Multi-task collaborative robot scheduling method and system for intelligent factory

Abstract

The invention discloses a multi-task collaborative robot scheduling method and system oriented to an intelligent factory, and the method comprises the steps: receiving the input data of a plurality of tasks to be executed by the intelligent factory, the input data comprising the task type, the dependency relationship between the tasks, the task priority, the resource demand of each task, and the expected completion time of each task; determining an initial task execution sequence of the tasks according to the task types, the dependency relationships and the task priorities; obtaining current resource state data of all available robots in the intelligent factory; according to the resource demand of each task, the expected completion time of each task, the initial task execution sequence and the current resource state data of all available robots, a scheduling algorithm is adopted to generate a multi-task cooperative scheduling rule; and controlling the plurality of target robots to execute the plurality of tasks according to the multi-task cooperative scheduling rule. According to the method, the resource utilization rate of the robot in the intelligent factory can be effectively improved, and the task execution efficiency is optimized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent manufacturing technology, and in particular to a multi-task collaborative robot scheduling method and system for intelligent factories. Background Art

[0002] With the rapid development of Industry 4.0 and smart manufacturing, smart factories have become an important development direction of modern manufacturing. In smart factories, robots, as important execution units, undertake various complex production tasks. However, with the diversification and complexity of production tasks, how to efficiently coordinate and schedule multiple robots to perform multiple tasks at the same time has become a key challenge in the operation of smart factories.

[0003] Traditional robot scheduling methods usually adopt fixed scheduling strategies, which are difficult to adapt to dynamically changing production environments and task requirements. These methods often fail to fully consider the dependencies between tasks, priority differences, and changes in resource requirements, resulting in low resource utilization and inefficient task execution. In addition, traditional methods often lack flexibility and scalability when dealing with large-scale and diverse tasks, making it difficult to meet the needs of smart factories for efficient collaborative production.

[0004] The inventors found that although some studies have begun to focus on the problem of multi-robot collaborative scheduling, there are still many limitations. For example, some methods fail to fully consider the dynamic characteristics of the task and the resource status of the robot, resulting in less than optimal scheduling results; other methods consider the task priority, but ignore the dependency between tasks, resulting in unreasonable task execution order. Therefore, how to design a multi-task collaborative robot scheduling method that can comprehensively consider multiple factors and dynamically adapt to the production needs of smart factories has become the focus and difficulty of current research. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a multi-task collaborative robot scheduling method and system for a smart factory to solve the above technical problems.

[0006] To achieve the above object, in a first aspect, a multi-task collaborative robot scheduling method for a smart factory is provided, which comprises the following steps:

[0007] Receiving input data of a plurality of tasks to be performed by the smart factory, wherein the input data of each task includes: task type, dependencies between tasks, task priority, resource requirements of each task, and expected completion time of each task;

[0008] Determining an initial task execution order of the plurality of tasks according to the task types, dependencies between the tasks and the task priorities;

[0009] Get the current resource status data of all available robots in the smart factory;

[0010] Generate a multi-task collaborative scheduling rule using a scheduling algorithm according to the resource requirements of each task, the expected completion time of each task, the initial task execution order, and the current resource status data of all available robots;

[0011] According to the multi-task collaborative scheduling rule, multiple target robots dynamically determined from all available robots in the smart factory are controlled to perform the multiple tasks.

[0012] In a second aspect, a multi-task collaborative robot scheduling system for a smart factory is provided, the system comprising:

[0013] A task input module is used to receive input data of multiple tasks to be executed by the smart factory, wherein the input data of each task includes: task type, dependencies between tasks, task priority, resource requirements of each task and expected completion time of each task;

[0014] An initial order determination module, used to determine an initial task execution order of a plurality of tasks according to the task types, dependencies between the tasks and the task priorities;

[0015] Resource status acquisition module, used to obtain the current resource status data of all available robots in the smart factory;

[0016] A scheduling rule generation module, used to generate multi-task collaborative scheduling rules using a scheduling algorithm according to the resource requirements of each task, the expected completion time of each task, the initial task execution order, and the current resource status data of all available robots;

[0017] The task execution control module is used to control multiple target robots dynamically determined from all available robots in the smart factory to execute the multiple tasks according to the multi-task collaborative scheduling rules.

[0018] In a third aspect, an electronic device is provided, comprising:

[0019] one or more processors;

[0020] a storage device for storing one or more programs,

[0021] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in the first aspect.

[0022] According to a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method described in the first aspect is implemented.

[0023] The above technical solution has the following beneficial technical effects:

[0024] The method can comprehensively consider multiple factors such as task type, inter-task dependency, task priority, resource requirements, and expected completion time, and generate optimized multi-task collaborative scheduling rules through a dynamic scheduling algorithm. The method of the invention can effectively improve the resource utilization of robots in smart factories, optimize task execution efficiency, and enhance the flexibility and scalability of the production system, thereby meeting the needs of smart factories for efficient, collaborative, and flexible production.

