Heterogeneous robot distributed collaborative assembly method based on intelligent prefabricated parts

Abstract

The invention discloses a heterogeneous robot distributed collaborative assembly method based on an intelligent prefabricated part. The method comprises the steps that an assembly task is started, the intelligent prefabricated part senses the state of the intelligent prefabricated part, the task requirement is autonomously determined, and a task request is issued to a heterogeneous robot; the robot judges whether to respond or not according to a self-ability model; the component selects an optimal response robot and sends a task execution instruction; the robot executes the task and feeds back a result, and the component records and uploads the result to the fabricated building component management platform; a distributed collaborative framework with intelligent prefabricated parts as decision-making cores is constructed, and a closed-loop process of task release, robot bidding, part decision-making and robot execution and feedback is realized; the method is suitable for multi-robot cooperative assembly in a large-scale complex scene, the scheduling bottleneck and the calculation pressure of centralized control are effectively relieved, the system expansibility and the resource utilization efficiency are improved, and the cooperative capability and the overall assembly performance of heterogeneous robots are remarkably enhanced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent construction, and particularly relates to a heterogeneous robot distributed collaborative assembly method based on intelligent prefabricated components. BACKGROUND

[0002] In recent years, prefabricated buildings have rapidly developed due to high construction efficiency and controllable quality. In order to improve the assembly efficiency, multi-robot collaborative work has been gradually introduced into the prefabricated component assembly task, especially in the component transportation, positioning, hoisting and assembly links. However, the existing multi-robot collaborative assembly method mostly adopts a centralized control mode. With the increase of the assembly task scale and the number of robots, obvious deficiencies are exposed: ① the central control unit has heavy computing burden, resulting in slow real-time response speed; ② task conflicts easily occur when the number of robots increases, resulting in waste of system resources; ③ the overall control strategy lacks flexibility and cannot respond to changes in the field environment in time, making it difficult to effectively improve the assembly efficiency.

[0003] Therefore, it is urgent to propose a new control method suitable for large-scale tasks and heterogeneous robot collaborative work to improve the operation efficiency, resource utilization and flexible response ability of the system. SUMMARY

[0004] The application proposes a heterogeneous robot distributed collaborative assembly method based on intelligent prefabricated components, which solves the efficiency bottleneck in large-scale tasks under the centralized control mode and improves the real-time performance, collaboration and scalability of the system.

[0005] Technical scheme: The heterogeneous robot distributed collaborative assembly method based on intelligent prefabricated components comprises the following steps:

[0006] After the assembly task is started, the intelligent prefabricated component real-time perceives the state information of itself; the intelligent prefabricated component comprises a task flow model unit, an RFID tag, an inertial measurement unit, an ultra-wideband positioning device and a lightweight AI inference model; the RFID tag, the inertial measurement unit and the ultra-wideband positioning device respectively acquire component identity information, position information, attitude information and assembly progress information; the task flow model unit is used for pre-storing and managing a BIM assembly task model;

[0007] The intelligent prefabricated component autonomously determines the assembly task demand according to the BIM assembly task model and issues a task request instruction to the heterogeneous robot in the corresponding work area; the task request is broadcast in real time to the heterogeneous robot through a communication network and enters a task listening stage;

[0008] After receiving the task request instruction of the intelligent prefabricated component, the heterogeneous robot autonomously determines whether to match the task request instruction according to the preset capability model of the robot; if the matching is successful, the confirmation information is fed back to the intelligent prefabricated component, and the state of the robot is automatically adjusted, and the assembly instruction is waited for

[0009] When receiving the task response of the heterogeneous robot, the intelligent prefabricated component adopts a lightweight AI inference model to prioritize all responding robots, and only sends a task execution instruction to the target robot with the highest score, and sends a task release signal to other responding robots to release the standby state

[0010] The target robot completes the assembly task according to the task execution instruction, and feeds back the task execution result to the intelligent prefabricated component; the intelligent prefabricated component records the assembly task completion and uploads it to the prefabricated building component management platform for storage and analysis.

