Building robot collaborative operation system

Abstract

The invention relates to the technical field of building engineering automation, and discloses a building robot collaborative operation system, a BIM modeling module is used for establishing a three-dimensional model of building engineering and dividing the three-dimensional model into a plurality of operation areas; the information acquisition module is used for acquiring state information and working environment information of the robot; the central control module is used for performing task allocation and operation planning by using an ant colony algorithm; the information sharing module is used for carrying out real-time communication between the robots through wireless communication in the operation process, and sharing operation progress, state information and environment information; the operation evaluation module is used for monitoring the operation progress and state of each robot in real time, evaluating the operation effect of the robot, adjusting task allocation and planning strategies in time according to the evaluation result, optimizing the collaborative operation process of the robot, and improving the operation efficiency of the robot. According to the invention, the problems of unreasonable task allocation, low communication efficiency, difficult operation coordination and the like in the collaborative operation of the existing building robot are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of construction engineering automation, and particularly relates to a collaborative operation system for construction robots. Background Art

[0002] In the field of construction engineering, with the development of automation technology, the application of construction robots is becoming more and more extensive. However, at present, when construction robots perform collaborative operations, there are problems such as unreasonable task allocation, low communication efficiency, and difficult operation coordination, resulting in low operation efficiency and the advantages of construction robots cannot be fully utilized. For example, when multiple robots collaborate to complete a construction task, it may occur that some robots have too heavy tasks while some robots are idle. At the same time, the communication between robots may be delayed or incorrect, affecting the smooth progress of the operation. Summary of the Invention

[0003] The purpose of the present invention is to solve the above problems and design a collaborative operation system for construction robots.

[0004] The present invention provides a collaborative operation system for construction robots, which includes: A BIM modeling module, which is used to establish a three-dimensional model of a construction project by using BIM technology and divide the BIM model into multiple operation areas; An information acquisition module, which is used to collect the status information of the robot and the operation environment information in real time through sensors equipped on each robot; A central control module, which is used to receive the status information and operation environment information sent by each robot, combine the operation area division and operation requirements in the BIM model, and use the ant colony algorithm for task allocation and operation planning; An information sharing module, which is used for each robot to move to the corresponding operation area for operation according to the assigned tasks and operation plans. During the operation process, the robots communicate in real time through wireless communication and share the operation progress, status information and environment information; An operation evaluation module, which is used to monitor the operation progress and status of each robot in real time, evaluate the operation effect of the robot, and adjust the task allocation and planning strategy in time according to the evaluation result to optimize the collaborative operation process of the robot.

[0005] Optionally, in the first implementation manner of the present invention, the BIM modeling module includes: A semantic segmentation sub-module, which is used to obtain the three-dimensional point cloud data of the BIM model, perform semantic segmentation on the three-dimensional point cloud data, extract geometric features and semantic features, and construct a multi-dimensional feature vector including spatial position and structural semantics; A learning sub-module, which is used to abstract 3D point cloud data into nodes in a graph structure based on the multi-dimensional feature vectors output by the semantic segmentation sub-module, learn the spatial dependence relationships between nodes through a graph neural network, and generate a graph representation containing structural logic; A partitioning sub-module, which is used to perform multi-scale feature extraction on the BIM model by using a 3D convolutional neural network based on the graph representation output by the learning sub-module, and introduce a hierarchical clustering algorithm to perform hierarchical partitioning of multi-scale spatial regions so as to partition into multiple operation regions.

[0006] Optionally, in the second implementation manner of the present invention, the partitioning sub-module includes: Performing multi-scale feature extraction on the BIM model by using a 3D convolutional neural network. The 3D convolutional neural network includes multiple convolutional layers and pooling layers. Local detail features are extracted by shallow convolutional layers, and global structural features are extracted by deep convolutional layers. Feature maps of different scales are integrated through a pyramid pooling module to generate a multi-scale feature representation; Based on the extracted multi-scale features, taking each small region as an initial clustering unit, calculating the feature similarity between each small region by using cosine similarity, and merging the clustering units with the highest similarity in a bottom-up manner to form a larger region, so as to obtain a region partitioning result.