[0025] This method improves the production efficiency of the smart factory and reduces unnecessary waiting time and resource waste by optimizing the task execution order and resource allocation. This method enhances the flexibility and robustness of the production system, can quickly respond to changes and emergencies in the production environment, and ensure the continuity and stability of production. In addition, by comprehensively considering multiple factors for scheduling, this method improves the resource utilization of the entire smart factory and reduces the problems of idle resources and over-allocation. In addition, the dynamic characteristics of this method enable smart factories to better cope with complex and changing production needs and improve the overall production quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0027] Figure 1 is a flow chart of a multi-task collaborative robot scheduling method for a smart factory according to an embodiment of the present invention;

[0028] Figure 2 is a specific flow chart of step S10 of an embodiment of the present invention;

[0029] Figure 3 is a specific flow chart of step S20 of an embodiment of the present invention;

[0030] Figure 4 is a specific flow chart of step S30 of an embodiment of the present invention;

[0031] Figure 5 is a specific flow chart of step S40 of an embodiment of the present invention;

[0032] Figure 6 is a specific flow chart of step S50 of an embodiment of the present invention;

[0033] Figure 7 It is a functional block diagram of a multi-task collaborative robot scheduling method for a smart factory according to an embodiment of the present invention;

[0034] Figure 8It is a schematic diagram of the structure of a computer system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0036] Embodiment 1

[0037] like Figure 1 As shown, this embodiment provides a multi-task collaborative robot scheduling method for a smart factory, which includes the following steps:

[0038] S10: receiving input data of a plurality of tasks to be executed in the smart factory, wherein the input data of each task includes: task type, dependencies between tasks, task priority, resource requirements of each task, and expected completion time of each task.

[0039] In this step, the central control system of the smart factory automatically receives input data of multiple tasks to be executed through the production management system or production planning system. The input data of each task contains detailed information such as the task type (welding, assembly, handling, spraying, etc.), the dependencies between tasks (e.g., task A must be completed before task B), the task priority (expressed by numbers 1-10, 10 is the highest priority), the resource requirements of each task (e.g., the required robot type, tools, materials, etc.), and the expected completion time of each task (accurate to the minute). For example, the system receives the following tasks: Task T001 (welding task, priority 8, no dependency, requires a welding robot and a welding gun, expected to be completed at 14:30), Task T002 (assembly task, priority 7, dependent on T001, requires an assembly robot and a screwdriver, expected to be completed at 16:00), and Task T003 (handling task, priority 9, no dependency, requires a handling robot and a forklift, expected to be completed at 15:00).

[0040] S20: Determine an initial task execution order of the plurality of tasks according to the task types, the dependencies between the tasks and the task priorities.

[0041] In this step, the system uses the task sorting algorithm to determine the initial task execution order based on the received task data. The algorithm considers the task type, dependency, and priority to generate the optimal initial execution order. In the above example, the system first checks the dependency and notices that T002 depends on T001 to complete. Then consider the task priority, T003 has the highest priority (priority 9), followed by T001 (priority 8), and finally T002 (priority 7). Finally, the system considers the task type and expected completion time to optimize the overall execution efficiency. Based on these factors, the system will derive the following initial task execution order: T003→T001→T002. This order not only meets the dependency requirements, but also takes into account the priority of the task, while also taking into account the expected completion time, so that the high-priority transport task (T003) can start as early as possible.

[0042] S30: Obtain current resource status data of all available robots in the smart factory.

[0043] In this step, the system obtains and updates the current resource status data of all available robots in real time through the factory's IoT system. This data can include information such as robot ID, robot type, current status of the robot, location of the robot, battery power of the robot, installed tools of the robot, recently completed tasks, and current load capacity. For example, the factory has the following available robots: R001 (welding robot, currently idle, located in area A, battery power 85%, welding gun installed), R002 (assembly robot, currently busy, expected to be idle in 15 minutes, located in area B, battery power 70%, screwdriver installed), R003 (handling robot, currently idle, located in area C, battery power 90%, forklift installed).

[0044] S40: Generate multi-task collaborative scheduling rules using a scheduling algorithm according to the resource requirements of each task, the expected completion time of each task, the initial task execution order, and the current resource status data of all available robots.

[0045] In this step, the system uses advanced scheduling algorithms, such as genetic algorithms, ant colony algorithms, or reinforcement learning algorithms, to comprehensively consider all relevant factors to generate multi-task collaborative scheduling rules. The goal of the algorithm is to minimize the overall completion time, maximize resource utilization, and meet the deadline requirements of all tasks. In this embodiment, considering the previously determined initial task execution order (T003→T001→T002) and the current state of the robot, the scheduling algorithm will generate the following collaborative scheduling rules: Task T003 (handling) is assigned to robot R003 and starts execution immediately; Task T001 (welding) is assigned to robot R001 and starts immediately after R003 completes T003; Task T002 (assembly) is assigned to robot R002 and starts after R001 completes T001 and R002 becomes idle. This scheduling rule takes into account the matching of the resource requirements of the task with the type of robot, the current state and position of the robot, the expected completion time of the task, the initial task execution order, and the collaborative efficiency between robots. The scheduling algorithm also generates estimated start and end times for each task, as well as alternative plans to deal with unexpected situations.

[0046] S50: According to the multi-task collaborative scheduling rule, control multiple target robots dynamically determined from all available robots in the smart factory to perform the multiple tasks.