[0011] Further, the BIM assembly task model is constructed in the form of combining BIM model data and component unique code to form a "component-task" mapping table; the mapping table contains the unique code of the prefabricated component and the corresponding specific assembly task list, including assembly task type, assembly position coordinates, key operation precision requirement, task priority and task execution time window and other information, which is used to clearly describe the specific assembly task and sequence of each component. The assembly task model is defined as follows:

[0012] M={(C i ,T i )|i∈[1,n]}

[0013] Where C i represents the component number, which is generated by the BIM component code; T i represents the assembly task set of component C i , which is arranged in order according to the assembly process; each t j ∈T i , contains task type, assembly position, precision requirement, priority and execution time window fields, and is defined as:

[0014] T i =[t1,t2,...,t m ]

[0015] t j =(type,pos,acc,prio,window)。

[0016] Further, the intelligent prefabricated component adopts a finite state machine to model the component task flow based on the BIM assembly task model, and converts the task state according to the current perception state, generates an assembly task request instruction, and realizes dynamic execution and update of the task flow driven.

[0017] each component C i corresponding finite state machine FSM i defined as:

[0018] FSM i = (S i , T i , G i , s i,0 )

[0019] wherein S i is a state set, including "to be assembled", "in task request", "in matching", "in execution", "completion confirmation"; T i is an assembly task set, indicating specific assembly tasks that can be triggered in each state; G i is a state transition function, which determines the next state according to the current state and task triggering condition; s i,0 is the initial state of the component, which is "to be assembled";

[0020] According to the current perception state s i of the component C t , the matching task t j , the state is updated as:

[0021] s t+1 = G i (s t , t j )

[0022] At the same time, according to the task mapping table, the task request is issued: if s t ∈ S i and "perception condition is met", the task t j is triggered, and the task request instruction is generated; the "perception condition is met" is that the state information collected by the component through its own perception device is consistent with the task triggering condition set in the BIM task mapping table.

[0023] Further, the communication mode between the intelligent prefabricated component and the heterogeneous robot includes WiFi, ZigBee or LoRa; the communication adopts a distributed publish-subscribe mode, a distributed communication network is built, and the communication interaction between the intelligent prefabricated component and the heterogeneous robot is realized.

[0024] Further, the heterogeneous robot includes a palletizing robot, a transportation robot, a hoisting robot, and an assembly robot, each of which has an independent capability model.

[0025] Further, the capability model includes robot type, maximum work load, key work precision, current state information, and endurance capability; the capability model Ri is defined as a five-tuple:

[0026] R i = (type i , load i , acc i , state i , power i ).

[0027] Further, after receiving the task request instruction, the heterogeneous robot performs adaptability evaluation on the task request by adopting a condition judgment-based rule, requiring that each index of the capability model meets the minimum requirement set by the task parameters; if all conditions are met, it is determined to be adaptable, and the component is fed back task confirmation information; otherwise, it does not respond; the condition judgment-based rule is that, specifically, the robot compares the type matching, load capacity, accuracy requirement, working state and endurance capability in the task request t j with the capability model R i maintained locally, and if , the robot is adaptable.

[0028] Further, the lightweight AI inference model is a multi-layer feedforward neural network model deployable at the edge, including an input layer, a hidden layer and an output layer; specifically, the input layer includes five feature variables, namely, the robot capability matching degree r i , the distance d i between the component and the robot, the task load rate l i , the response time delay t i and the historical task completion quality score q i , constituting an input vector X i = [r i , d i , l i , t i , q i ]; the model inference calculation process is as follows:

[0029] H i = ReLU(W (1) X i + b (1) )

[0030] S i = σ(W (2) H i + b (2) )

[0031] wherein W (1) is a hidden layer weight matrix, W (2) is an output layer weight vector, and b (1) , b(2) H is the corresponding bias vector, i is the hidden layer output vector after the i-th robot input, and σ is a Sigmoid activation function, i is the i-th robot robot task adaptation score, indicating the matching degree of the robot to the current component task; the intelligent prefabricated component selects the one with the highest score according to the score results of multiple responding robots to execute the task.

[0032] Further, a single task locking mechanism is arranged in the heterogeneous robot, that is, each heterogeneous robot can only respond to one component at any time, send a locking confirmation information, and reject or ignore the task request of other components. After receiving the release information, it restores the standby state and can respond to the task request of the next component.

[0033] Further, the assembly management platform is a BIM-based data management platform for storing, analyzing and optimizing the data fed back by the intelligent prefabricated component during the assembly task execution process; a contract net protocol is used for task request distribution and coordination, including task broadcast, robot bidding, component evaluation and task allocation, robot execution and result feedback, and a closed-loop distributed collaborative control mechanism is constructed.