[0007] Optionally, in the third implementation manner of the present invention, the information acquisition module includes: A establishing sub-module, which is used to perform timestamp alignment on the original data of the sensor, establish the conversion relationship between the coordinate systems of each sensor and the robot body coordinate system, and generate a spatio-temporal alignment data set; A processing sub-module, which is used to construct a spatio-temporal attention network to process the multi-modal data in the spatio-temporal alignment data set, extract the spatial features of the visual image by using a convolutional neural network, extract the 3D spatial features of the lidar data by using a point cloud processing algorithm, and utilize the self-attention mechanism of the Transformer encoder to capture the dependence relationships between data of different sensors at different times, and output a fusion feature vector containing spatio-temporal context information; An analyzing sub-module, which is used to analyze the fusion feature vector to obtain the state information of the robot and the operation environment information.

[0008] Optionally, in the fourth implementation manner of the present invention, the analyzing sub-module includes: Constructing a feature distribution model in a normal operation state by using an auto-encoder, inputting the fusion feature vector into the fusion feature vector, and detecting abnormal events by calculating the reconstruction error; Performing semantic segmentation on the fusion feature vector by using a ViT-BERT model to identify the categories and positions of objects, and generating an environmental map with semantic labels; Integrate the output of the feature distribution model and the ViT-BERT model to obtain the state information of the robot and the information of the working environment.

[0009] Optionally, in the fifth implementation manner of the present invention, the central control module includes: An initialization sub-module, configured to decompose the construction engineering task into multiple sub-tasks, each sub-task corresponding to a working area, and initialize the ant colony parameters; A selection sub-module, where each ant represents a robot, and according to the pheromone concentration and heuristic information of each current sub-task, selects the next sub-task to be executed. When an ant selects a task, it follows the roulette wheel selection strategy; An allocation sub-module, configured to update the pheromone concentration between each sub-task according to the result of the task allocation after all ants complete a task allocation; An iteration sub-module, configured to obtain an optimal task allocation scheme after multiple iterations, allocate sub-tasks and working areas to each robot, and plan the working sequence and path of the robot.

[0010] Optionally, in the sixth implementation manner of the present invention, the ant colony parameters at least include the number of ants, the initial concentration of pheromone, the heuristic factor, and the pheromone evaporation factor.

[0011] Optionally, in the seventh implementation manner of the present invention, the information sharing module includes: An update sub-module, configured to, when the robot moves, update the neighbor node list in real time through signal strength detection, and use the Q-learning algorithm to dynamically adjust the communication link to form an adaptive Mesh network topology; A calculation sub-module, configured to, when two robots request the same resource, calculate the conflict resolution weight according to the task priority, the current load of the robot, and the resource occupation duration, and put the robot with a lower priority into the waiting queue; An implementation sub-module, configured to, if a path planning conflict occurs, obtain an optimal concession strategy through the Nash equilibrium algorithm to realize real-time communication and collaborative control of multiple robots.

[0012] Optionally, in the eighth implementation manner of the present invention, the implementation sub-module includes: Each robot, based on the current position, speed, and target path, uses a Kalman filter to predict the sequence of trajectory points within a future time window, and generates a spatio-temporal occupancy grid map; Spatially superimpose the spatio-temporal occupancy grid maps of all robots, identify the areas with spatio-temporal overlap, and construct a conflict state vector; Based on the conflict state vector, define an optional strategy set for each robot, and for each strategy combination, calculate the benefit of the robot to construct a multi-dimensional benefit matrix; Traverse the multi-dimensional revenue matrix to solve the Nash equilibrium point, determine the final Nash equilibrium strategy combination to be executed, and obtain the optimal concession strategy.

[0013] Optionally, in the ninth implementation manner of the present invention, a method for implementing the described construction robot collaborative operation system includes the following steps: Use BIM technology to establish a three-dimensional model of the construction project and divide the BIM model into multiple operation areas; Real-time collect the status information of the robot and the operation environment information through the sensors equipped on each robot; Receive the status information and operation environment information sent by each robot, combine the operation area division and operation requirements in the BIM model, and use the ant colony algorithm for task allocation and operation planning; Each robot moves to the corresponding operation area for operation according to the assigned tasks and operation plans. During the operation process, the robots communicate in real time through wireless communication, sharing the operation progress, status information, and environment information; Real-time monitor the operation progress and status of each robot, evaluate the operation effect of the robot, and according to the evaluation results, timely adjust the task allocation and planning strategies to optimize the collaborative operation process of the robots.

