the thickness of the floor tiles. For example, in the construction step of installing door frames, the door frames are building components, and the comparison is to see whether there is an object similar to the door frame in the door area, thereby judging the progress of the construction step. The error value X 误 , Y 误 , Z 误 It can be defined according to different construction steps. For example, the main error of floor tiles is thickness, so Z 误 Make corresponding adjustments, but this value cannot be close to the thickness of the floor tile. For example, the door frame has a special shape and the probability of misjudgment is low, so the error value X 误 and Y 误 Both can be set to a larger degree.

[0129] The preset supervision interval can be determined based on the time required for the construction step, or it can be uniformly set to a fixed value, such as ten minutes or half an hour.

[0130] A method for calculating how many three-dimensional coordinates of construction components fall within the error range of the three-dimensional coordinates of the environmental scan map or the three-dimensional adjacent environmental scan map can use a merge elimination method, that is, combining two sets of point cloud data. If they fall within the range, they are retained; if they do not fall within the range, both sets of point cloud data are deleted at the same time, and the number of remaining point cloud data is calculated.

[0131] As a further improvement of the present invention, the specific steps of selecting a robot based on the process stage label and robot data in the cloud include:

[0132] If the robot's construction step set contains the same construction step as that in the process stage label, and the robot is not in the charging state, the robot is selected, and multiple pre-selected construction sets are formed based on each construction step in the process stage label.

[0133] If all robots of a certain type are in the charging state, they will be selected again after a period of time to avoid the selection of robots that have just entered the charging state, and then entering the construction state again without sufficient power. After entering the construction state, the power level will fall below the limit threshold again and need to be charged, which will seriously affect the construction efficiency.

[0134] Each construction step is preset with a standard number of construction machines. If there is a robot in the preparation state in this construction step, the robot is selected and the preparation state is switched to the construction state. If the robot is also marked as charging, it is switched to the construction state after the charging task is completed. The number of robots required for each construction step is N. z ;

[0135] N z =P 标 ·S 目 -N ready , z∈[1, 2, …, j],

[0136] Among them, P 标 is the number of robots required per thousand square meters, S 目 is the construction size of this step, N ready is the number of robots in the preparation state, j is the number of construction steps;

[0137] If the number of robots required for a construction step is greater than the number of robots in the corresponding pre-selected construction set, all robots in the pre-selected construction set are marked as in the construction state;

[0138] At this point, it's important to note that if at least one robot of the same type is included in the preselected construction set and its battery level drops below the maximum charge threshold during construction and needs to be recharged, the number of robots of the same type for that construction step will be zero. Since status tags are synchronized with the cloud, the cloud will reselect robots upon receiving this information. A robot that just completed a charging task may be idle and can be directly assigned to the construction task. Furthermore, if a robot of the same type is still charging, it will be reselected after a certain period of time. This ensures efficient charging and avoids frequent unplugging and recharging, which can affect overall construction efficiency.

[0139] If the number of robots required for a construction step is less than the number of robots in the corresponding pre-selected construction set, calculate the construction score M of each robot in the pre-selected construction set. i ;

[0140]

[0141] Among them, L0 is the standard moving length, L i is the path length of the i-th robot to the construction area, Q i is the battery percentage of the i-th robot, N max is the total number of construction steps in this process stage, N i is the number of the i-th robot selected by the pre-selected construction set, γ1, γ2 and γ3 are the selection score weight factors;

[0142] According to the optional construction score M i Select robots from largest to smallest and mark them as construction status.

[0143] By calculating the optional construction score M i , giving priority to robots with close proximity, high battery life, and the fewest robots selected by the pre-selected construction set. Regarding the number of robots selected by the pre-selected construction set, the principle is that, for example, within a process stage, multiple construction steps can be performed by type A robots, while some can be performed by type B robots. If all these steps are performed by type A robots, the type A robots' battery life is likely insufficient to complete them. Therefore, for each robot in the same preferred construction set, robots with the fewest construction steps should be prioritized. Log processing and combining this with the total number of construction steps in the calculation make the results of this item more robust and reliable.