[0047] In this step, based on the generated multi-task collaborative scheduling rules, the system starts to control the corresponding robot to perform the task. This process is dynamic, and the system will monitor the task progress and robot status in real time, and adjust the scheduling rules when necessary. The specific execution process is as follows: First, the system sends an instruction to R003 to execute T003, including task details, start time, and required tools (forklift) and other information. After receiving the instruction, R003 moves to the specified position and prepares to start the handling task. At the same time, the system notifies R001 to prepare for T001 and ensure that the required welding materials and tools are ready. After R003 completes T003, the system immediately instructs R001 to start T001. The system continuously monitors the status of R002. Once its current task is completed and becomes idle, and T001 has been completed, it immediately instructs R002 to start T002. During the entire execution process, if there is an unexpected situation (such as robot failure or task delay), the system will dynamically adjust the scheduling rules, reallocate tasks, or adjust the execution order. After each task is completed, the corresponding robot reports the completion status to the system and prepares to perform the next task or return to standby status. Through this dynamic collaborative scheduling method, the system can flexibly respond to various situations and ensure that multiple tasks can be completed efficiently and coordinated, while maximizing resource utilization and improving overall production efficiency.

[0048] like Figure 2As shown, the step S10 of receiving input data of multiple tasks to be performed by the smart factory specifically includes the following steps:

[0049] S11: Receive a data packet containing key information of multiple tasks from the production management system of the smart factory.

[0050] Specifically, in this step, the central control system of the smart factory establishes a secure data connection with the production management system and receives a data packet containing multiple task information through a predefined interface or API. This data packet is a structured file, such as JSON or XML format, containing detailed information of all tasks that the factory needs to perform. For example, a data packet contains the following structure: {"tasks":[{"taskID":"T001","type":"welding","description":"Weld component A to component B","dueTime":"2024-11-15T14:30:00","scale":"medium","urgency":"high"},{"taskID":"T002","type":"assembly","description":"Assemble parts C,D,and E after T001 is completed","dueTime":"2024-11-15T16:00:00","scale":"large","urgency":"medium"},...]}.

[0051] S12: Parse the data packet and extract key information of each task from the data packet, wherein the key information includes: task identification, task type, task description text, expected completion time of each task, task scale information, and task urgency.

[0052] Specifically, the system uses a parser to process the received data packets and extract the key information of each task. For the above JSON formatted data packets, the system uses the JSON parsing library to read the information of each task one by one. For example, for task T001, the system extracts the following information: task ID (T001), task type (welding), task description text (Weldcomponent A to component B), expected completion time (2024-11-15T14:30:00), task scale information (medium), and task urgency (high). The system repeats this process for each task in the data packet, storing the extracted information in memory or a temporary database for subsequent processing.

[0053] S13: directly extracting the task type and the expected completion time of each task from the key information.

[0054] Specifically, in this step, the system directly extracts the task type and expected completion time from the parsed key information without further processing. For example, for task T001, the system directly obtains the task type as welding and the expected completion time as "2024-11-15T14:30:00". This information will be stored in an easily accessible data structure, such as a hash table or a dedicated object, so that the subsequent scheduling algorithm can quickly access and use it.

[0055] S14: Based on the task description text, a natural language processing algorithm is used to determine the dependency relationship between tasks.

[0056] Specifically, the system uses natural language processing (NLP) algorithms to analyze task description texts and identify keywords and phrases that represent task dependencies. For example, for the description of task T002, "Assemble parts C, D, and E after T001 is completed", the NLP algorithm will recognize the phrase "after T001 is completed", thereby determining that T002 depends on the completion of T001. The system uses techniques such as named entity recognition and dependency syntax analysis to identify task IDs and words that represent timing or dependencies. In this way, the system can build a task dependency graph that clearly shows which tasks must be completed before or after other tasks.

[0057] Among them, the natural language processing algorithm includes one or more of the following: a sequence-to-sequence model based on deep learning, a graph neural network, a relationship extraction model based on an attention mechanism, and a conditional random field. When determining the dependencies between tasks, a variety of natural language processing algorithms can be used. The sequence-to-sequence model (Seq2Seq) based on deep learning takes the task description as the input sequence and the dependency as the output sequence, which is suitable for processing complex task descriptions and long-distance dependencies. The graph neural network (GNN) represents tasks as nodes in a graph and potential dependencies as edges, which can capture the complex relationships between tasks.

[0058] Relational network. The relation extraction model based on the attention mechanism identifies entities (task IDs) in the task description and the relationships between them, and can focus on the most relevant parts of the description to improve accuracy. The Conditional Random Field (CRF) is suitable for sequence labeling tasks, marking dependent keywords in the task description, taking into account the dependency between tags, and is suitable for structured prediction.

[0059] S15: Determine the task priority according to the task urgency and the task type.

[0060] Specifically, the system uses predefined rules or machine learning models to determine the priority of tasks based on task urgency and task type. For example, a simple rule can be set: tasks with high urgency get a basic priority of 8-10, tasks with medium urgency get a basic priority of 5-7, and tasks with low urgency get a basic priority of 1-4. Then, fine-tune according to the task type, for example, a welding task has a slightly higher priority than an assembly task. For task T001, given its high urgency and task type of welding, the system assigns it a priority of 9. This process assigns a specific numerical priority to each task for subsequent task sorting and scheduling.

[0061] In one embodiment, determining the task priority according to the task urgency and task type specifically includes: mapping the task urgency of each task into a preset numerical range to obtain a task urgency value for each task; assigning a corresponding weight coefficient to each different task type; for each task, multiplying the urgency value of the task by the weight coefficient of its corresponding task type to obtain a priority score for the task; and sorting multiple tasks according to the priority scores of each task, wherein a task with a higher score has a higher priority.

