Three-dimensional modeling design method and system based on multi-machine cooperation of construction robots

Abstract

The application proposes a three-dimensional modeling design method based on multi-machine cooperation of construction robots, which comprises: allocating the task and path set assigned to each robot as task allocation information; based on the error correction of the real-time feedback and the task allocation information, a task execution scheme; for the corrected task execution scheme, globally analyzing the execution error of all robots; based on the optimized task execution scheme, constructing a task-robot cooperation matrix, using a cooperation scheduling algorithm to schedule the task-robot cooperation matrix; and based on the adjusted point cloud data and the optimized three-dimensional model, merging into a complete three-dimensional model. The application optimizes the cooperative working mechanism of the robot cluster, introduces a dynamic task allocation and real-time feedback error correction mechanism, overcomes the shortcomings in the existing construction robot technology, and significantly improves the efficiency and modeling accuracy of construction.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of three-dimensional modeling design, and particularly relates to a three-dimensional modeling design method and system based on multi-robot cooperation of building robots. BACKGROUND

[0002] With the increasing demand for construction efficiency, precision, and automation in the construction industry, the application of building robot technology in the field of construction has attracted widespread attention. Especially in three-dimensional modeling and automated construction, building robots have significant advantages. Through autonomous navigation, environmental perception, and operation execution of robots, the precision of construction can be significantly improved, manual intervention can be reduced, and construction processes can be optimized. However, current building robot technology still faces many challenges, especially in multi-robot cooperation, dynamic task allocation, spatial error correction, and robot adaptability to complex environments.

[0003] Existing multi-robot cooperation systems generally rely on pre-set path planning and task allocation mechanisms. Although these systems can achieve a certain degree of automation, they still have some significant shortcomings. First, task allocation often lacks flexibility, and robots can only execute tasks according to predetermined order and path, lacking real-time adjustment and dynamic optimization capabilities when the construction environment changes. For example, when obstacles appear on the construction site, progress lags behind, or construction parameters change, existing systems often cannot adjust tasks or robot paths in a timely manner, resulting in low task execution efficiency and even wasting a lot of time and resources. Second, when building robots perform three-dimensional modeling tasks, they usually rely on high-precision sensors for spatial perception and positioning. However, due to sensor errors during robot movement, environmental interference (such as changes in light, vibration, thermal expansion, etc.), and external object occlusion, positioning errors can easily occur, which will gradually magnify over time, ultimately leading to deviations in modeling results and affecting the final construction precision. Although some advanced error correction algorithms such as filtering algorithms, Kalman filtering, etc. have been proposed and applied in single-machine systems, the error correction effect of these technologies is still limited by system size, cooperation precision, and real-time feedback mechanisms when multiple robots work together, making it difficult to achieve effective error control in a global range. Finally, in the cooperative work of multi-robot clusters, although modern robot scheduling systems can achieve basic task allocation, they lack global optimization strategies across robots and tasks. In complex construction sites, a single robot may be inefficient or conflict due to factors such as location, task progress, and workload, and traditional scheduling systems often lack flexibility in responding to these dynamic changes, making it difficult to achieve efficient cooperation and optimal resource allocation among robots.

[0004] Therefore, in the current building robot technology, there are still a series of problems such as inflexible task allocation, insufficient spatial error correction, low efficiency of robot cooperation, and the like, which limit the automation and intelligent level in the construction process, especially in complex environments, the robot cluster cooperation and the guarantee of three-dimensional modeling accuracy are still a difficult problem to be solved. SUMMARY

[0005] The purpose of the present application is to design a three-dimensional modeling design method and system based on the multi-machine cooperation of building robots, which optimizes the cooperative working mechanism of the robot cluster, introduces a dynamic task allocation and real-time feedback error correction mechanism, overcomes the shortcomings of existing building robot technology, and significantly improves the efficiency and modeling accuracy of building construction.

[0006] To achieve the above purpose, in the first aspect of the present application, a three-dimensional modeling design method based on the multi-machine cooperation of building robots is provided, the method comprising:

[0007] S1, according to the environmental data of the construction site, the robot state information, and the task demand data, dynamic task allocation and path planning are carried out, and the set of tasks and paths allocated to each robot is obtained as task allocation information; wherein the robot state information includes the current position and the target task position;

[0008] In step S1, the multi-objective optimization function robot is used for task allocation, and the hybrid method of dynamic programming and A algorithm is used for path optimization;

[0009] S2, collect real-time feedback of each robot executing tasks, monitor local spatial error of each robot based on real-time feedback and error of task allocation information, use machine learning to calculate path correction amount of the current robot according to the size of the local spatial error and the urgency of the task, and the robot updates the current path according to the correction amount to obtain the corrected task execution scheme; wherein the corrected task execution scheme includes task number, robot number, corrected path cost and updated robot position;

[0010] S3, for the corrected task execution scheme, the execution error of all robots is globally analyzed to obtain a global error, and the task allocation and path planning are dynamically adjusted based on the global error to obtain a task execution scheme after global task rescheduling and path optimization; wherein the optimized task execution scheme includes task number, robot number, optimized path planning and optimized path cost;

[0011] S4, based on the optimized task execution scheme, a task-robot cooperation matrix is constructed, a cooperation scheduling algorithm is used to schedule the task-robot cooperation matrix, and a task allocation scheme after optimization scheduling and an adjusted current position of each robot are determined;

[0012] S5, the original point cloud data is obtained, a preliminary three-dimensional model is generated, and a task execution error when the robot executes the task is obtained, the preliminary original point cloud data is corrected based on the task execution error, and the task allocation scheme is adjusted, to obtain adjusted point cloud data and the position of each robot, and an optimized three-dimensional model, and the adjusted point cloud data and the optimized three-dimensional model are combined into a complete three-dimensional model.

[0013] Further, the multi-objective optimization function dynamically calculates the matching degree of the task and the robot by weighted average to perform task allocation; the mixed method of dynamic programming and A algorithm is used for path optimization, which is represented as:

[0014] The current position of the robot is defined as r j =(x k ,y k ), and the target task position is t i =(x i ,y i ); wherein the goal of path planning is to minimize the path cost C j from r i to t path , and the path cost C path considers the path length, obstacle avoidance and load capacity of the robot, and is represented as:

[0015]

[0016] Wherein (x k ,y k ) is the coordinate of each node in the path; is an indicator function, which indicates whether there is an obstacle at the path node k, wherein the obstacle is 1 and the blank area is 0; λ4 is the weighted factor of obstacle avoidance, and n is the total number of nodes.