[0034] Beneficial effects: Compared with the prior art, the beneficial effects of the present application are: the above-mentioned heterogeneous robot distributed collaborative assembly method based on intelligent prefabricated component provided by the present application takes intelligent prefabricated component as the center node of task decision and scheduling, generates task request by itself through real-time sensing state information, accurately and dynamically allocates heterogeneous robots based on built-in AI inference model, forms a distributed task-driven collaborative control mechanism, effectively solves the scheduling bottleneck, calculation pressure concentration, poor scalability and low resource utilization of traditional centralized control mode, and significantly improves the collaborative efficiency of heterogeneous robots and the overall assembly performance; at the same time, it effectively reduces the dependence of assembly task on on-site manpower, helps to realize less-personalized or unmanned intelligent construction, actively responds to the labor shortage challenge in the construction industry, and promotes the integrated development of building industrialization and intelligent construction. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 is the flowchart of the present application;

[0036] Figure 2 is the collaborative assembly construction scene schematic diagram of intelligent prefabricated component and heterogeneous robot interaction proposed by the present application;

[0037] Figure 3 is the distributed collaborative control mechanism diagram of heterogeneous robot based on intelligent prefabricated component proposed by the present application. DETAILED DESCRIPTION

[0038] The application will be described in further detail below with reference to the drawings:

[0039] As shown in the drawings, Figure 1 The application proposes a heterogeneous robot distributed collaborative assembly method based on intelligent prefabricated components, including the following steps:

[0040] Step 1: After the start of the assembly task, the intelligent prefabricated component perceives its state information in real time.

[0041] The intelligent prefabricated component is embedded with an RFID tag, an inertial measurement unit, and an ultra-wideband (UWB) positioning device for real-time collection of its identity, attitude, position information, and assembly progress information.

[0042] Step 2: The intelligent prefabricated component autonomously determines the assembly task requirements based on the pre-stored BIM assembly task model and issues task request instructions to the task requirement-related heterogeneous robots; the task request is broadcast to the heterogeneous robots through a communication network and enters the task listening stage.

[0043] The collaborative assembly construction scene of the intelligent prefabricated component interacting with the heterogeneous robot is shown in the drawings: Figure 2 The assembly construction scene is composed of three areas: a material yard, a hoisting area, and an assembly area; the assembly construction object is a plurality of intelligent prefabricated components; the heterogeneous robots include stacking robots, transport robots, hoisting robots, and assembly robots, and the number of each type of robot is not fixed. In order to reduce the calculation difficulty and improve the decision-making efficiency, the prefabricated component will only issue task requests to the task-related heterogeneous robots in the corresponding area when it is working in a specific area.

[0044] The interaction relationship between the intelligent prefabricated component and the heterogeneous robot is as follows: the stacking robot picks up the component in the material yard and places it on the transport robot AGV; after the component is loaded, the transport robot transports the component to the hoisting area. The hoisting robot hoists the component on the transport robot and transports it to the assembly area; after the component is hoisted, the transport robot can return. The hoisting robot hoists the component to the assembly point and completes the positioning and assembly tasks with the assembly robot; after the component is positioned and fixed, the hoisting robot can leave.

[0045] The intelligent prefabricated component is built-in with a task flow model unit for pre-storing and managing the BIM assembly task model. The BIM assembly task model is constructed in the form of a "component-task" mapping table by combining BIM model data with component unique codes; the mapping table contains the unique codes of prefabricated components and the corresponding specific assembly task list, including assembly task type, assembly position coordinates, key operation precision requirements, task priority, and task execution time window information, for clearly describing the specific assembly tasks and sequence of each component. The assembly task model is defined as follows:

[0046] M = {(C i , T i )|i∈[1,n]}

[0047] where C i represents the component number, generated by BIM component coding; T i represents the assembly task set of component C i , arranged in order according to the assembly process; each t j ∈T i , contains task type, assembly position, accuracy requirement, priority and execution time window fields, defined as:

[0048] T i = [t1,t2,...,t m ]

[0049] t j = (type,pos,acc,prio,window)

[0050] Intelligent prefabricated components based on BIM assembly task model, using finite state machine (Finite State Machine, FSM) to model the component task flow, and according to the current perception state to convert task state, generate assembly task request instruction, realize the dynamic execution and update of task flow driven. Each component C i corresponding finite state machine FSM i is defined as:

[0051] FSM i = (S i ,T i ,G i ,s i,0 )

[0052] Where: S i is the state set, including "to be assembled", "in task request", "in matching", "in execution", "completion confirmation" and other typical states; T i is the assembly task set, indicating the specific assembly task that can be triggered in each state; G i is the state transition function, which determines the next state according to the current state and task trigger condition; s i,0 is the initial state of the component, usually "to be assembled".