[0014] In the technical solution provided by the present invention, use BIM technology to establish a three-dimensional model of the construction project and divide the BIM model into multiple operation areas; real-time collect the status information of the robot and the operation environment information through the sensors equipped on each robot; receive the status information and operation environment information sent by each robot, combine the operation area division and operation requirements in the BIM model, and use the ant colony algorithm for task allocation and operation planning; each robot moves to the corresponding operation area for operation according to the assigned tasks and operation plans. During the operation process, the robots communicate in real time through wireless communication, sharing the operation progress, status information, and environment information; real-time monitor the operation progress and status of each robot, evaluate the operation effect of the robot, and according to the evaluation results, timely adjust the task allocation and planning strategies to optimize the collaborative operation process of the robots; the present invention provides accurate basic data for task allocation by establishing a BIM model, uses the ant colony algorithm for task allocation and planning, improves the rationality and efficiency of task allocation, realizes the efficient collaborative operation of robots through real-time communication and collaborative control algorithms between robots. At the same time, operation monitoring and evaluation can timely adjust operation strategies, ensure the quality and efficiency of operations, solve the problems of unreasonable task allocation, low communication efficiency, and difficult operation coordination existing in the existing construction robot collaborative operation, and have broad application prospects. Description of the Drawings

[0015] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention.

[0016] Figure 1 Schematic diagram of the structure of the building robot collaborative operation system provided by the embodiment of the present invention; Figure 2 Schematic diagram of the structure of the BIM modeling module provided by the embodiment of the present invention; Figure 3 Schematic diagram of the structure of the information acquisition module provided by the embodiment of the present invention. Detailed implementation manners

[0017] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and do not necessarily have to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order different from that shown or described herein. In addition, the terms "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0018] For ease of understanding, the specific processes of the embodiments of the present invention are described below. Please refer to Figure 1 Schematic diagram of the structure of the building robot collaborative operation system provided by the embodiment of the present invention. The system includes: A BIM modeling module for using BIM technology to establish a three-dimensional model of a construction project and dividing the BIM model into multiple operation areas; An information acquisition module for real-time collecting the status information of the robot and the operation environment information through sensors equipped on each robot; A central control module for receiving the status information and operation environment information sent by each robot, combining the operation area division and operation requirements in the BIM model, and using the ant colony algorithm for task allocation and operation planning; An information sharing module for each robot to move to the corresponding operation area for operation according to the assigned tasks and operation plans. During the operation process, the robots communicate in real time through wireless communication to share the operation progress, status information, and environment information; The job evaluation module is used to monitor the job progress and status of each robot in real time, evaluate the job effect of the robot, and adjust the task allocation and planning strategy in a timely manner according to the evaluation results, so as to optimize the collaborative job process of the robot.

[0019] In this embodiment, please refer to Figure 2 , the BIM modeling module includes: The semantic segmentation sub-module is used to obtain the three-dimensional point cloud data of the BIM model, perform semantic segmentation on the three-dimensional point cloud data, extract geometric features and semantic features, and construct a multi-dimensional feature vector containing spatial position and structural semantics; The learning sub-module is used to abstract the three-dimensional point cloud data into nodes in the graph structure based on the multi-dimensional feature vector output by the semantic segmentation sub-module, and learn the spatial dependence relationship between nodes through the graph neural network to generate a graph representation containing structural logic; The division sub-module is used to perform multi-scale feature extraction on the BIM model by using a three-dimensional convolutional neural network based on the graph representation output by the learning sub-module, and introduce a hierarchical clustering algorithm to perform hierarchical division of multi-scale spatial regions to divide into multiple job regions.

[0020] In this embodiment, the division sub-module includes: Perform multi-scale feature extraction on the BIM model by using a three-dimensional convolutional neural network. The three-dimensional convolutional neural network contains multiple convolutional layers and pooling layers. Local detail features are extracted by the shallow convolutional layer, and global structural features are extracted by the deep convolutional layer. The feature maps of different scales are integrated through the pyramid pooling module to generate a multi-scale feature representation; Based on the extracted multi-scale features, each small region is used as an initial clustering unit, the feature similarity between each small region is calculated by using cosine similarity, and the clustering units with the highest similarity are merged in a bottom-up manner to form a larger region to obtain the region division result.

[0021] In this embodiment, please refer to Figure 3 , the information acquisition module includes: The establishment sub-module is used to align the timestamps of the original data of the sensors, establish the conversion relationship between the coordinate systems of each sensor and the robot body coordinate system, and generate a spatio-temporal alignment data set; The processing sub-module is used to construct a spatio-temporal attention network to process the multi-modal data in the spatio-temporal alignment data set, extract the spatial features of the visual image through a convolutional neural network, extract the three-dimensional spatial features of the lidar data through a point cloud processing algorithm, and use the self-attention mechanism of the Transformer encoder to capture the dependencies between data at different times and different sensors, and output a fusion feature vector containing spatio-temporal context information; An analysis sub-module, which is used to analyze the fused feature vector to obtain the state information of the robot and the operation environment information.