[0144] The path length for the robot to the construction area is planned based on the robot's current position, the construction coordinate position, and the real-time three-dimensional scene graph. Specifically, Monte Carlo path planning is used. If there are obstacles in the real-time three-dimensional scene graph, the probability of selecting the path is reduced, and the path with the highest probability is selected. A path planning method can be referred to in CN116519005B.

[0145] As a further improvement of the present invention, the specific steps of the robot going to the construction coordinate to perform the construction task include: the three-dimensional construction scene set includes a step scene graph of each construction step in each process stage and a real-time three-dimensional scene graph, the step scene graph having a construction area;

[0146] Multiple coordinates in the step scene graph are randomly selected as construction coordinates, and the construction coordinates are sent to the robot one by one. The robot performs Monte Carlo path planning based on the real-time 3D scene graph and the construction coordinates. During the process of moving to the construction coordinates, the robot scans the environment in real time and performs Monte Carlo path planning to avoid obstacles.

[0147] Randomly selecting construction coordinates is not only to avoid congestion after the robots arrive at the construction site, but also to enable the construction robots to work together around the entire construction object, rather than all the robots gathering in one place to work together, which is prone to construction conflicts.

[0148] If there are at least two robots in the same construction step, and all robots establish signal connections with each other when they reach the construction coordinates, each robot performs an environmental scan to identify whether there are construction components for that construction step in the environmental scan map. If so, the construction component area is set as the material area and synchronized to other robots in the same construction step. If not, all robots move randomly and re-scan the environment. When there are material areas and construction areas, each robot performs construction tasks according to the preset program.

[0149] Establishing a signal connection after reaching the construction coordinates reduces the burden of establishing communications. Specifically, after establishing a signal connection, the robots transmit their own data, such as power level, time, status, and location. Because the robots arrive at the site using random construction coordinates, some may not be able to identify the location of materials. By scanning and communicating with each robot, the location of construction components can be quickly determined, helping the robots to smoothly execute their construction tasks.

[0150] As a further improvement of the present invention, when the power level of the robot marked as construction state is lower than the limit power threshold, the construction state is switched to charging state and synchronized with the main robot;

[0151] The main robot calculates the power P required for the remaining tasks r ;

[0152]

[0153] Among them, P start is the average power of each robot of the same type as the robot switched to charging state at the beginning of the construction step, P out is the average power of each robot of the same type when the robot switches to charging state and exits the construction task, N 真 is the number of robots of the same type as the robot switched to the charging state at the beginning of this construction step;

[0154] Calculate the total available power P of the robot currently in the construction state and of the same type as the robot switched to the charging state 可 ;

[0155]

[0156] Among them, P t is the current power of the t-th robot of the same type, P极 is the limit power threshold, P 余 Is the remaining power, if the total available power P 可 Less than the power P required for the remaining task r , a robot of the same type that is in the charging state and not in the ready state is selected and sent the construction coordinates. After the robot completes the charging task, it will go to the construction coordinates to carry out the construction task. This allows the entire construction step to be completed without interruption.

[0157] The present invention also proposes an intelligent selection and scheduling system for construction robots based on multi-machine linkage, which uses the above-mentioned intelligent selection and scheduling method for construction robots based on multi-machine linkage, including the following modules:

[0158] Multiple robots, pre-programmed with construction programs, are used to scan the environment and perform construction tasks;

[0159] The cloud connects to multiple robots and receives architectural drawings, defines the overall coordinate system, and sends construction coordinates and tasks to the robots.

[0160] A 3D construction scene building module, located in the cloud, builds a 3D construction scene set based on a set of architectural drawings or updates the 3D construction scene set based on fixed obstacles identified by the robot;

[0161] The process stage label update module is used to update the process stage labels manually or through the process stage model, so that the cloud can select robots for construction based on the updated process stage labels;

[0162] The path planning module is installed in the robot and plans the path to the construction coordinates based on the robot's global coordinate system and the construction coordinates;

[0163] The task selection module is located in the cloud and selects the robot to perform the construction task based on the process stage label and robot data;

[0164] The status monitoring module is used to detect the operating status of the robot and change the status label of the robot;

[0165] The process stage feedback module is used to feed back the robot's construction results to the cloud and trigger the allocation of tasks for the next process stage.