[0062] In this embodiment, the determination of task priority involves multiple steps, the first of which is to map the task urgency to a preset numerical range. For example, the urgency can be divided into four levels: very urgent, urgent, ordinary and not urgent, corresponding to the values ​​10, 7, 4 and 1 respectively. In this way, a task marked as urgent will be assigned an urgency value of 7. Next, the system assigns weight coefficients to different task types. This reflects the importance of different types of tasks in the overall production process. For example, a quality inspection task is assigned a weight of 1.5, while a routine maintenance task is assigned a weight of 0.8. This differentiated weight assignment ensures that critical tasks are given proper attention in the priority sorting. Then, the system multiplies the urgency value of each task by the weight coefficient of its corresponding task type to obtain the priority score of the task. For example, a quality inspection task (weight 1.5) with an urgency of urgent (value 7) will have a priority score of 7*1.5=10.5. And a routine maintenance task (weight 0.8) with an urgency of very urgent (value 10) will have a priority score of 10*0.8=8. This calculation method takes into account the urgency and importance of the task to obtain a more comprehensive priority assessment. Finally, the system sorts all tasks according to the calculated priority score, and the higher the score, the higher the priority. In the above example, although the maintenance task has a higher urgency, the quality inspection task is given a higher final priority due to its higher weight. This sorting method ensures that the most critical and urgent tasks are handled first, while also taking into account the relative importance of different types of tasks, improving overall production efficiency and quality control.

[0063] S16: Determine the resource requirement of each task according to the task scale information.

[0064] Specifically, the system uses a predefined mapping table or heuristic algorithm to determine the specific resource requirements for each task based on the task scale information. For example, for a welding task T001 with a medium task scale, the system will determine the following resource requirements: 1 welding robot, 1 set of welding tools, 30 minutes of estimated execution time, and 2 auxiliary workstations. For a large-scale assembly task T002, the system will determine the need for: 2 assembly robots, 1 set of general tools, 1 set of special assembly tools, 60 minutes of estimated execution time, and 3 auxiliary workstations.

[0065] like Figure 3 As shown, the step S20 of determining the initial task execution order of the multiple tasks according to the task types, the dependencies between the tasks and the task priorities specifically includes:

[0066] S21: Based on the dependencies between tasks, a directed acyclic graph is constructed, in which each node in the directed acyclic graph represents a task, and the edge represents the dependencies between tasks. Specifically, in this step, the system uses a graph data structure to represent the dependencies between tasks. Assume that there are the following tasks: T001 (welding task A), T002 (assembly task B, dependent on T001), T003 (welding task C), T004 (assembly task D, dependent on T003), T005 (intermediate quality inspection task E, dependent on T002 and T004), T006 (painting task F, dependent on T005), T007 (final quality inspection task G, dependent on T006), T008 (assembly task H). The system creates a directed graph with an edge from T001 to T002, an edge from T003 to T004, an edge from T002 and T004 to T005, an edge from T005 to T006, and an edge from T006 to T007. T008 has no dependencies and is an independent node. This graph structure ensures that the execution order of tasks complies with the dependencies, while also allowing tasks that have no dependencies on each other (such as T001 and T003, or T008) to be executed in parallel.

[0067] S22: Convert the task priority of each task into a numerical form to obtain the priority value of each task. Specifically, the system converts the task priority into a numerical range of 1-10, where 10 represents the highest priority and 1 represents the lowest priority. Assume that the priorities of the tasks are as follows: T001(8), T002(7), T003(8), T004(7), T005(6), T006(5), T007(1), T008(7). This numerical priority enables the system to more easily compare and process the importance of tasks.

[0068] S23: Topologically sort the directed acyclic graph to obtain a preliminary task sequence that satisfies the dependencies between tasks. Specifically, the system topologically sorts the constructed directed acyclic graph using the Kahn algorithm or depth-first search (DFS). Using the above example, the result of the topological sorting is: [T001, T003, T008, T002, T004, T005, T006, T007]. This preliminary sequence ensures that each task will not be executed until all its predecessor tasks are completed, thereby satisfying the constraints of the dependencies. As an independent task, T008 can be flexibly placed in a position that does not affect the dependencies of other tasks.

[0069] S24: Based on the preliminary task sequence, combined with the priority value and the task type of each task, the following processing is performed to obtain the processed task sequence: on the premise of maintaining the dependency relationship between tasks, the tasks with high priority values ​​are arranged at the front of the preliminary task sequence; the order of tasks with the same priority value is adjusted so that tasks of different task types appear alternately to achieve a balanced distribution of task types. Specifically, the system first considers the task priority, and advances the high priority task without violating the dependency relationship. Then, the system will evenly distribute tasks of different types. The processed task sequence becomes: [T001 (welding A), T003 (welding C), T002 (assembly B), T008 (assembly H), T004 (assembly D), T005 (intermediate quality inspection E), T006 (painting F), T007 (final quality inspection G)]. This adjustment not only takes into account the priority, but also makes different types of tasks appear alternately (welding and assembly tasks alternate, followed by quality inspection and painting) to optimize resource utilization.

[0070] S25: Use the processed task sequence as the initial task execution order of multiple tasks. Specifically, the system determines the processed task sequence [T001, T003, T002, T008, T004, T005, T006, T007] obtained in the previous step as the initial task execution order. This order will serve as the basis for subsequent task scheduling and resource allocation. This initial order comprehensively considers task dependencies, priorities, and type distribution. During the actual execution process, the system will dynamically adjust this order according to real-time conditions (such as resource availability, new high-priority tasks, etc.) to ensure the efficiency and flexibility of the production process.