[0053] State machine driven logic is: according to the current perception state s i of component C t , matching task t j , state update is:

[0054] s t+1 = Gi (s t ,t j )

[0055] Meanwhile, according to the task mapping table, a task request is issued: If s t ∈S i and the perception condition is met, trigger task t j , generate a task request instruction. The "perception condition is met" is that the state information collected by the component through its own perception device is consistent with the task trigger condition set in the BIM task mapping table.

[0056] Step 3: After receiving the task request instruction of the intelligent prefabricated component, the heterogeneous robot autonomously determines whether to match the task request instruction according to its own preset capability model; if the matching is successful, it feeds back the confirmation information to the intelligent prefabricated component and automatically adjusts its own state, waiting for the assembly instruction.

[0057] The intelligent prefabricated component broadcasts its assembly task demand t j to the heterogeneous robot in real time; each robot has an independent capability model. The capability model includes robot type, maximum working load, working accuracy, current state information, and endurance capability. The capability model R i of each heterogeneous robot is defined as a five-tuple:

[0058] R i =(type i ,load i ,acc i ,state i ,power i )

[0059] After receiving the task request instruction, the heterogeneous robot performs adaptability evaluation on the task request by using the condition judgment-based rule, which requires that each index of its capability model meets the minimum requirement set by the task parameters; if all conditions are met, it is determined to be adaptable, and the task confirmation information is fed back to the component; otherwise, it does not respond. Based on the condition judgment-based rule, specifically, the robot compares the type matching, load capacity, accuracy requirement, working state, and endurance capability in the task request t j with the local maintained capability model R i item by item, and if then the robot is adaptable.

[0060] Step 4: After receiving the confirmation information from the heterogeneous robot, the intelligent prefabricated component uses the embedded lightweight AI inference model to score multiple responding robots, determines the optimal robot to execute the task, and sends a task execution instruction to the determined robot.

[0061] When receiving multiple robot task responses, the intelligent prefabricated component prioritizes all responding robots using an embedded lightweight AI inference model, and only sends task execution instructions to the target robot with the highest score, while sending task release signals to other responding robots to release the standby state.

[0062] The lightweight AI inference model is a multi-layer perceptron (MLP) edge-deployable neural network model, which is composed of an input layer, a hidden layer and an output layer. Specifically, the input layer includes five feature variables, namely robot capability matching degree r i , component-robot distance d i , task load rate l i , response delay t i , and historical task completion quality score q i , which constitute an input vector X i =[r i , d i , l i , t i , q i ]. The model inference calculation process is as follows:

[0063] H i =ReLU(W (1) X i +b (1) )

[0064] S i =σ(W (2) H i +b (2) )

[0065] Where W (1) is the hidden layer weight matrix, W (2) is the output layer weight vector, b (1) and b (2) are the corresponding bias vectors, H i is the hidden layer output vector after input of the i-th robot, σ is the Sigmoid activation function, S i ∈[0,1] is the i-th robot task adaptation score, indicating the matching degree of the robot to the current component task; the intelligent prefabricated component selects the highest score to execute the task according to the score results of multiple responding robots; the AI inference model is trained offline and compressed to be burned into the MCU embedded in the component, supporting on-site rapid inference and autonomous assignment, and realizing the end-side intelligence of the component.

[0066] Step 5: The selected robot completes the assembly task according to the task execution instruction, and feeds back the task execution result to the intelligent prefabricated component; the intelligent prefabricated component records the completion of the assembly task, and uploads it to the assembly management platform for storage and analysis.

[0067] The communication mode between the intelligent prefabricated component and the heterogeneous robot includes but is not limited to WiFi, ZigBee or LoRa; the communication adopts a distributed publish-subscribe mode, and a distributed communication network is constructed for realizing the communication interaction between the intelligent prefabricated component and the heterogeneous robot.