[0017] Further, the spatial error monitoring is the local spatial error Δr j between the current position of the robot r i and the task target t j ; then

[0018] An adaptive correction amount is obtained based on the local spatial error Δr j ; wherein the adaptive correction amount is according to the local spatial error Δr j, and the urgency of the task to dynamically adjust; the robot obtains the adaptive correction amount, and updates the current path by a cost function according to the adaptive correction amount to obtain a new path cost; the cost function is an obstacle collision judgment function, indicating whether the robot collides with the obstacle when the path is corrected.

[0019] Further, the global error is a total error of the local space error Δr j ; a global error matrix of the global error is calculated based on the distance between the robot and the task target position, and gives the total error of all tasks.

[0020] Further, the global error is used to dynamically adjust the task allocation and path planning to obtain a task execution scheme after global task rescheduling and path optimization, including:

[0021] obtaining a global error;

[0022] correcting the priority of the task based on the global error to obtain an error correction amount of the task i;

[0023] obtaining a new path cost, combining the error correction amount of the task i, to determine a new task allocation scheme.

[0024] Further, the new path cost is obtained, and the error correction amount of the task i is combined to determine a new task allocation scheme:

[0025] calculated by minimizing the task execution cost, including:

[0026]

[0027] wherein T′ assign is the new task allocation scheme; α j is the adjustment weight of the robot j; C′ psth (j, i) is the path cost of the robot j executing the task i after correction; and γ is a path cost adjustment factor, used to balance the weights of path optimization and error correction.

[0028] Further, the task-robot cooperation matrix represents the cooperation relationship between the task and the robot; an element of the matrix represents the cooperation effect between the task i and the robot j; and the task-robot cooperation matrix is represented as:

[0029]

[0030] wherein α i is the priority weight of the task i, β j is the state adjustment factor of the robot j, d(r′ j , t i) is the error distance of robot j to the target of task i, d max is the maximum tolerance error distance of task i;

[0031] Then, based on the task-robot coordination matrix, the most suitable robot is assigned to each task by using a dynamic scheduling algorithm; a resource adjustment factor is introduced into the dynamic scheduling algorithm to optimize task scheduling, and a dynamic feedback correction factor is introduced to adjust the task allocation of robot j in real time.

[0032] Further, the task execution error E j (t) of the robot when performing the task is represented as:

[0033] E j (t) = λ1∥r′ j (t) - t i ∥ + λ2|ΔT j (t) | + λ3∥ΔL j (t)∥

[0034] Wherein, r′ j (t) is the actual position of robot j at time t; t i is the expected position of the target task i; ΔT j (t) is the time error of robot j in performing the task; ΔL j (t) is the load error; λ1, λ2, λ3 are weighting coefficients;

[0035] Then, based on the task execution error E j (t), the point cloud data is corrected, represented as:

[0036]

[0037] Wherein, P i corrected is the corrected point cloud data, P i preprocessed is the original point cloud data; μ(t) is a dynamic adjustment factor, which adjusts the correction strength based on the resources of the robot; E max is the maximum error allowed in the task;

[0038] According to the corrected point cloud data P i corrected , the original three-dimensional model is weighted and optimized to obtain the optimized three-dimensional model M optimized (t) at the current time;

[0039] The corrected point cloud data P i corrected is fused with the optimized three-dimensional model M optimized (t) at the current time to obtain the final complete three-dimensional model.

[0040] In a second aspect of the present application, a three-dimensional modeling design system based on multi-robot cooperation of construction robots is provided, and the system comprises:

[0041] An environment data acquisition unit is configured to perform dynamic task allocation and path planning according to environment data of a construction site, robot state information, and task demand data, and obtain a set of tasks and paths allocated to each robot as task allocation information, wherein the robot state information comprises a current position and a target task position.

[0042] In the environment data acquisition unit, a multi-objective optimization function robot is used to perform task allocation, and a hybrid method of dynamic programming and A algorithm is used to perform path optimization.

[0043] A task scheduling unit is configured to acquire real-time feedback of each robot performing a task, perform local space error monitoring on each robot based on an error of the real-time feedback and the task allocation information, calculate a path correction amount of a current robot according to a size of the local space error and an urgency of the task using machine learning, and update a current path of the robot according to the correction amount to obtain a corrected task execution scheme, wherein the corrected task execution scheme comprises a task number, a robot number, a corrected path cost, and an updated robot position.

[0044] A global optimization unit is configured to perform global analysis on execution errors of all robots for the corrected task execution scheme to obtain a global error, and perform dynamic adjustment on task allocation and path planning based on the global error to obtain a task execution scheme after global task rescheduling and path optimization, wherein the optimized task execution scheme comprises a task number, a robot number, an optimized path planning, and an optimized path cost.

[0045] A cooperative optimization unit is configured to construct a task-robot cooperation matrix based on the optimized task execution scheme, perform task scheduling on the task-robot cooperation matrix using a cooperative scheduling algorithm, and determine a task allocation scheme after optimization scheduling and an updated current position of each robot.

[0046] A three-dimensional modeling unit is configured to obtain original point cloud data, generate a preliminary three-dimensional model, obtain a task execution error when a robot performs a task, correct the preliminary original point cloud data based on the task execution error and adjust the task allocation scheme, obtain adjusted point cloud data and a position of each robot, and an optimized three-dimensional model, and combine the adjusted point cloud data and the optimized three-dimensional model into a complete three-dimensional model.

[0047] The present application has at least the following beneficial technical effects:

[0048] Task-driven dynamic task allocation mechanism: Traditional robot swarm task allocation mechanisms usually rely on preset paths and fixed task allocation, which can easily lead to robots failing to adapt to changes in real time in environments with large changes in construction sites. To overcome this problem, the present invention proposes a task-driven dynamic task allocation algorithm that dynamically adjusts task allocation and path planning based on real-time construction requirements, task progress, and robot status (including power, work progress, sensor status, etc.). This mechanism can flexibly respond to the complex changes in the construction site, avoid resource waste, and ensure that robots can work efficiently in collaboration.