[0166] Obviously, the above embodiments of the present invention are merely examples for the purpose of clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the claims of the present invention.

Claims

1. A method for intelligent selection and scheduling of construction robots based on multi-machine linkage, characterized in that: The invention comprises a plurality of robots equipped with camera modules, and a cloud connected to signals of the plurality of robots; The following steps are involved: S1. Input a set of architectural drawings to the cloud. The cloud constructs a 3D construction scene set based on the architectural drawings and defines a global coordinate system. The 3D construction scene set and the global coordinate system are then updated to all robots on the construction site. S2: The robot scans the environment, identifies fixed obstacles in the space, and sends them to the cloud. The cloud updates and integrates them into the 3D construction scene set, and updates the 3D construction scene to other robots. S3. The process stage labels are updated manually or the robot updates the process stage labels through the process stage model. The cloud selects the robot to go to the construction coordinates to perform the construction task based on the process stage labels and robot data.

2. The intelligent selection and scheduling method for construction robots based on multi-machine linkage according to claim 1 is characterized in that: Each of the robots has a status tag synchronized with the cloud, and the status tag includes construction status, charging status and idle status; If the robot has a construction task, it is marked as construction status; if it has no construction task or charging task, it is marked as idle status; When the battery level of a robot marked as idle is lower than a preset charging threshold, a charging task is executed to switch the idle state to the charging state. The robot is also preset with a limit power threshold that is lower than the charging threshold. When the power of the robot marked as the construction state is lower than the limit power threshold, the charging task is performed and the construction state is marked as the charging state.

3. The intelligent selection and scheduling method for construction robots based on multi-machine linkage according to claim 1 is characterized in that: Each process stage label includes a plurality of construction steps, and the robots are preset with a construction step set including at least one construction step; The step S3 further includes a step S4: The robot has a display module, and the display module displays the current construction step of the robot; When there are at least two robots performing the same construction step, each robot scans the environment through a camera module and performs posture correction to obtain an environment scan image. The environment scan image is then compared with the step scene image corresponding to the previous construction step in the 3D construction scene set for similarity calculation; If the similarity is greater than the construction scene threshold, the robot is selected and all selected robots are integrated to form a pre-selection set; If the similarities are all less than the construction scene threshold, each robot will move around a circle with itself as the origin and the preset election scene value as the radius. During the movement, it will stop multiple times to scan the environment and correct its posture to obtain a 3D adjacent environment scan map. The 3D adjacent environment scan map is then compared with the step scene map corresponding to the previous construction step for similarity calculation. The robot with the highest similarity value is selected as the master robot. The master robot will scan the environment in real time to construct a 3D adjacent environment scan map. If there is a pre-selected election set, the collaborative election score W is calculated for each robot in it. i ,Select the robot with the highest collaborative election score as the master robot, and record the coordinates of the robot when scanning the environment as the supervision coordinates; Among them, Q i is the battery percentage of the i-th robot, α1 is the battery weight coefficient, is the sum of the distances between the i-th robot and the j robots in the same construction task, α2 is the weight coefficient of the average transmission distance, and E i is the construction efficiency of the i-th robot, α3 is the construction efficiency weight coefficient, H i is the average construction amount of the i-th robot, Η a is the average construction volume of robots of the same type as the i-th robot, W 极 is the attenuation of construction efficiency in extreme weather conditions, H ie is the average construction volume of the e-th extreme weather type; If the current construction step requires at least two robots to execute together, the master robot sends an action synchronization instruction to the robots that execute together.