[0071] like Figure 4 As shown, in step S30, the current resource status data of all available robots in the smart factory is obtained, specifically including:

[0072] S31: Send a robot list query request to the production management system of the smart factory to obtain a list of all available robots in the current smart factory. Specifically, the system sends an API request to the production management system (MES system) of the smart factory to obtain a list of currently available robots. For example, the system sends an HTTP GET request to "http: / / mes.factory.com / api / robots / available". After receiving this request, the production management system returns a JSON array containing all currently available robot identifiers, such as [RB001, RB002, RB003, RB004, RB005]. This list includes all robots that are not currently in a faulty state, are on standby, or can accept new tasks.

[0073] S32: Traverse the identification list of all available robots, and perform the following sub-steps for each available robot: send a resource status query request for the available robot to the production management system; receive and parse the current resource status data of the available robot returned by the production management system based on the resource status query request. Specifically, the system traverses the robot identification list obtained in the previous step and performs a separate status query for each robot. For example, for a robot identified as RB001, the system sends an HTTP GET request to "http: / / mes.factory.com / api / robots / RB001 / status". The production management system returns a JSON object containing the current status of the robot, which may include the following information: current power level (such as 85%), current position coordinates (such as {x:10.5, y:20.3, z:1.5}), currently executed task ID (such as "TASK123" or null if no task is executed), continuous working time (such as 7.5 hours), last maintenance time (such as "2024-11-2708:00:00"), etc. The system will parse this JSON object and extract the required status information.

[0074] In some embodiments, the current resource status data may include: the current processing capacity of the available robot, the tools currently equipped with the available robot, the raw material data available to the available robot, the battery level of the available robot, the physical workspace currently available to the available robot, the current operating status of the available robot, the current position coordinates of the available robot, and the status of each functional module of the available robot.

[0075] The current processing capacity of an available robot refers to the type and efficiency of tasks that the robot can perform in its current state, for example, a welding robot has the processing capacity to complete 20 welds per hour. The tools currently equipped by an available robot refer to the tools and accessories currently installed or immediately available for use by the robot, such as welding guns, fixtures, or painting equipment. The raw material data available to an available robot indicates the material information that the robot can currently use directly, including type, quantity, and location, such as 500 meters of welding wire remaining or 10 liters of paint reserve.

[0076] The battery level of an available robot refers to the percentage of battery power remaining in the robot, for example, 85%. The currently available physical workspace of an available robot describes the range of areas in which the robot can move and operate freely, expressed in coordinates or area descriptions, for example, the workspace is 10 meters long, 8 meters wide, and 3 meters high.

[0077] The current operating state of the available robot indicates whether the robot is working, standing by, charging or undergoing maintenance, for example, the state is executing task TASK123 or the state is standing by. The current position coordinates of the available robot accurately locate the position of the robot in the workspace, expressed in three-dimensional coordinates, for example, {x:10.5, y:20.3, z:1.5}.

[0078] The status of each functional module of the available robot describes in detail the working status of each component of the robot, including but not limited to the motion control system, sensor system, communication module, etc. For example, motion control system: normal; visual sensor: calibration required; communication module: signal strength 90%.

[0079] S33: After completing the resource status query of all available robots, the current resource status data of all available robots are integrated to form a data set containing the current resource status data of all available robots. Specifically, the system integrates the status data of each robot obtained in the previous steps into a unified data structure. This data structure can be a large JSON object, in which the identifier of each robot is used as a key, and the corresponding value is the detailed status data of the robot. For example:

[0080] {

[0081] "RB001":{

[0082] "batteryLevel":85,

[0083] "position":{"x":10.5,"y":20.3,"z":1.5},

[0084] "currentTask":"TASK123",

[0085] "continuousWorkHours":7.5,

[0086] "lastMaintenance":"2024-11-2708:00:00"

[0087] },

[0088] "RB002":{

[0089] "batteryLevel":92,

[0090] "position":{"x":15.2,"y":8.7,"z":0.5},

[0091] "currentTask":null,

[0092] "continuousWorkHours":2.3,

[0093] "lastMaintenance":"2024-11-2620:00:00"

[0094] },

[0095] / / ...data of other robots

[0096] }.

[0097] like Figure 5 As shown, in step S40, according to the resource requirements of each task, the expected completion time of each task, the initial task execution order and the current resource status data of all available robots, a scheduling algorithm is used to generate a multi-task collaborative scheduling rule, which specifically includes:

[0098] S41: Create a task resource requirement matrix and a robot resource status matrix, wherein the rows of the task resource requirement matrix represent the task identifier, the columns represent the resource type, and the matrix elements represent the demand for the resource type corresponding to each task identifier; the rows of the robot resource status matrix represent the robot identifier, the columns represent the resource type, and the matrix elements represent the available amount of the resource type corresponding to each robot identifier. Specifically, the system first creates two matrices to represent the task requirements and the robot resource status. For example, assuming there are 3 tasks (T1, T2, T3) and 4 resource types (welding capacity, assembly capacity, painting capacity, handling capacity), the task resource requirement matrix is ​​as follows: T1: [2, 0, 1, 1], T2: [0, 3, 0, 2], T3: [1, 1, 2, 0]. This means that T1 requires 2 units of welding capacity, 1 unit of painting capacity, and 1 unit of handling capacity. Similarly, suppose there are 4 robots (R1, R2, R3, R4), and the robot resource state matrix is ​​as follows: R1: [3, 0, 0, 2], R2: [0, 2, 3, 1], R3: [2, 2, 0, 2], R4: [1, 1, 2, 3]. This means that R1 has 3 units of welding capacity and 2 units of handling capacity.