[0022] In this embodiment, the analysis sub-module includes: Construct a feature distribution model under normal operation conditions using an autoencoder, input the fused feature vector into the fused feature vector, and detect abnormal events by calculating the reconstruction error; Use the ViT-BERT model to perform semantic segmentation on the fused feature vector, identify the category and location of objects, and generate an environmental map with semantic labels; Integrate the outputs of the feature distribution model and the ViT-BERT model to obtain the state information of the robot and the operation environment information.

[0023] In this embodiment, the central control module includes: An initialization sub-module, which is used to decompose the construction engineering task into multiple sub-tasks, each sub-task corresponding to an operation area, and initialize the ant colony parameters; A selection sub-module, where each ant represents a robot, and according to the pheromone concentration and heuristic information of each current sub-task, select the next sub-task to be executed. When an ant selects a task, it follows the roulette wheel selection strategy; An allocation sub-module, which is used to update the pheromone concentration between each sub-task according to the task allocation result when all ants complete a task allocation; An iteration sub-module, which is used to obtain the optimal task allocation plan after multiple iterations, allocate sub-tasks and operation areas to each robot, and plan the operation sequence and path of the robot.

[0024] In this embodiment, the ant colony parameters at least include the number of ants, the initial concentration of pheromone, the heuristic factor, and the pheromone evaporation factor.

[0025] In this embodiment, the information sharing module includes: An update sub-module, which is used to update the neighbor node list in real time by signal strength detection when the robot moves, and dynamically adjust the communication link using the Q-learning algorithm to form an adaptive Mesh network topology; A calculation sub-module, which is used to calculate the conflict resolution weight according to the task priority, the current load of the robot, and the resource occupation duration when two robots request the same resource, and the robot with a lower priority enters the waiting queue; An implementation sub-module, which is used to obtain the optimal concession strategy through the Nash equilibrium algorithm if a path planning conflict occurs, and realize multi-robot real-time communication and cooperative control.

[0026] In this embodiment, the implementation sub-module includes: Each robot predicts a sequence of trajectory points within a future time window using a Kalman filter based on its current position, speed, and target path, and generates a spatio-temporal occupancy grid map; Spatially superimpose the spatio-temporal occupancy grid maps of all robots, identify regions with spatio-temporal overlap, and construct a conflict state vector; Based on the conflict state vector, define a set of optional strategies for each robot, calculate the payoffs of the robots for each strategy combination, and construct a multi-dimensional payoff matrix; Traverse the multi-dimensional payoff matrix to solve for the Nash equilibrium point, determine the final Nash equilibrium strategy combination to be executed, and obtain the optimal concession strategy.