4. The intelligent selection and scheduling method for construction robots based on multi-machine linkage according to claim 3 is characterized in that: The specific steps for updating the process stage labels through the process stage model include: The progress of the construction step is analyzed based on the environment scan map or the 3D adjacent environment scan map and the step scene map corresponding to the construction step in the 3D construction scene set. When the progress of the construction step exceeds a preset step threshold, the master robot broadcasts to the robot of the next construction step; If the current construction step is the last construction step of the process stage, the comparison between the progress of the construction step and the step threshold is stopped. When the progress of the construction step is greater than the preset stage threshold, the master robot sends the process stage tag marked as ready to the cloud. The cloud selects the robot marked as idle based on the process stage tag marked as ready and the robot data, and switches the robot's idle state to ready. When the battery level of a robot in the ready state drops below the preset ready power threshold, it remains in the ready state and is marked as charging to perform the charging task. When the progress of the last construction step of a process stage is greater than a preset stage completion threshold, the master robot sends an environmental scan or a 3D adjacent environmental scan, as well as the next process stage label to the cloud. The cloud determines whether the progress of the step is greater than the stage completion threshold based on the environmental scan or the 3D adjacent environmental scan and the step scene diagram corresponding to the construction step in the 3D construction scene. If so, the process stage label is updated. Based on the environmental scan or the 3D adjacent environmental scan, and the step scene diagram corresponding to the construction step in the 3D construction scene set, the specific steps for analyzing and deriving the progress of the construction step include: The master robot goes to the supervisory coordinates according to the preset supervisory interval to scan the environment and obtains the environment scan map through posture correction, or the master robot performs real-time environment scanning to construct a 3D adjacent environment scan map. Both the environment scan map and the 3D adjacent environment scan map are point cloud data. The step scene map corresponding to the construction step has a construction component in the form of point cloud data. The progress of the construction step P ij ; P m，graph =(X m ±X 误 ,Y m ±Y 误 ,Z m ±Z 误 ), Among them, P ijg is the nth three-dimensional coordinate of the construction component in the jth construction step in the i-th process stage, P m，graph is the error range of the mth three-dimensional coordinate in the environmental scan or the three-dimensional adjacent environmental scan, in millimeters, num(P ijn ∈P m，graph ) is to calculate how many three-dimensional coordinates of the construction components fall within the error range of the three-dimensional coordinates of the environmental scan map or the three-dimensional adjacent environmental scan map, P 总ij It is the total amount of point cloud data of the construction components in the jth construction step in the i-th process stage.

5. The intelligent selection and scheduling method for construction robots based on multi-machine linkage according to claim 4 is characterized in that: The specific steps for the cloud to select a robot based on process stage labels and robot data include: If the robot's construction step set contains the same construction step as that in the process stage label, and the robot is not in the charging state, the robot is selected, and multiple pre-selected construction sets are formed based on each construction step in the process stage label. Each construction step is preset with a standard number of construction machines. If there is a robot in the preparation state in this construction step, the robot is selected and the preparation state is switched to the construction state. If the robot is also marked as charging, it is switched to the construction state after the charging task is completed. The number of robots required for each construction step is N. z ; N z =P 标 ·S 目 -N ready ,z∈[1、2、…、j], Among them, P 标 is the number of robots required per thousand square meters, S 目 is the construction size of this step, N ready is the number of robots in the preparation state, j is the number of construction steps; If the number of robots required for a construction step is greater than the number of robots in the corresponding pre-selected construction set, all robots in the pre-selected construction set are marked as in the construction state; If the number of robots required for a construction step is less than the number of robots in the corresponding pre-selected construction set, calculate the construction score M of each robot in the pre-selected construction set. i ; Among them, L0 is the standard moving length, L i is the path length of the i-th robot to the construction area, Q i is the battery percentage of the i-th robot, N max is the total number of construction steps in this process stage, N i is the number of the i-th robot selected by the pre-selected construction set, γ1, γ2 and γ3 are the selection score weight factors; According to the optional construction score M i Select robots from largest to smallest and mark them as construction status.

6. The intelligent selection and scheduling method for construction robots based on multi-machine linkage according to claim 1 is characterized in that: The specific steps of the robot going to the construction coordinate to perform the construction task include: the three-dimensional construction scene set includes a step scene graph of each construction step in each process stage and a real-time three-dimensional scene graph, and the step scene graph has a construction area; Multiple coordinates in the step scene graph are randomly selected as construction coordinates, and the construction coordinates are sent to the robot one by one. The robot performs Monte Carlo path planning based on the real-time 3D scene graph and the construction coordinates. During the process of moving to the construction coordinates, the robot scans the environment in real time and performs Monte Carlo path planning to avoid obstacles. If there are at least two robots in the same construction step, and all robots establish signal connections with each other when they reach the construction coordinates, each robot performs an environmental scan to identify whether there are construction components for that construction step in the environmental scan map. If so, the construction component area is set as the material area and synchronized to other robots in the same construction step. If not, all robots move randomly and re-scan the environment. When there are material areas and construction areas, each robot performs construction tasks according to the preset program.