[0049] Spatial error correction and real-time feedback optimization mechanism: Existing error correction methods mainly rely on self-correction of a single robot, which is difficult to effectively control global errors when multiple robots work together. To address this problem, the present invention introduces a spatial error correction and feedback optimization mechanism to achieve global error correction for robots when performing modeling tasks. Each robot in the process of performing tasks, real-time acquisition of sensor data of the surrounding environment, compared with the predetermined building model, detect and correct spatial error. When a robot has an error, the system will real-time feedback error information to other robots, to adjust the path and task in collaboration, to ensure that the entire robot swarm maintains high modeling accuracy in the global range.

[0050] Global collaboration and resource optimization scheduling: To further improve the efficiency of multi-robot collaboration, the present invention proposes a global optimization scheduling mechanism that can integrate and optimize the scheduling of multiple robot tasks, progress, and resource status to ensure that the collaborative work between robots does not conflict, avoid resource waste, and maximize task execution efficiency. This mechanism calculates the task load, spatial positioning, and working status of each robot in real time, intelligently schedules tasks between robots, ensures that construction site tasks are balancedly distributed to each robot, and adjusts the priority of tasks according to the construction progress to avoid construction delays caused by excessive load or path conflicts of a robot.

[0051] Through these innovations, the present invention can real-time adjust robot tasks and paths in complex construction environments, reduce the impact of spatial errors, and optimize the overall collaboration efficiency of the robot swarm. Ultimately, it can significantly improve the automation, precision, and construction speed of construction, thereby solving the problems of inflexible task allocation, insufficient spatial error correction, and low efficiency of robot collaboration in existing technologies, and promoting the development of the construction industry towards more intelligent and efficient direction. BRIEF DESCRIPTION OF DRAWINGS

[0052] The application is further illustrated by the accompanying drawings, but the embodiments in the drawings do not constitute any limitation to the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of the following drawings.

[0053] Figure 1 A flow chart of a three-dimensional modeling design method based on multi-machine cooperation of construction robots is provided.

[0054] Figure 2 A system framework diagram of a three-dimensional modeling design system based on multi-machine cooperation of construction robots is provided. DETAILED DESCRIPTION

[0055] The embodiments of the application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, in which the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain the application, and cannot be understood as a limitation to the application.

[0056] In one or more embodiments, as shown in Figure 1 A three-dimensional modeling design method based on multi-machine cooperation of construction robots is disclosed, and the method comprises the following steps S1-S5:

[0057] S1, according to the environmental data of the construction site, the robot state information, and the task demand data, dynamic task allocation and path planning are performed, and the set of tasks and paths allocated to each robot is obtained as task allocation information; wherein the robot state information includes the current position and the target task position.

[0058] Specifically, the input data of step S1 comes from the overall environmental perception and task allocation module of the system, and these data contain the following aspects:

[0059] Environmental data: real-time data acquisition system from the construction site, including structure diagram, obstacle distribution, construction progress, etc. These data are collected in real time by on-site sensors, laser scanners, visual sensors, etc. and are processed by data fusion algorithms.

[0060] Robot state data: real-time state data of each robot, including power, load, current position, available tools, task priority, etc. These information are obtained by sensors and internal computing modules on the robot and transmitted to the control system.

[0061] Task requirement data: Based on the construction plan and building progress, the system dynamically generates specific requirements for each task, including the timeliness, priority, and required resources of the task. These task requirements are generated by the scheduling system and combined with the robot's state information to provide a priority ranking of tasks.

[0062] Further, based on the resource status of the robot cluster and environmental conditions, dynamically allocate the tasks that each robot needs to complete at a specific time. Consider multiple factors such as task priority, robot idle resources, and current task load of the robot.

[0063] Among them, the task allocation strategy:

[0064] Set a multi-objective optimization function for balancing the objectives of task allocation. This objective function includes but is not limited to task priority, spatial location of the task, and robot load conditions.

[0065] Define the priority function of the task as P i , where i represents the number of tasks, P i represents the importance or timeliness of the task.

[0066] Define the weight function of the robot state as W j , where j represents the number of robots, W j represents the current idle resources of the robot, and W j is evaluated based on the robot's power, load, and task history.

[0067] Based on these inputs, the matching degree between tasks and robots is dynamically calculated through weighted averaging:

[0068]

[0069] where P i is the priority of task i; W | is the current resource weight of robot j; d(i,j) is the distance between task i and robot j (which can be the Euclidean distance between the task location and the current location of the robot); λ1, λ2, λ3 are the weighting factors for adjusting the priority of the task, the resources of the robot, and the distance.

[0070] The optimization objective function represents the comprehensive priority of task allocation, and each robot will match with the task according to the value of this function, so as to efficiently allocate tasks at each time.

[0071] Further, for each task, the system needs to plan an optimal path for the robot from its current location to the target task location, ensuring the path avoids obstacles and takes into account the length of the path and the workload of the robot.

[0072] Path planning method: a hybrid method of dynamic programming (DP) and A algorithm is used for path optimization. Each robot, based on its current task and location, uses A algorithm for preliminary path planning, and combines dynamic programming to fine-tune the path to avoid congestion areas and obstacles.

[0073] Define the current location of the robot as r j =(x j ,y j ), and the target task location as t i =(x i ,y i ). The goal of path planning is to minimize the path cost C j from r i to t path , which takes into account the path length, obstacle avoidance, and the load capacity of the robot:

[0074]

[0075] where (x k ,y k ) is the coordinate of each node in the path; is an indicator function that indicates whether there is an obstacle at path node k (obstacle is 1, blank area is 0); λ4 is the weighted factor of obstacle avoidance.

[0076] The optimization goal of this path cost function is to minimize the path cost from the current location of the robot to the target task location, ensuring that the robot can efficiently and safely reach the target location.

[0077] Finally, the output of step S1 is the specific task assigned to each robot and the corresponding path planning. These information will be used as input for step S2, providing the task execution framework and path planning data for the robot.

[0078] where each task i is assigned to robot j, and robot j will move from its current location r j to the target task location t i according to the calculated path.