[0027] In this embodiment, the specific extensions for the BIM modeling module include: Downsample the 3D point cloud data exported from the BIM model, reduce data redundancy through voxel grid filtering while retaining key geometric features; remove outliers from the downsampled point cloud using a statistical outlier filtering algorithm, calculate the average distance of each point to its neighboring points, and consider points deviating from the statistical distribution as noise points and remove them; finally, convert the processed point cloud data into a regular voxel grid representation to provide a unified input format for subsequent convolutional operations; Use the PointNet++ network architecture for semantic segmentation of the voxelized point cloud. This network captures local geometric features at different scales through a multi-layer feature extractor; first, use the Set Abstraction module to downsample the input point cloud and extract the features of local regions; then, upsample the low-resolution features to the original point cloud density through the Feature Propagation module to achieve per-point classification; during the training process, use the cross-entropy loss function and combine data augmentation techniques (random rotation, scaling) to improve the generalization ability of the model, and finally assign semantic labels (walls, floors, beams, columns, etc.) to each point; Extract the geometric features of the point cloud, including the normal vector, curvature, local surface roughness, etc. of the points, obtained by calculating the covariance matrix of each point's neighborhood and performing eigen-decomposition; at the same time, extract semantic features such as the class probability distribution and semantic boundary information from the semantic segmentation results; concatenate the geometric features and semantic features to construct a high-dimensional feature vector; to reduce feature redundancy, use principal component analysis to perform dimensionality reduction on the feature vector, retain the principal components with the largest variance contribution, and generate a multi-dimensional feature vector containing spatial position and structural semantics; The point cloud data after semantic segmentation and feature extraction is abstracted into nodes in a graph structure, where each node represents a semantic region (a wall, a column); node attributes include the geometric center coordinates, dimension parameters (length, width, height), semantic category labels, and statistical features extracted from the point cloud (average normal vector, curvature distribution) of the region; at the same time, construction process constraint attributes, such as pouring sequence requirements and load-bearing dependency relationships, are assigned to each node, and these attributes come from the design parameters of the BIM model and construction specifications; Edges are constructed between nodes to represent spatial dependency relationships, and the weights of the edges reflect the association strength between nodes; first, an initial graph structure is constructed based on spatial proximity, and if the minimum distance between two nodes is less than a set threshold, an edge is established between them; then, the weights of the edges are calculated, comprehensively considering geometric correlation (coplanarity, proportion of adjacent area), semantic correlation (support relationship between walls and beams / columns), and construction sequence constraints (foundation construction must be completed first before the main structure construction can be carried out); for nodes with direct functional dependencies (doors / windows and walls), higher edge weights are assigned; The graph convolutional network (GCN) is used to learn the spatial dependency relationships between nodes, and the node feature representations are updated through multi-layer graph convolutional operations; in each layer of convolution, a node updates its own features by aggregating the feature information of its neighbor nodes, and the aggregation function considers the weights of the edges to distinguish the importance of different neighbors; during the training process, self-supervised learning tasks (node classification, link prediction) are used to optimize the network parameters, enabling the model to capture the potential dependency relationships in the building structure; by predicting the semantic categories of the nodes, the model is forced to learn the association between the spatial layout and structural functions; finally, a graph representation containing spatial dependency relationships is output, providing structural constraints for subsequent work area division; A three-dimensional convolutional neural network is used to perform multi-scale feature extraction on the BIM model; the network contains multiple convolutional layers and pooling layers, and spatial features at different scales are captured through convolutional kernels of different sizes; shallow convolutional layers extract local detail features (component edges, holes), and deep convolutional layers extract global structural features (floor layout, functional partition); after each layer of convolution, batch normalization and ReLU activation functions are applied to enhance the model's stability and non-linear expression ability; different-scale feature maps are integrated through a pyramid pooling module to generate a multi-scale feature representation; Based on the extracted multi-scale features, a hierarchical clustering algorithm is used for spatial region division. First, each voxel or small region is used as an initial clustering unit, and the feature similarity (cosine similarity) between them is calculated. Then, in a bottom-up manner, the clustering units with the highest similarity are merged to form larger regions. During the merging process, the Ward method is used to minimize the within-class variance to ensure feature consistency within the same region. At the same time, a spatial continuity constraint is introduced to preferentially merge adjacent regions to avoid generating fragmented region division results. By setting different clustering stopping conditions, multi-level region division results (floors, rooms, functional areas, etc.) are obtained. The initially divided regions are optimized in combination with construction technology constraints. First, according to the construction sequence requirements (main body first and then decoration), the region boundaries are adjusted to ensure that the tasks within the same region can be continuously executed in sequence. Then, considering the limitations of the robot's operation ability, the regions are decomposed or merged to ensure that the size and complexity of each region are within the operable range of the robot. The large-area wall is divided into multiple sub-regions, and the height and width of each sub-region do not exceed the maximum operation radius of the robotic arm. Finally, based on the spatial dependency relationship learned by the graph neural network, it is checked whether the region division result meets the structural construction logic, and the regions that violate the dependency relationship are readjusted to ensure the construction feasibility of the divided operation regions. In this embodiment, the specific extensions for the information acquisition module include: The raw data of sensors such as cameras, lidars, and IMUs are time-stamped and aligned. The transformation relationship between the coordinate systems of each sensor and the robot's body coordinate system is established through the hand-eye calibration algorithm. For asynchronously acquired sensors, linear interpolation is used to synchronize the data in time to ensure the consistency of the environmental perception data at the same moment in the spatial coordinate system, and a spatio-temporal aligned data set including position, attitude, environmental point cloud, and visual image is generated. A spatio-temporal attention network is constructed. First, the spatial features (obstacle contours, material positions) of the visual image are extracted through a convolutional neural network, and the three-dimensional spatial features of the lidar data are extracted through a point cloud processing algorithm. Then, the multi-frame features in the time series are input into the Transformer encoder, and the self-attention mechanism is used to capture the dependency relationships (the movement trajectories of dynamic obstacles) between the data at different moments and different sensors, and a fused feature vector containing spatio-temporal context information is output to describe the dynamic changes of the environment around the robot. An autoencoder is used to construct a feature distribution model under normal operating conditions. The real-time fused features are input into the model, and abnormal events (equipment failures, personnel intrusions) are detected by calculating the reconstruction error. At the same time, the pre-trained ViT-BERT model is used to perform semantic segmentation on the fused scene image to identify the categories and positions of key objects (scaffolding boards, steel bars), and an environmental map with semantic labels is generated to provide interactive semantic information for task planning. Fuse the data of the robot's own sensors (encoders, force sensors) through extended Kalman filtering to estimate the state parameters of the robot in real time, such as position, attitude, and joint torque. Combine historical fault data to establish an equipment reliability model based on the hidden Markov model, predict parameters such as battery life and the degree of wear of the robotic arm, and output a comprehensive perception report including the robot's health status and environmental risk level.