[0068] In order to avoid the repeated response of the robot or the resource conflict when multiple intelligent prefabricated components broadcast task requests at the same time, a single task locking mechanism is set for each heterogeneous robot, that is, the robot can only respond to one component at any time, sends a locking confirmation information, and refuses or ignores the task request of other components, and after receiving the release information, the robot restores the standby state, and can respond to the task request of the next component.

[0069] The assembly management platform is a BIM-based data management platform for storing, analyzing and optimizing the data generated in the assembly task execution process. The improved contract network protocol (CNP) is used for task request distribution and coordination, and the protocol process includes task broadcasting, robot bidding, component evaluation and task allocation, robot execution and result feedback, and a closed-loop distributed collaborative control mechanism is constructed, as shown in Figure 3 .

[0070] In summary, the present application effectively solves the scheduling bottleneck, the centralized calculation pressure, the poor expansibility and the low resource utilization rate of the traditional centralized control mode, has the potential to improve the collaborative efficiency of the heterogeneous robot and the overall assembly performance; at the same time, it effectively reduces the dependence of the assembly task on the on-site manual work, helps to realize the few-person or unmanned intelligent construction, actively responds to the labor shortage challenge in the construction industry, and promotes the integrated development of the building industrialization and the intelligent construction.

[0071] The above only illustrates the technical idea of the present application, and cannot limit the protection scope of the present application, any modification made on the basis of the technical scheme belongs to the technical idea proposed by the present application, and falls within the protection scope of the claims of the present application.

Claims

1. A distributed collaborative assembly method for heterogeneous robots based on intelligent prefabricated components, characterized in that, The implementation process is as follows: After the assembly task is started, the intelligent prefabricated component senses its own status information in real time; the intelligent prefabricated component includes a task flow model unit, an RFID tag, an inertial measurement unit, an ultra-wideband positioning device, and a lightweight AI inference model; the RFID tag, inertial measurement unit, and ultra-wideband positioning device respectively acquire the component's identity information, location information, attitude information, and assembly progress information. The task flow model unit is used to pre-store and manage BIM assembly task models; The intelligent prefabricated components autonomously determine the assembly task requirements based on the BIM assembly task model and issue task request instructions to the heterogeneous robots in the corresponding work areas. The task request is broadcast to the heterogeneous robot in real time via the communication network, and the task listening phase begins. After receiving a task request instruction from a smart prefabricated component, the heterogeneous robot autonomously determines whether it matches the task request instruction based on its own preset capability model. If the match is successful, it sends a confirmation message to the intelligent prefabricated component and automatically adjusts its own status, waiting for assembly instructions. When receiving task responses from heterogeneous robots, the intelligent prefabricated component uses a lightweight AI inference model to prioritize all responding robots and sends task execution instructions only to the target robot with the highest score. At the same time, it sends task release signals to the other responding robots to deactivate their standby status. The target robot completes the assembly task according to the task execution instructions and feeds back the task execution results to the intelligent prefabricated component; the intelligent prefabricated component records the completion status of the assembly task and uploads it to the prefabricated building component management platform for storage and analysis.

2. The method for distributed collaborative assembly of heterogeneous robots based on intelligent prefabricated components according to claim 1, characterized in that, The BIM assembly task model is constructed by combining BIM model data with unique component codes to form a "component-task" mapping table. This mapping table contains the unique codes of prefabricated components and the corresponding list of specific assembly tasks, including information such as assembly task type, assembly location coordinates, critical operation accuracy requirements, task priority, and task execution time window. It is used to clearly describe the specific assembly tasks and sequence of each component. The assembly task model is defined as follows: M={(C i ,T i )|i∈[1,n]} Where C i Indicates the component number, generated from the BIM component code; T i Indicates component C i The assembly task set is arranged in order of assembly process; each t j ∈T i It includes fields such as task type, assembly location, accuracy requirements, priority, and execution time window, and is defined as follows: T i =[t1,t2,...,t m ] t j =(type,pos,acc,prio,window)。 3. The method for distributed collaborative assembly of heterogeneous robots based on intelligent prefabricated components according to claim 2, characterized in that, The intelligent prefabricated components are based on the BIM assembly task model. The finite state machine is used to model the component task process, and the task state is changed according to the current perception state to generate assembly task request instructions, so as to realize the dynamic execution and update driven by the task process. Each component C i The corresponding finite state machine (FSM) i Defined as: FSM i =(S i ,T i ,G i ,s i,0 ) Among them, S i This is a set of states, including "Pending Assembly", "Task Requesting", "Matching", "Executing", and "Completion Confirmation"; T i G represents the set of assembly tasks, indicating the specific assembly tasks that can be triggered in each state; i This is a state transition function that determines the next state based on the current state and the task triggering conditions; s i,0 This is the initial state of the component, designated as "to be assembled". According to component C i Current perception state s t Matching task t j The status has been updated to: s t+1 =G i (s t ,t j ) Simultaneously, a task request is issued based on the task mapping table: if s t ∈S i And if the "perception condition is met", then task t is triggered. j The task request instruction is generated; the "sense condition is met" means that the status information collected by the component through its own sensing device is consistent with the task triggering conditions set in the BIM task mapping table.