7. The intelligent selection and scheduling method for construction robots based on multi-machine linkage according to claim 1 is characterized in that: The steps for building a 3D construction scene set based on the architectural drawing set and defining the global coordinate system in the cloud include: Using NLP natural language processing technology, the names of the drawings in the architectural drawing collection are identified and classified to obtain the plot master plan, individual building design drawings, and building construction drawings. The drawings in the individual building design drawings and building construction drawings are arranged according to the drawing numbers and imported into the BIM software to generate a 3D construction scene set as point cloud data. The southwest corner of the plot master plan is selected as the origin to generate the overall coordinate system. The 3D construction scene set includes step scene diagrams of each construction step in each process stage and real-time 3D scene diagrams.

8. The intelligent selection and scheduling method for construction robots based on multi-machine linkage according to claim 1 is characterized in that: The robot scans the environment, identifies fixed obstacles in the space, and sends them to the cloud. The cloud updates and integrates them into the 3D construction scene. The specific steps for updating the 3D construction scene to other robots include: The robot has a preset obstacle detection interval. When the robot scans the environment and identifies the existence of an obstacle in a certain space, it will continuously scan the environment. If the obstacle moves within the obstacle detection interval, it is considered to be a dynamic obstacle and the continuous scanning of the environment will be stopped. If the obstacle remains stationary for more than the obstacle detection interval, it is considered to be a fixed obstacle. The robot will take multiple views of the fixed obstacle and send them to the cloud. The cloud will generate 3D reconstructed point cloud data through COLMAP and integrate and update it into the 3D construction scene set.

9. The method for intelligent selection and scheduling of construction robots based on multi-machine linkage according to claim 5 is characterized in that: When the battery level of the robot marked as construction state falls below the limit battery level threshold, the construction state is switched to charging state and synchronized with the main robot; The main robot calculates the power P required for the remaining tasks r ; Among them, P start is the average power of each robot of the same type as the robot switched to charging state at the beginning of the construction step, P out is the average power of each robot of the same type when the robot switches to charging state and exits the construction task, N 真 is the number of robots of the same type as the robot switched to the charging state at the beginning of this construction step; Calculate the total available power P of the robot currently in the construction state and of the same type as the robot switched to the charging state 可 ; Among them, P t is the current power of the t-th robot of the same type, P 极 is the limit power threshold, P 余 Is the remaining power, if the total available power P 可 Less than the power P required for the remaining task r , then select a robot of the same type that is in charging state and not in ready state and send the construction coordinates. After the robot completes the charging task, it will go to the construction coordinates to perform the construction task.

10. A construction robot intelligent matching and scheduling system based on multi-machine linkage, using a construction robot intelligent matching and scheduling method based on multi-machine linkage as described in any one of claims 1 to 9, characterized in that: Includes the following modules: Multiple robots, pre-programmed with construction programs, are used to scan the environment and perform construction tasks; The cloud connects to multiple robots and receives architectural drawings, defines the overall coordinate system, and sends construction coordinates and tasks to the robots. A 3D construction scene building module, located in the cloud, builds a 3D construction scene set based on a set of architectural drawings or updates the 3D construction scene set based on fixed obstacles identified by the robot; The process stage label update module is used to update the process stage labels manually or through the process stage model, so that the cloud can select robots for construction based on the updated process stage labels; The path planning module is installed in the robot and plans the path to the construction coordinates based on the robot's global coordinate system and the construction coordinates; The task selection module is located in the cloud and selects the robot to perform the construction task based on the process stage label and robot data; The status monitoring module is used to detect the operating status of the robot and change the status label of the robot; The process stage feedback module is used to feed back the robot's construction results to the cloud and trigger the allocation of tasks for the next process stage.

Images