[0079] The output data format is a set of tasks and paths

[0080]

[0081] where i represents the task number; j represents the robot number assigned to the task; r j and t i are the current position and target task position of robot j, respectively; C path (j, i) is the path cost from robot j to task i.

[0082] In the entire scheme, the core role of step S1 is to perform dynamic task allocation and path planning according to existing environmental data, robot state and task requirements. Through this process, the system can optimize the work efficiency of the robot cluster and the priority of task execution, providing accurate task and path information for subsequent steps. The output of this step (task allocation and path planning results) will be used as the input of step 2 for further task execution and spatial error correction.

[0083] S2, collect real-time feedback of each robot executing the task, monitor local spatial error of each robot based on real-time feedback and task allocation information error, use machine learning to calculate the path correction amount of the current robot according to the size of the local spatial error and the urgency of the task, and update the current path according to the correction amount to obtain the corrected task execution scheme; wherein the corrected task execution scheme includes task number, robot number, corrected path cost and updated robot position.

[0084] Specifically, the input of step 2 comes directly from the output of step 1 - task allocation and path planning results. Specifically, the input data includes:

[0085] Task allocation information Target position t i of each task, corresponding robot number j, task priority and path planning cost C path (j, i), etc.

[0086] Robot state information r j : current position, load, power, available tools, etc. of each robot, and real-time state in task execution.

[0087] These information will be passed as input to the error monitoring and feedback module, the goal is to monitor the spatial error according to the execution state of the task, the real-time position of the robot, and provide necessary feedback adjustment for task execution.

[0088] Further, error monitoring: real-time spatial error calculation and environmental disturbance assessment:

[0089] Through the fusion of multi-sensor data in the system, the current position r j of the robot and the target t iThe spatial error between the current position of the robot and the target position of the task is defined as an error vector:

[0090] Δr j = r j -t i (4)

[0091] wherein r j is the current position of the robot j; t i is the target position of the task i; and Δr j is the error vector between the current position of the robot and the target position of the task.

[0092] In addition to the basic error calculation, the influence of dynamic changes in the building environment (such as obstacles, the behavior of other robots, etc.) on the error needs to be considered. The present invention designs an environmental interference factor γ j to dynamically correct the error value, as follows:

[0093]

[0094] wherein γ j is a dynamic adjustment factor, reflecting the influence of the urgency of the robot's task and the workload; is a complex nonlinear function for the current position of the robot, the target position, and the distribution of environmental obstacles , used to estimate the influence of the environment on the error. The adjusted error provides a more accurate measurement of the spatial error, taking into account the influence of dynamic changes in the environment, and is the basis for error feedback and path correction.

[0095] Further, on the basis of real-time error monitoring, the present invention proposes a feedback mechanism based on machine learning adaptive correction, which can dynamically adjust the path correction amount according to the size of the error and the urgency of the task. In traditional methods, the error correction amount is usually static, but in the present invention, an adaptive correction amount c j is introduced, allowing the feedback mechanism to flexibly adjust with changes in the environment and the task. It is defined as:

[0096]

[0097] wherein α is the weighting coefficient of error correction, adjusting the strength of error correction; β is the weight factor of task urgency; is the urgency indicator function of the task (1 for urgent, 0 for ordinary tasks); λ j is the weight coefficient adjusting the path correction amount and the error correction; ΔC path (j, i) is the change in path cost, representing the change in cost from the original path to the corrected path. By adding the path cost change term ΔC path(j,i), this scheme can dynamically balance task priority and the cost of path correction, ensuring that path correction does not lead to excessive efficiency loss. Adaptive correction amount c j This is the feedback quantity for the robot's current path adjustment. By applying this quantity, the robot's path adjustment can better balance the relationship between task urgency, path cost, and error correction.

[0098] Furthermore, after obtaining the adaptive correction amount, the robot needs to adjust the correction amount c accordingly. j Update the current path. This invention proposes a weighted shortest path optimization method, specifically adjusting the robot's trajectory by optimizing the path cost function. The new path cost C′ path (j,i) is determined by the original path cost C. path (j,i) and correction c j Joint decision:

[0099]

[0100] Among them, ∥c j ∥ is the correction amount c j The Euclidean norm represents the strength of the error correction; γ is the influence coefficient of the correction amount on the path cost. ν is the obstacle collision determination function, indicating whether the robot encounters an obstacle during path correction; ν is the weight of the collision correction term, reflecting the priority of avoiding obstacle collisions.

[0101] By adding collision correction items This solution can ensure error correction while avoiding collisions between the robot and obstacles caused by path correction.

[0102] Finally, the corrected path cost C′ path (j,i) provides new path planning results, optimizing the robot's navigation performance in dynamic environments.

[0103] Furthermore, the output of step 2 is the revised task execution plan. Each task and robot path planning result undergoes error correction. The output includes task number i, robot number j, and the corrected path cost C′. path (j,i) and the adjusted robot position r′ j :

[0104]

[0105] This task and path information will serve as input for step 3 (task execution and spatial accuracy optimization), helping the robot complete tasks more accurately and effectively reducing errors and efficiency losses during execution.

[0106] Step S2, real-time spatial error monitoring and feedback, is the core module of this patent. It follows the task allocation and path planning results from Step 1 and solves the accuracy problem of construction robot swarms performing tasks in dynamic environments through precise error monitoring, dynamic error correction, and path optimization. Adaptive correction strategies, weighted path optimization, and collision correction ensure that the robots can perform tasks efficiently and accurately in complex and uncertain construction environments. The output of this step provides more accurate task execution information for Step 3 and is a crucial part of ensuring task success rate and execution efficiency throughout the system.

[0107] S3. For the corrected task execution plan, perform a global analysis of the execution errors of all robots to obtain the global error. Based on the global error, dynamically adjust the task allocation and path planning to obtain the task execution plan after global task rescheduling and path optimization. The optimized task execution plan includes the task number, robot number, optimized path planning, and optimized path cost.

[0108] Specifically, in step 2, the present invention has already performed local corrections for the errors between each robot and the task through real-time spatial error monitoring. However, errors are not only local but may accumulate into global errors, affecting the execution efficiency of the entire task. Therefore, the primary task of step 3 is to perform a global analysis of the execution errors of all robots.