[0099] S42: Determine the preliminary matching relationship between the multiple tasks and the multiple target robots according to the corresponding relationship between the task resource requirement matrix and the robot resource state matrix, and determine the preliminary execution start time and preliminary estimated execution duration of each task according to the initial task execution sequence, and convert the preliminary matching relationship, the preliminary execution start time of each task and the preliminary estimated execution duration of each task into a matrix.

[0100] The execution time is used as the preliminary task allocation result. Specifically, the system will make a preliminary match based on the task requirements and robot capabilities. For example, T1 mainly requires welding capabilities, so R1 will be matched first. The system obtains the following preliminary matches: T1->R1, T2->R2, T3->R4. Then, according to the initial task execution order (assuming it is T1, T2, T3), the system will assign a preliminary start time and estimated execution time to each task. For example: T1: start time = 0, estimated execution time = 2 hours; T2: start time = 2, estimated execution time = 3 hours; T3: start time = 5, estimated execution time = 2.5 hours.

[0101] S43: Based on the preliminary task allocation result, combined with the expected completion time of each task and the resource requirements of each task, the preliminary matching relationship, the initial task execution order and the preliminary execution start time of each task are adjusted by the optimization algorithm to determine the optimized matching relationship between multiple tasks and multiple target robots, the optimized task execution order of multiple tasks, the optimized execution start time of each task and the optimized estimated execution time of each task. Specifically, the system uses an optimization algorithm (such as a genetic algorithm or a simulated annealing algorithm) to improve the preliminary allocation results. Assuming that T2 has a more urgent expected completion time, the system will adjust the execution order to T2, T1, T3, and reallocate the robots to improve efficiency. The optimized results are as follows: T2: R2+R4, start time = 0, estimated execution time = 2 hours; T1: R1+R3, start time = 1, estimated execution time = 1.5 hours; T3: R4, start time = 2, estimated execution time = 2 hours. This optimization takes into account task urgency and improves efficiency through parallel processing (for example, T2 uses R2 and R4 simultaneously).

[0102] Specifically, the optimization algorithm includes at least one of the following: a hybrid particle swarm optimization algorithm, a genetic algorithm, an ant colony optimization algorithm, a simulated annealing algorithm, and a hybrid heuristic algorithm.

[0103] In one embodiment, when S43 adopts a hybrid particle swarm optimization algorithm (Hybrid Particle Swarm Optimization, HPSO), it can be decomposed into the following specific sub-steps:

[0104] S43.1: First, the problem is encoded to convert the task scheduling problem into a format suitable for HPSO processing. Each particle represents a complete scheduling solution, including the task-robot matching relationship, task execution order, and execution start time. For example, a particle can be represented as [T1, R2, 0.5, T2, R1, 0, T3, R3, 1.2], where each three elements represent the task, the assigned robot, and the execution start time.

[0105] S43.2: Then initialize the particle swarm and generate a set of initial particles based on the preliminary task allocation results, each particle represents a possible scheduling solution. Ensure that the initial particle swarm has sufficient diversity to cover different regions of the solution space.

[0106] S43.3: Then define the fitness function and design a function that comprehensively evaluates the quality of the scheduling plan, taking into account factors such as minimization of total completion time, maximization of resource utilization, satisfaction of the expected task completion time, rationality of task execution sequence, and balance of resource allocation.

[0107] S43.4: Then, a particle evaluation is performed, a fitness function is applied to each particle, its fitness value is calculated, and the global best particle (gBest) and the individual best position of each particle (pBest) are recorded.

[0108] S43.5: Update the velocity and position of each particle according to the HPSO formula:

[0109] V(t+1)=w*V(t)+c1*r1*(pBest-X(t))+c2*r2*(gBest-X(t)), X(t+1)=X(t)+V(t+1), where w is the inertia weight, c1 and c2 are acceleration constants,

[0110] Among them, V(t+1) represents the particle velocity vector at the next moment (t+1), V(t) represents the particle velocity vector at the current moment (t), X(t+1) represents the particle position vector at the next moment (t+1), and X(t) represents the particle position vector at the current moment (t); pBest represents the best position in the history of the particle, that is, the optimal solution found by the particle so far; gBest represents the global best position in the history of the entire particle swarm, that is, the optimal solution found by all particles so far; w is the inertia weight, which is used to control the degree to which the particle maintains the current motion trend. A larger w is conducive to global search, and a smaller w is conducive to local search. c1 is a cognitive factor, which indicates the tendency of particles to move toward the individual optimal position; c2 is a social learning factor, which indicates the tendency of particles to move toward the group optimal position. c1 and c2 are set to constants; r1 and r2 are random numbers between [0,1], which are used to increase the randomness of the search; (pBest-X(t)) represents the vector from the current position of the particle to its individual historical optimal position; (gBest-X(t)) represents the vector from the current position of the particle to the global optimal position.