[0028] In this embodiment, the specific extensions for the central control module include: Use the improved DBSCAN algorithm to cluster all micro-task units. Take the central coordinates of the task area as the spatial dimension and the construction time window as the time dimension to calculate the spatio-temporal distance between tasks; introduce density peak clustering to optimize the initial clustering center and identify task clusters with high correlation (wall masonry and concrete pouring tasks on the same floor); the subtasks within each task cluster have spatial proximity and process dependence, serving as the basic units for macro-layer task allocation. Use the building structure tree as the macro-layer planning framework to construct a task hierarchical dependency graph; when the ant colony searches in the structure tree, it preferentially selects task clusters on the critical path (main structure construction), and the pheromone update rule combines the construction progress delay risk (probability of weather impact) and resource occupancy cost (crane usage duration); generate the execution order of task clusters that conforms to the logic of building construction through ant colony iteration to ensure that tasks across floors and functional areas are sorted according to the principle of "from top to bottom, from main to decoration". For the subtasks within each task cluster, construct a robot-task matching matrix, where the matrix elements include the time cost for the robot to complete the task (operation efficiency model fitted based on historical data) and the ability matching degree (professional adaptation of the welding robot to steel structure tasks); when the ant colony searches at the micro-layer, each ant represents a robot and selects subtasks according to the roulette strategy, and the selection probability combines the pheromone concentration and the heuristic function (task urgency / robot's current load); avoid overloading a single robot task through local search optimization to ensure load balancing in task allocation. Divide the robots into sub-populations according to functional types, such as handling, masonry, spraying, etc. The ant colonies of different sub-populations achieve cooperation by sharing "collaborative pheromones across task types"; after the handling robot completes the material transportation, it releases high-concentration pheromones to the masonry robot sub-population to guide it to preferentially select the corresponding masonry task; when a robot failure or environmental mutation is detected, trigger local replanning: screen high-priority subtasks from the tasks not completed by the failed robot, quickly reallocate them through the elite ant colony, and update the pheromone matrix to reflect the latest task status.

[0029] In this embodiment, the specific extensions for the information sharing module include: Each robot's local maintenance includes a knowledge graph of task progress, equipment status, and environmental semantics. It periodically aggregates global knowledge through a federated learning framework. In the parameter aggregation stage, secure multi-party computation is used to encrypt the local knowledge vectors. The global knowledge graph feeds back to each robot to update its local construction knowledge base (standard process parameters, safety specifications). Establish a multi-attribute decision-making model to handle task execution conflicts: When two robots request the same resource (material stacking area), calculate the conflict resolution weight based on the task priority (urgency × duration weight), the current load of the robot, and the resource occupation duration. The robot with a lower priority enters the waiting queue. If a path planning conflict occurs, solve the optimal concession strategy through the Nash equilibrium algorithm in game theory, such as adjusting the traveling speed or detouring path, to ensure the minimization of the cooperation cost (time delay + increased energy consumption). Adopt multi-agent reinforcement learning to train the collaborative strategy of the robot team. Each robot observes the local environment (its own position, the positions of neighboring robots, and the distribution of obstacles), and shares key state information through an attention mechanism. In the trajectory planning stage, generate an initial path based on the artificial potential field method, and then optimize it through the Deep Deterministic Policy Gradient (DDPG) algorithm. Introduce a collision prediction module (spatiotemporal occupancy grid method) to detect the intersection points of the motion trajectories in real time, and dynamically adjust the speed and direction to ensure collision-free cooperation of multiple robots in the shared working area.