[0109] Furthermore, the present invention first defines a global error matrix E. global This is used to evaluate the overall error of the robot swarm in the current task execution. Global Error Matrix E global Based on the robot j and the task target position t i The distance between them is calculated to give the total error of all tasks. The formula is:

[0110]

[0111] Where, r′ j It is the current position (updated position) of robot j, t i It is the target location of task i. This is the set of all robots. This matrix evaluates the spatial deviation of all robots in task execution. If the error is large, it indicates that the robot cluster failed to complete the task accurately, which provides a basis for subsequent global adjustments and corrections.

[0112] Furthermore, error correction and global adjustment based on task priority:

[0113] The correction of errors is not only to simply reduce the deviation between the robot and the target position, but also needs to consider the urgency of the task. For those tasks that need to be completed as soon as possible (such as emergency repair tasks on construction sites), the strength of error correction should be greater to ensure that the task can be completed on time.

[0114] To achieve this, the present application introduces a priority factor a i for each task, which is used to adjust the strength of error correction. In addition, the correction of the task deviation of each robot j is also affected by its current workload, resource situation and other factors, so the present application introduces an adjustment factor b j for each robot.

[0115] On this basis, the present application introduces a weight and an adjustment factor for the error correction amount of each task, and finally defines the error correction amount c i of task i as:

[0116]

[0117] where c i is the error correction amount of task i; a i is the priority adjustment factor of task i; b j is the correction factor of robot j, which reflects its load and state. In this way, the urgency of the task and the state of the robot will jointly determine the error correction strategy of each task. This design can ensure that higher priority tasks receive stronger correction support, while robots with heavier loads or fewer resources will receive lower correction strength.

[0118] Further, global task rescheduling and path re-planning:

[0119] Based on the error correction amount c i of the previous step, the present application will dynamically adjust the task allocation and path planning. This is the core part of ensuring that the robot can flexibly respond to changes in the construction site, the priority of the task and the real-time state.

[0120] Suppose each task i has a corresponding path cost C′ path (j, i), which represents the path time or consumption when robot j performs task i. The present application proposes a dynamic task scheduling method based on error correction, which combines error correction with path cost to calculate the optimized path and execution strategy of each task.

[0121] Specifically, the new task allocation matrix T′ assign can be calculated by minimizing the task execution cost. The task allocation matrix T′ assign can be obtained by the following optimization formula:

[0122]

[0123] where T' is the new task allocation scheme; a is the adjustment weight of robot j; C' is the path cost of robot j performing task i; and g is the adjustment factor of path cost, used to balance the weight of path optimization and error correction. assign j path is the path cost of robot j performing task i; and g is the adjustment factor of path cost, used to balance the weight of path optimization and error correction.

[0124] Through this optimization process, the system not only can re-allocate tasks, but also can quickly adjust the robot path after each task adjustment, so as to maximize the execution efficiency of the task. In particular, the optimization of the robot path not only depends on the current error correction amount, but also considers the urgency of the task and the workload of the robot, further improving the collaborative efficiency of global task execution.

[0125] Further, the output of step S3 is the task execution scheme after global task rescheduling and path optimization and path planning result r' j . These information will be used as input for subsequent step 4 (task execution and feedback correction) to help the system dynamically adjust the cooperation strategy between tasks and robots during execution. The output data includes:

[0126]

[0127] These data will continue to be used to guide the task execution of the robot in the subsequent steps, and further optimize the execution process through the feedback correction mechanism.

[0128] S4, based on the optimized task execution scheme, construct a task-robot cooperation matrix, use the cooperation scheduling algorithm to schedule the task-robot cooperation matrix, determine the optimized scheduling task allocation scheme and the current position of each robot after adjustment.

[0129] Specifically, the input of step 4 comes from the output of step 3, that is, the task allocation scheme after global error correction and task adjustment and path planning r' j . These outputs contain the work undertaken by each robot in a specific task, as well as its path, load, execution time, etc.

[0130] Further, the construction of the task-robot cooperation matrix:

[0131] Construct a task-robot cooperation matrix C collab to describe the cooperation relationship between tasks and robots. The element C collab (i,j) of the matrix represents the cooperation effect between task i and robot j.​​

[0132] The calculation of the synergy effect is not only based on the physical capability of the robot, but also includes the task priority, the current state of the robot (e.g. power, load, etc.) and the error corrected result.

[0133] The elements of the synergy matrix are denoted as:

[0134]

[0135] where a i is the priority weight of task i, b j is the state adjustment factor of robot j, d(r′ j , t i ) is the error distance between robot j and the target of task i, d max is the maximum tolerable error distance of task i.

[0136] Further, the synergy scheduling algorithm is:

[0137] Based on the synergy matrix C collab , a dynamic scheduling algorithm is used to assign the most suitable robot for each task. The scheduling model is denoted as:

[0138]

[0139] where T schrfi;r is the task scheduling scheme, is the set of all robots, is the set of all tasks, r′ k is the current position of robot j, t o is the target position of task i.

[0140] Further, resource optimized scheduling:

[0141] A resource adjustment factor g k is introduced through the information of robot power, battery consumption rate, load, etc. The resource adjustment factor is denoted as:

[0142]

[0143] where E k is the current power of robot j, E max is the maximum power of robot j, L k is the current load of robot j, L max is the maximum load of robot j, and l is the adjustment factor. This factor reflects the influence of the state of the robot on the task allocation, and is used to avoid allocating high priority tasks to robots with insufficient resources.

[0144] Further, task and resource dynamic feedback correction:

[0145] Introducing dynamic feedback correction factor δ j (t), real-time adjusting the task allocation of robot j. The feedback correction factor is represented as:

[0146]

[0147] where ΔE j (t) is the power change of robot j at time t, ΔL j (t) is the load change of robot j at time t, and μ is the dynamic adjustment factor. Through real-time feedback, the system can adjust the task allocation, enabling robots to better adapt to changes occurring during execution.