[0111] S43.6: Perform local search around the updated particle position, using simulated annealing or a small-scale genetic algorithm as a strategy to improve the convergence speed of the algorithm and the quality of the solution.

[0112] S43.7: Boundary processing and constraint satisfaction are then performed to check whether the updated particles meet the problem constraints, such as resource capacity constraints, task priorities, etc., and particles that violate the constraints are corrected or punished.

[0113] S43.8: Re-evaluate the particle swarm and update the global best particle (gBest) and the individual best position of each particle (pBest).

[0114] S43.9: Dynamically adjust algorithm parameters, such as inertia weight w and acceleration constants c1 and c2, according to the iteration process to balance global exploration and local development capabilities.

[0115] S43.10: Check whether the termination condition is met, such as reaching the maximum number of iterations, finding a satisfactory solution, or no significant improvement after multiple consecutive iterations. If not, return to step S43.5 to continue iterating.

[0116] S43.11: When the termination condition is met, the global optimal particle (gBest) is decoded into the actual scheduling plan to obtain the optimized task-robot matching relationship, task execution order, execution start time of each task and estimated execution time.

[0117] S43.12: Perform final verification on the decoded scheduling plan to ensure that all constraints are met, and make minor adjustments if necessary to adapt to the special requirements of the actual production environment.

[0118] S43.13: Finally, the final optimized scheduling plan is output, including the optimized matching relationship between multiple tasks and multiple target robots, the optimized task execution order of multiple tasks, the optimized execution start time of each task and the optimized estimated execution time of each task.

[0119] S44: Based on the resource requirements of each task and the current resource status data of all available robots, a hybrid meta-heuristic algorithm is used to obtain a resource allocation plan for multiple tasks and multiple target robots. Specifically, the system will plan the specific resources used for each task in detail. For example, for the optimized T1 task: R1 provides 2 units of welding capacity and 1 unit of handling capacity; R3 provides 1 unit of painting capacity. The system will ensure that the resource requirements of each task are met while avoiding resource conflicts.

[0120] In one embodiment, step S44 may include the following sub-steps:

[0121] S44.1: Calculate the initial resource allocation based on the task-robot matching relationship. In the initialization phase, the system first reads the task-robot matching relationship output by S43.13, denoted as M = {(ti, rj)}, where ti represents the task and rj represents the matching robot. At the same time, obtain the resource demand vector Di = (Di,1, Di,2, ..., Di,k) of each task ti and the current available resource state vector Cj = (Cj,1, Cj,2, ..., Cj,k) of each robot rj. In the resource allocation calculation phase, for each matching pair (ti, rj), perform the following steps: initialize the resource allocation vector Xij = (0,0, ...,0) of task ti on robot rj, and then allocate resource type k item by item. If Di,k ≤ Cj,k, the allocation amount is Xij,k = Di,k; otherwise, Xij,k = Cj,k, and the resource is marked as insufficient. After each allocation, the available resource state of robot rj is updated to Cj = Cj-Xij. In order to ensure the dynamic adaptation of constraints, the system checks the resource capacity constraints and task requirements in real time during the allocation process, and dynamically adjusts the constraints based on task priority and resource status, such as lowering the allocation limit for robots with limited resources or giving priority to key tasks. In the case of insufficient resources or constraint violations, the system will give priority to finding compatible robots with sufficient resources for resource reallocation, adjust task priorities when necessary, or try to split and execute tasks in parallel, and finally generate the initial resource allocation matrix X0, and output task resource allocation details and related reports.

[0122] S44.2: Construct a comprehensive multi-objective optimization function. The form of the multi-objective optimization function is F(X) = w1f1(X) + w2f2(X) + w3f3(X) + w4f4(X), where f1(X) is the total completion time target, f2(X) is the resource utilization target, f3(X) is the load balancing target, and f4(X) is the time optimization target. Weights w1 to w4 reflect the priority of each target. To improve the optimization effect, the system dynamically adjusts the weights according to the task requirements of the current stage. For example, when the task time pressure is high, the weight of f1 is increased, when resources are tight, the weight of f2 is increased, and when the load difference is large, the weight of f3 is increased, thereby achieving flexible switching of optimization targets and reasonable allocation of resources.

[0123] S44.3: Define and dynamically adjust the problem constraints. Constraints include three categories: resource capacity constraints, that is, the total amount of resources allocated to all robots cannot exceed their available resources; task requirement satisfaction constraints, that is, the total amount of resources allocated to each task must meet its requirements; task dependency constraints, that is, if task ti depends on task tj, then the completion time of ti must be later than the completion time of tj. During the resource allocation and task execution process, the system dynamically updates these constraints in real time, and dynamically adjusts the resource capacity constraints or task priorities according to the robot load and resource status to ensure the flexibility of allocation and system stability.

[0124] S44.4: Global optimization based on hybrid metaheuristic algorithm. Global optimization takes the initial resource allocation plan X0 as the starting point and uses a hybrid metaheuristic algorithm for iterative optimization. First, a population P containing N feasible solutions is generated, including the initial solution X0. The population is then subjected to genetic algorithm operations, including the selection of excellent individuals, single-point crossover, and random gene mutation. To avoid generating infeasible solutions, constraint correction operations are performed on the individuals after mutation to ensure that they meet task requirements and resource constraints. For the current optimal solution Xbest, a local search method is further used for optimization to improve the fitness of the solution by adjusting the task allocation. When the maximum number of iterations is reached or there is no better solution for multiple consecutive generations, the algorithm stops and outputs the global optimal solution.