[0030] In this embodiment, the solution of the Nash equilibrium point includes: Pure strategy Nash equilibrium detection: Traverse all strategy combinations to check if there is a combination such that no participant can obtain a higher payoff by unilaterally changing the strategy. Mixed strategy Nash equilibrium calculation: If there is no pure strategy equilibrium, calculate the mixed strategy equilibrium through linear programming, that is, the optimal probability distribution for each participant to choose different strategies. Equilibrium point screening: If there are multiple Nash equilibrium points, select the Pareto-optimal equilibrium point (that is, no other equilibrium point can increase the payoffs of all participants simultaneously).

[0031] In this embodiment, the specific expansion of the operation evaluation module includes: Establish a digital twin model of the construction robot and the construction environment. Real-time collect the robot's position, joint angles, energy consumption data, and environmental parameters such as temperature, humidity, and dust concentration in the working area through the Internet of Things interface. Use the space mapping algorithm to convert the physical coordinates into the virtual coordinates of the digital twin body, and interpolate the missing data through the state estimation model to ensure that the state synchronization accuracy between the virtual space and the physical world reaches the centimeter level, and conduct multi-dimensional construction performance evaluation: Quality Dimension: Identify construction defects (wall verticality deviation, missing welds) through computer vision algorithms, and calculate the quality deviation rate in combination with the design parameters of the BIM model; Efficiency Dimension: Establish a construction process model based on queuing theory, analyze material transportation bottlenecks (elevator usage conflicts) and robot waiting times, and identify efficiency bottleneck points on the critical path; Safety Dimension: Use spatio-temporal collision detection algorithms to predict the collision probability between robots and personnel / obstacles, construct a safety risk assessment matrix in combination with historical accident data, and output the real-time safety warning level; Construct a causal diagram of the construction process, analyze the causal relationships between task allocation strategies (robot load balance), environmental parameters and construction efficiency / quality; when abnormal performance indicators are detected, simulate the counterfactual results of different intervention measures (reallocate tasks, adjust robot motion parameters) through causal inference algorithms, select the optimization plan that maximally improves performance, and generate control instructions including the parameter adjustment range and execution time.

[0032] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only preferred embodiments of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A collaborative operation system for construction robots, characterized in that, The system includes: A BIM modeling module, which is used to establish a three-dimensional model of a construction project using BIM technology and divide the BIM model into multiple operation areas; An information acquisition module, which is used to collect the status information of the robot and the operation environment information in real time through the sensors equipped on each robot; A central control module, which is used to receive the status information and operation environment information sent by each robot, combine the operation area division and operation requirements in the BIM model, and use the ant colony algorithm for task allocation and operation planning; An information sharing module, which is used for each robot to move to the corresponding operation area for operation according to the assigned tasks and operation plans. During the operation process, the robots communicate in real time through wireless communication to share the operation progress, status information and environment information; An operation evaluation module, which is used to monitor the operation progress and status of each robot in real time, evaluate the operation effect of the robot, and adjust the task allocation and planning strategy in time according to the evaluation results to optimize the collaborative operation process of the robot.

2. The collaborative operation system of a construction robot according to claim 1, wherein The BIM modeling module includes: A semantic segmentation sub-module, which is used to obtain the three-dimensional point cloud data of the BIM model, perform semantic segmentation on the three-dimensional point cloud data, extract geometric features and semantic features, and construct a multi-dimensional feature vector containing spatial position and structural semantics; A learning sub-module, which is used to abstract the three-dimensional point cloud data into nodes in a graph structure based on the multi-dimensional feature vector output by the semantic segmentation sub-module, and learn the spatial dependence relationship between nodes through a graph neural network to generate a graph representation containing structural logic; A division sub-module, which is used to perform multi-scale feature extraction on the BIM model using a three-dimensional convolutional neural network based on the graph representation output by the learning sub-module, and introduce a hierarchical clustering algorithm to perform hierarchical division of multi-scale spatial regions to divide into multiple operation areas.

3. The collaborative operation system of a construction robot according to claim 2, wherein, The division sub-module includes: Performing multi-scale feature extraction on the BIM model using a three-dimensional convolutional neural network. The three-dimensional convolutional neural network contains multiple convolutional layers and pooling layers. Local detail features are extracted through shallow convolutional layers, and global structural features are extracted through deep convolutional layers. Different-scale feature maps are integrated through a pyramid pooling module to generate a multi-scale feature representation; Based on the extracted multi-scale features, each small area is used as an initial clustering unit, the feature similarity between each small area is calculated using cosine similarity, and the clustering units with the highest similarity are merged in a bottom-up manner to form larger areas to obtain the area division result.