[0148] The output of step S4 is the optimized scheduling task allocation scheme T schedule and the adjusted robot state r j . These information will be used as input in subsequent steps to ensure that robots can efficiently and accurately execute tasks. The output data format is:

[0149] T schedule ″={(i,j,r j ,t i ,γ j )} (17)。

[0150] S5, obtaining the original point cloud data, generating a preliminary three-dimensional model, and obtaining the task execution error when the robot executes the task, correcting the preliminary original point cloud data based on the task execution error and adjusting the task allocation scheme, obtaining the adjusted point cloud data and the position of each robot, and the optimized three-dimensional model, and merging the adjusted point cloud data and the optimized three-dimensional model into a complete three-dimensional model.

[0151] Specifically, the robot collects original point cloud data P i through sensors, which reflects the environmental information of the robot. Then, the invention needs to process these data to generate a preliminary three-dimensional model.

[0152] Input: original sensor data P o .

[0153] Processing process:

[0154] De-noising: using filtering algorithms (such as statistical filtering, voxel grid filtering, etc.) to remove unnecessary noise.

[0155] Registration: if the data comes from multiple sensors or perspectives, it needs to be aligned through registration algorithms.

[0156] Simplify: Reduce the redundancy of point cloud by sampling or clustering algorithm to improve the computational efficiency.

[0157] Preprocessing formula:

[0158] P i preprocessed = f preprocess (P i ) (18)

[0159] Where f preprocess is the combination of denoising, filtering, registration, etc.

[0160] Further, after preliminary modeling, the robot may be affected by many factors when performing tasks (e.g. positioning error, environmental interference, etc.), resulting in point cloud data deviating from the actual situation. At this time, the present application needs to correct these errors through feedback information in task execution.

[0161] Further, when performing tasks, each robot may have position error, execution time error, etc. For example, when robot j performs a task, the error between its actual position and target position may affect the final three-dimensional modeling accuracy. This error is called task execution error, and the present application adds this information to the model for correction.

[0162] Further, the error E j (t) of task execution can be quantified by the following formula:

[0163] E j (t) = λ1∥r′ j (t) - t i ∥ + λ2|ΔT j (t)| + λ3∥ΔL j (t)∥ (19)

[0164] Where r′ k (t) is the actual position of robot j at time t. t i is the expected position of target task i. ΔT j (t) is the time error of robot j performing the task. ΔL k (t) is the load error. λ1, λ2, λ3 are weighting coefficients.

[0165] Further, according to the execution error E j (t), the present application corrects the point cloud data. The correction strength depends on the size of the error and the resource status of the robot. The corrected point cloud data is:

[0166]

[0167] where μ(t) is a dynamic adjustment factor that adjusts the correction strength based on the robot's resources (such as power, load).E max is the maximum error allowed in the task.

[0168] Further, the adaptive adjustment factor μ(t) reflects the influence of the robot's resource state on error correction. It can be calculated using the following formula:

[0169]

[0170] where λ is a constant that adjusts the sensitivity of the correction factor.R j (t) is the resource consumption of robot j at time t.R max is the maximum value of resource consumption in the task.

[0171] Further, the task scheduling scheme T schedule and the adjusted robot state r j play an important role in this step. They determine the order, method and resource allocation of the robot in executing the task, thus affecting the accuracy of the final three-dimensional model.

[0172] Task scheduling and state adjustment formula:

[0173] Task scheduling takes into account the current state of the robot and task constraints:

[0174]

[0175] where T schedule final is the adjusted task scheduling, c j (t) is a cost function related to resource consumption, d j (t) is the cost related to task execution delay or error.

[0176] Based on the scheduling, the state of the robot r j (t) is dynamically adjusted by the scheduling scheme:

[0177] r j (t) = r j (t-1) + Δr j (t) (23)

[0178] where Δr j (t) is the state change amount adjusted according to the scheduling scheme and feedback error.

[0179] Further, through task scheduling and error correction, the closed-loop feedback mechanism will continuously optimize the three-dimensional modeling process. The execution result of each task will provide feedback for the execution of the next task, thereby improving the accuracy of the final model.

[0180] point cloud data P after error correction and task scheduling optimization i corrected and updated robot state r j With feedback information (error feedback and task progress feedback), the three-dimensional model is weighted and optimized. The optimized three-dimensional model gradually fits the actual environment. Closed-loop optimization formula:

[0181] M optimized (t)=M optimized (t-1)+α(P i corrected -M optimized (t-1))(24)

[0182] Where M optimized (t) is the current time optimized three-dimensional model. Alpha is the optimization step, used to control the amplitude of correction. P i corrected is the error-corrected point cloud data.

[0183] Further, during the task execution process, the robot collects point cloud data at different angles. Through task scheduling T schedule ', the invention determines which data has higher priority for merging, thereby optimizing the final model.

[0184] Input: error-corrected point cloud data P i corrected and optimized three-dimensional model M optimized .

[0185] Data fusion process: using weighted average method, de-redundancy and filling gaps, etc. Technology, different sources of data are merged into a complete three-dimensional model.

[0186] Three-dimensional model fusion formula:

[0187] M 3D =merge(M optimized ,P i corrected )(25)

[0188] Further, the specific fusion method is:

[0189] Step 1: Registration of point cloud data, aligning data from different sensors.

[0190] Step 2: Eliminate redundant points, use weighted average method to fuse overlapping area data.

[0191] Step 3: Fill in the blank area, use interpolation or image filling method to process the unscanned area.

[0192] In one or more embodiments, as shown in Figure 2 A three-dimensional modeling design system based on multi-robot cooperation of construction robots is disclosed, and the system comprises:

[0193] An environment data acquisition unit 31 is configured to perform dynamic task allocation and path planning according to environment data of a construction site, robot state information, and task demand data, to obtain a set of tasks and paths allocated to each robot as task allocation information, wherein the robot state information includes a current position and a target task position.

[0194] In the environment data acquisition unit, a multi-objective optimization function robot is used to perform task allocation, and a hybrid method of dynamic programming and A algorithm is used to perform path optimization.

[0195] A task scheduling unit 32 is configured to acquire real-time feedback of each robot performing a task, to perform local space error monitoring on each robot based on an error of the real-time feedback and the task allocation information, to calculate a path correction amount of a current robot dynamically according to a size of the local space error and an urgency of the task using machine learning, and to update a current path of the robot according to the correction amount to obtain a corrected task execution scheme, wherein the corrected task execution scheme includes a task number, a robot number, a corrected path cost, and an updated robot position.