4. The method for distributed collaborative assembly of heterogeneous robots based on intelligent prefabricated components according to claim 1, characterized in that, The communication methods between intelligent prefabricated components and heterogeneous robots include WiFi, ZigBee, or LoRa; the communication adopts a distributed publish-subscribe model to build a distributed communication network and realize the communication interaction between intelligent prefabricated components and heterogeneous robots.

5. The method for distributed collaborative assembly of heterogeneous robots based on intelligent prefabricated components according to claim 1, characterized in that, The heterogeneous robots include palletizing robots, transport robots, hoisting robots, and assembly robots, each with its own independent capability model.

6. The method for distributed collaborative assembly of heterogeneous robots based on intelligent prefabricated components according to claim 5, characterized in that, The capability model includes robot type, maximum workload, critical task accuracy, current status information, and endurance; the capability model R for each heterogeneous robot... i Defined as a quintuple: R i =(type i ,load i ,acc i ,state i ,power i )。 7. The method for distributed collaborative assembly of heterogeneous robots based on intelligent prefabricated components according to claim 1, characterized in that, After receiving the task request instruction, the heterogeneous robot uses a condition-based judgment rule to evaluate the adaptability of the task request, requiring that all indicators of its capability model meet the minimum requirements set by the task parameters; if all conditions are met, it is determined to be adaptable and sends task confirmation information back to the component. Otherwise, no response is made; specifically, the rule based on conditional judgment refers to the robot's local maintained capability model R. i According to the task request t j The type matching, load capacity, accuracy requirements, working status and endurance are compared item by item. If the type i =type tj and load i ≥load tj min andacc i ≤acc tj max and state i = "Idle" and power i ≥power min Then the robot will adapt.

8. The method for distributed collaborative assembly of heterogeneous robots based on intelligent prefabricated components according to claim 1, characterized in that, The lightweight AI inference model is an edge-deployable multi-layer feedforward neural network model, including an input layer, hidden layers, and an output layer; specifically, the input layer includes five feature variables, namely the robot capability matching degree r. i d, the distance between the component and the robot i Task load rate i Response delay t i And the historical mission completion quality score q i , forming the input vector X i =[r i ,d i ,l i ,t i ,q i The model inference and calculation process is as follows: H i =ReLU(W (1) X i +b (1) ) S i =σ(W (2) H i +b (2) ) Among them, W (1) W is the hidden layer weight matrix. (2) b is the output layer weight vector. (1) b (2) For the corresponding bias vector, H i Let S be the hidden layer output vector after the input of the i-th robot, and σ be the sigmoid activation function. i ∈[0,1] represents the robot task adaptation score of the i-th robot, indicating the degree of matching between the robot and the current component task; the intelligent prefabricated component selects the robot with the highest score based on the scoring results of multiple responding robots to perform the task.

9. A method for distributed collaborative assembly of heterogeneous robots based on intelligent prefabricated components according to claim 1, characterized in that, The heterogeneous robot is equipped with a single task locking mechanism, which means that each heterogeneous robot can only respond to one component at any given time, send a lock confirmation message, and reject or ignore task requests from other components. Only after receiving a release message and returning to standby state can it respond to the task request from the next component.

10. A method for distributed collaborative assembly of heterogeneous robots based on intelligent prefabricated components according to claim 1, characterized in that, The assembly management platform is a BIM-based data management platform used to store, analyze, and optimize data fed back by intelligent prefabricated components during the assembly task execution process; The Contract Network protocol is used for task request distribution and coordination, including task broadcasting, robot bidding, component evaluation and task assignment, robot execution and result feedback, to build a closed-loop distributed collaborative control mechanism.

Images