4. A collaborative operation system for construction robots according to claim 1, characterized in that, The information acquisition module includes: An establishment sub-module, which is used to align the timestamps of the original data of the sensors, establish the conversion relationship between the coordinate systems of each sensor and the robot body coordinate system, and generate a spatio-temporal alignment data set; A processing sub-module, which is used to construct a spatio-temporal attention network to process the multi-modal data in the spatio-temporal alignment data set, extract the spatial features of the visual image through a convolutional neural network, extract the three-dimensional spatial features of the lidar data through a point cloud processing algorithm, and use the self-attention mechanism of the Transformer encoder to capture the dependencies between data at different times and different sensors, and output a fusion feature vector containing spatio-temporal context information; An analysis sub-module, which is used to analyze the fused feature vector to obtain the status information of the robot and the working environment information.

5. The collaborative operation system of a construction robot according to claim 4, wherein The analysis sub-module includes: Construct a feature distribution model in the normal working state by using an autoencoder, input the fused feature vector into the fused feature vector, and detect abnormal events by calculating the reconstruction error; Use the ViT-BERT model to perform semantic segmentation on the fused feature vector, identify the category and location of the object, and generate an environmental map with semantic labels; Integrate the outputs of the feature distribution model and the ViT-BERT model to obtain the status information of the robot and the working environment information.

6. The collaborative operation system of a construction robot according to claim 1, wherein, The central control module includes: An initialization sub-module, which is used to decompose the construction engineering task into multiple sub-tasks, each sub-task corresponding to a working area, and initialize the ant colony parameters; A selection sub-module, where each ant represents a robot, and according to the pheromone concentration and heuristic information of each current sub-task, select the next sub-task to be executed. When an ant selects a task, it follows the roulette wheel selection strategy; An allocation sub-module, which is used to update the pheromone concentration between each sub-task according to the result of the task allocation after all ants complete a task allocation; An iteration sub-module, which is used to obtain the optimal task allocation plan after multiple iterations, allocate sub-tasks and working areas to each robot, and plan the working sequence and path of the robot.

7. The collaborative operation system of a construction robot according to claim 1, characterized in that, The ant colony parameters at least include the number of ants, the initial concentration of pheromone, the heuristic factor, and the pheromone evaporation factor.

8. The collaborative operation system of a construction robot according to claim 1, characterized in that, The information sharing module includes: An update sub-module, which is used to update the neighbor node list in real time by signal strength detection when the robot moves, and dynamically adjust the communication link by using the Q-learning algorithm to form an adaptive Mesh network topology; A calculation sub-module, which is used to calculate the conflict resolution weight according to the task priority, the current load of the robot, and the resource occupation duration when two robots request the same resource, and the robot with a lower priority enters the waiting queue; An implementation sub-module, which is used to obtain the optimal concession strategy through the Nash equilibrium algorithm in case of a path planning conflict, and realize real-time communication and cooperative control of multiple robots.

9. The collaborative operation system of a construction robot according to claim 8, characterized in that, The implementation sub-module includes: Each robot uses the Kalman filter to predict the sequence of trajectory points within the future time window based on the current position, speed, and target path, and generates a spatio-temporal occupancy grid map; Spatially superimpose the spatio-temporal occupancy grid maps of all robots, identify the areas with spatio-temporal overlap, and construct a conflict state vector; Based on the conflict state vector, define an optional strategy set for each robot, calculate the benefits of the robot for each strategy combination, and construct a multi-dimensional benefit matrix; Traverse the multi-dimensional benefit matrix to solve for the Nash equilibrium point, determine the final executed Nash equilibrium strategy combination, and obtain the optimal concession strategy.

10. A method for implementing a cooperative operation system of a construction robot as described in claim 1, characterized in that, This method includes the following steps: Use BIM technology to establish a three-dimensional model of the construction project, and divide the BIM model into multiple working areas; Real-time collect the status information of the robot and the working environment information through the sensors equipped on each robot; Receive the status information and working environment information sent by each robot, combine the working area division and working requirements in the BIM model, and use the ant colony algorithm for task allocation and operation planning; Each robot moves to the corresponding working area for operation according to the assigned tasks and operation plans. During the operation process, the robots communicate in real time through wireless communication to share the operation progress, status information and environment information; Monitor the operation progress and status of each robot in real time, evaluate the operation effect of the robot, and according to the evaluation results, adjust the task allocation and planning strategy in a timely manner to optimize the collaborative operation process of the robot.

Images