[0196] A global optimization unit 33 is configured to perform global analysis on execution errors of all robots for the corrected task execution scheme to obtain a global error, to perform dynamic adjustment on task allocation and path planning based on the global error, and to obtain a task execution scheme after global task rescheduling and path optimization, wherein the optimized task execution scheme includes a task number, a robot number, an optimized path planning, and an optimized path cost.

[0197] A cooperative optimization unit 34 is configured to construct a task-robot cooperation matrix based on the optimized task execution scheme, to perform task scheduling on the task-robot cooperation matrix using a cooperative scheduling algorithm, to determine an optimized scheduling task allocation scheme and an updated current position of each robot.

[0198] A three-dimensional modeling unit 35 is configured to obtain original point cloud data, to generate a preliminary three-dimensional model, to obtain a task execution error when a robot performs a task, to correct the preliminary original point cloud data based on the task execution error and to adjust the task allocation scheme, to obtain adjusted point cloud data and a position of each robot, and an optimized three-dimensional model, and to merge the adjusted point cloud data and the optimized three-dimensional model into a complete three-dimensional model.

[0199] It should be noted that the specific work flow of the three-dimensional modeling design system based on multi-robot cooperation of construction robots provided by the embodiments of the present application is the same as the flow of the three-dimensional modeling design method based on multi-robot cooperation of construction robots described in the above embodiments, and will not be repeated here.

[0200] Compared with the prior art, the three-dimensional modeling design system based on multi-robot cooperation of construction robots provided by the embodiments of the present application performs dynamic task allocation and path planning according to the environmental data of the construction site, the robot state information, and the task demand data, obtains the set of tasks and paths allocated to each robot as task allocation information, wherein the robot state information includes the current position and the target task position; the multi-objective optimization function robot performs task allocation, and the mixed method of dynamic programming and A algorithm is used for path optimization; real-time feedback of each robot executing the task is collected, local space error monitoring is performed on each robot based on the error of the real-time feedback and the task allocation information, the path correction amount of the current robot is calculated dynamically according to the size of the local space error and the urgency of the task using machine learning, and the robot updates the current path according to the correction amount to obtain a corrected task execution scheme; wherein the corrected task execution scheme includes task number, robot number, corrected path cost and updated robot position; for the corrected task execution scheme, the execution error of all robots is globally analyzed to obtain a global error, and the task allocation and path planning are dynamically adjusted based on the global error to obtain a task execution scheme after global task rescheduling and path optimization; wherein the optimized task execution scheme includes task number, robot number, optimized path planning and optimized path cost; based on the optimized task execution scheme, a task-robot cooperation matrix is constructed, a task scheduling algorithm is used to schedule the task-robot cooperation matrix to determine the optimized scheduling task allocation scheme and the current position of each robot after adjustment; the original point cloud data is obtained, a preliminary three-dimensional model is generated, and the task execution error when the robot executes the task is obtained, the preliminary original point cloud data is corrected based on the task execution error and the task allocation scheme is adjusted to obtain adjusted point cloud data and the position of each robot, and an optimized three-dimensional model, and the adjusted point cloud data and the optimized three-dimensional model are merged into a complete three-dimensional model.

[0201] The embodiments of the present application also provide a three-dimensional modeling design device based on multi-robot cooperation of construction robots, which comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the steps in the above three-dimensional modeling design method based on multi-robot cooperation of construction robots, for example Figure 1The processor implements the functions of the modules in the above system embodiments when executing the computer program.

[0202] The computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the three-dimensional modeling design device based on the multi-machine cooperation of construction robots.

[0203] The three-dimensional modeling design device based on the multi-machine cooperation of construction robots can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices. The three-dimensional modeling design device based on the multi-machine cooperation of construction robots can include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the three-dimensional modeling design device based on the multi-machine cooperation of construction robots can also include input and output devices, network access devices, buses and the like.

[0204] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is the control center of the three-dimensional modeling design device based on the multi-machine cooperation of construction robots, and connects all parts of the three-dimensional modeling design device based on the multi-machine cooperation of construction robots through various interfaces and lines.

[0205] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the three-dimensional modeling design device based on the multi-machine cooperation of the construction robot by running or executing the computer program and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, etc.; and the data storage area can store data created according to the running of the air conditioner controller, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash storage device, or other volatile solid-state storage devices.

[0206] The modules integrated in the three-dimensional modeling design device based on the multi-machine cooperation of the construction robot can be stored in a computer-readable storage medium if they are realized in the form of software function units and sold or used as independent products. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer-readable storage medium. The computer program can implement the steps of the above-mentioned various method embodiments when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer-readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0207] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium, and the program can include the processes of the above-mentioned various method embodiments when executed. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0208] The above is the preferred embodiment of the present application, it should be pointed out that, for those skilled in the art, without departing from the principles of the present application, can also make a number of improvements and refinements, these improvements and refinements are also considered to be within the scope of the present application.

Claims

1. A 3D modeling and design method based on multi-robot collaboration in construction, characterized in that, The method includes: S1. Based on the environmental data of the construction site, robot status information, and task requirement data, perform dynamic task allocation and path planning to obtain a set of tasks and paths assigned to each robot as task allocation information; wherein, the robot status information includes the current position and the target task position. In step S1, a multi-objective optimization function is used to assign tasks to the robot, and a hybrid method of dynamic programming and A-algorithm is used for path optimization. The multi-objective optimization function dynamically calculates the matching degree between the task and the robot using a weighted average method for task allocation. The hybrid method of dynamic programming and A-algorithm for path optimization is expressed as follows: Define the robot's current position as The target mission location is The goal of path planning is to minimize the distance from the starting point of the path. arrive Path cost The path cost Taking into account path length, obstacle avoidance, and robot load capacity, it is expressed as: ； in, These are the coordinates of each node in the path; It is an indicator function that indicates whether a path node is present. There are obstacles at the location, where obstacles are represented by 1 and blank areas by 0; It is the weighting factor for obstacle avoidance, where n is the total number of nodes; S2. Collect real-time feedback on the task execution of each robot, and monitor the local spatial error of each robot based on the error of the real-time feedback and task allocation information. Use machine learning to dynamically calculate the path correction amount of the current robot according to the magnitude of the local spatial error and the urgency of the task. The robot updates the current path according to the correction amount to obtain the corrected task execution plan. The corrected task execution plan includes the task number, robot number, corrected path cost and updated robot position. S3. For the corrected task execution plan, perform a global analysis of the execution errors of all robots to obtain the global error. Based on the global error, dynamically adjust the task allocation and path planning to obtain the task execution plan after global task rescheduling and path optimization. The optimized task execution plan includes the task number, robot number, optimized path planning, and optimized path cost. S4. Based on the optimized task execution plan, construct the task-robot collaboration matrix, use the collaborative scheduling algorithm to schedule tasks on the task-robot collaboration matrix, and determine the optimized task allocation scheme and the current position of each robot after adjustment. S5. Acquire raw point cloud data, generate a preliminary 3D model, and simultaneously acquire the task execution error of the robot when performing the task. Based on the task execution error, correct the preliminary raw point cloud data and adjust the task allocation scheme to obtain the adjusted point cloud data, the position of each robot, and the optimized 3D model. Merge the adjusted point cloud data and the optimized 3D model into a complete 3D model.

2. The 3D modeling and design method based on multi-robot collaboration in construction as described in claim 1, characterized in that, The local spatial error is the robot's current position. With mission objectives Local spatial error between ;but Based on the local spatial error Obtain the adaptive correction amount; wherein, the adaptive correction amount is based on the local spatial error. The robot dynamically adjusts its path based on the urgency of the task; it acquires the adaptive correction amount and updates the current path using a cost function based on the adaptive correction amount to obtain a new path cost; the cost function is an obstacle collision determination function, indicating whether the robot encounters an obstacle during path correction.

3. The 3D modeling and design method based on multi-robot collaboration in construction as described in claim 2, characterized in that, The global error is the local spatial error. The total error; the global error is calculated based on the distance between the robot and the target position, giving the total error for all tasks.

4. The 3D modeling and design method based on multi-robot collaboration in construction as described in claim 3, characterized in that, The dynamic adjustment of task allocation and path planning based on global error to obtain a task execution plan after global task rescheduling and path optimization includes: Obtain the global error; The priority of tasks is globally adjusted based on the global error to obtain the task... Error correction amount; Obtain the cost of the new path, combined with the task. The error correction amount is used to determine a new task allocation scheme.

5. The 3D modeling and design method based on multi-robot collaboration in construction as described in claim 4, characterized in that, The cost of obtaining a new path, combined with the task The error correction amount is used to determine the new task allocation scheme: Calculated by minimizing task execution cost, including: ； in, It's a new task allocation scheme; It's a robot. Adjusting the weights; It is a modified robot Execute the task Path cost; It is a path cost adjustment factor used to balance the weights of path optimization and error correction; Let i be the target location of task i; This is the adjusted robot position.

6. The 3D modeling and design method based on multi-robot collaboration in construction as described in claim 1, characterized in that, The task-robot collaboration matrix represents the collaborative relationship between tasks and robots; the elements of the matrix represent tasks. With robots The synergistic effect between them; the task-robot synergy matrix is ​​represented as: ； in, For the task Priority weights, For robots State regulation factor, For robots With the task Target error distance, It is a task Maximum tolerance error distance; Based on the task-robot collaboration matrix, a dynamic scheduling algorithm is used to assign the most suitable robot to each task. The dynamic scheduling algorithm incorporates a resource adjustment factor for task scheduling optimization and a dynamic feedback correction factor to adjust the robot in real time. Task allocation.

7. The 3D modeling and design method based on multi-robot collaboration in construction as described in claim 6, characterized in that, Task execution error when the robot performs a task , is represented as: ； in, It's a robot. At any moment The actual location; The target task The expected location; It's a robot. Time error in task execution; It is a load error; These are weighting coefficients; Based on task execution error The point cloud data is corrected and represented as follows: ； in, For the corrected point cloud data, This is the raw point cloud data; It is a dynamic adjustment factor that adjusts the correction intensity based on the robot's resources; This is the maximum allowable error in the task; Based on the corrected point cloud data The original 3D model is weighted and optimized to obtain the optimized 3D model at the current time step. ; The corrected point cloud data The optimized 3D model at the current moment Data fusion is performed to obtain the final complete 3D model.

8. A system for implementing the 3D modeling and design method based on multi-robot collaboration in construction as described in claim 1, characterized in that, The system includes: The environmental data acquisition unit is used to dynamically allocate tasks and plan paths based on environmental data from the construction site, robot status information, and task requirement data, obtaining a set of tasks and paths assigned to each robot as task allocation information; wherein, the robot status information includes the current position and the target task position; In the environmental data acquisition unit, a multi-objective optimization function is used to allocate tasks to the robot, and a hybrid method of dynamic programming and A algorithm is used for path optimization. The task scheduling unit is used to collect real-time feedback on the tasks executed by each robot. Based on the errors in the real-time feedback and task allocation information, it performs local spatial error monitoring on each robot. Using machine learning, it dynamically calculates the path correction amount for the current robot based on the magnitude of the local spatial error and the urgency of the task. The robot updates its current path based on the correction amount to obtain the corrected task execution plan. The corrected task execution plan includes the task number, robot number, corrected path cost, and updated robot position. The global optimization unit is used to perform a global analysis of the execution errors of all robots for the corrected task execution plan, obtain the global error, and dynamically adjust the task allocation and path planning based on the global error to obtain the task execution plan after global task rescheduling and path optimization; wherein, the optimized task execution plan includes task number, robot number, optimized path planning and optimized path cost; The collaborative optimization unit is used to construct a task-robot collaborative matrix based on the optimized task execution plan, use a collaborative scheduling algorithm to schedule tasks on the task-robot collaborative matrix, and determine the optimized task allocation plan and the current position of each robot after adjustment. The 3D modeling unit is used to acquire raw point cloud data, generate a preliminary 3D model, and acquire the task execution error when the robot performs the task. Based on the task execution error, the preliminary raw point cloud data is corrected and the task allocation scheme is adjusted to obtain the adjusted point cloud data, the position of each robot, and the optimized 3D model. The adjusted point cloud data and the optimized 3D model are then merged into a complete 3D model.

Images