[0125] Specifically, the specific working process of step S44.4 includes the following: First, the initial population P is created in the initialization phase, which contains N feasible solutions, of which the initial solution X0 generated in S44.1 is directly added to the population, and the remaining N-1 solutions are obtained by random generation or slight modification of X0. In the genetic algorithm operation, the selection operation is first performed, using the roulette selection method, and the fitness is calculated according to the multi-objective function F(X) defined in S44.2, and the selection probability is allocated according to the fitness. The higher the fitness, the greater the probability of the individual being selected. Then the crossover operation is performed, using the single-point crossover method, randomly selecting two parent individuals with a certain probability (for example, 0.8) for crossover, randomly selecting a crossover point, and exchanging the gene sequences of the two parents after the point. Then the mutation operation is performed, setting the mutation probability (for example, 0.1), and mutating each gene with a certain probability. The mutation may include changing the robot assigned to the task or adjusting the resource allocation amount. To ensure the feasibility of the solution, the constraints of the mutated individuals are corrected to check whether they meet the task requirements and resource constraints. If the constraints are violated, the resource allocation or task order is readjusted. A new population P' is generated through selection, crossover, mutation and constraint correction. Subsequently, the fitness of each solution in the new population is evaluated, and the current optimal solution Xbest is optimized through local search. A task is randomly selected and tried to be assigned to different robots. If a better solution is found, Xbest is updated. In the population update phase, populations P and P' are merged, and the N individuals with the highest fitness are selected to form the next generation population P, and the above steps are repeated. The termination condition is that the maximum number of iterations (for example, 1000 times) is reached or the optimal solution has not been improved for multiple consecutive generations (for example, 50 generations). Finally, the global optimal solution X* is output after the algorithm ends. This method combines the global search capability of genetic algorithms with the fine optimization capability of local searches, effectively improving the quality of solutions and optimization efficiency.

[0126] S44.5: Execute dynamic adjustment mechanism. The dynamic adjustment mechanism is used to respond to changes in real-time resource status or task progress. Trigger conditions include the deviation between the robot resource status and the expected value exceeding a threshold, the deviation between the actual task execution time and the expected duration exceeding a threshold, or the addition of a high-priority task. When the trigger condition is met, the adjustment function adjust(X,R,T) is called to perform local re-optimization of the affected tasks, giving priority to adjusting the allocation of new or changed tasks, and optimizing the resource utilization of the affected parts without destroying the allocation of other tasks. The system records the history of each adjustment to avoid repeated optimization of the same task or robot and ensure stable operation of the system. Among them, X represents the current resource allocation plan, which is a multidimensional matrix that describes the allocation relationship between tasks, robots and resource types, R represents the current resource status set of all robots, and T represents the relevant information of the task. T includes relevant information of all tasks, such as the resource requirements of the task, task priority, dependencies, current completion progress, etc.

[0127] S44.6: Output the optimal resource allocation plan and visualization results. The system generates the optimal resource allocation plan X* and outputs information including the resource utilization rate of each robot, the task allocation Gantt chart, and the time allocation plan for each type of resource (start time, duration, release time). At the same time, it provides dynamic adjustment strategy records, including trigger conditions, adjustment records, and local optimization plans, providing comprehensive support for the visualization and dynamic analysis of system operation.

[0128] S45: Integrate the optimized matching relationship between the multiple tasks and the multiple target robots, the optimized execution start time of each task and the optimized estimated execution time of each task, the optimized task execution order of the multiple tasks, and the resource allocation scheme for the multiple tasks and the multiple target robots to form a multi-task collaborative scheduling rule. Specifically, the system integrates all the optimized information into a complete scheduling rule. This rule is expressed in JSON format and contains detailed information of each task, such as task ID, assigned robot, start time, duration, and specific resource allocation. For example, the expression for the T2 task is: {"taskId":"T2","robots":["R2","R4"],"startTime":"2024-11-28T09:00:00","duration":"PT2H","resources":{"R2":["2 units of assembly capacity","1 unit of handling capacity"],"R4":["1 unit of assembly capacity","1 unit of handling capacity"]}}. This comprehensive scheduling rule provides detailed execution guidelines for the production management system of the smart factory, ensuring that multiple tasks can be completed efficiently and in a coordinated manner.

[0129] In some embodiments, the resource types include: the processing power of the robot, the tools required for the task, the raw materials consumed during the task execution, the power energy required for the robot to operate, and the physical workspace required for the robot to perform the task. In practical applications, these resource types can be defined more specifically. For example, the processing power of the robot can include welding capacity (unit: welding points / hour), assembly capacity (unit: parts / hour), painting capacity (unit: square meters / hour), etc. The tools required for the task may include welding guns, painting equipment, screwdrivers, etc. Raw materials may include sheet metal, paint, screws, etc. Power energy is measured in kilowatt-hours (kWh). Physical workspace can be measured in square meters or cubic meters. This detailed definition of resource types enables the system to allocate and schedule resources more accurately. In addition, the processing power of the robot can also include environmental adaptability, collaboration capabilities, etc.

[0130] In some embodiments, the resource allocation scheme includes: the resource type and resource quantity required for each task; the time dimension allocation plan of each type of resource during the task execution process, which includes the start time of resource use, the duration of resource use and the release time of the resource; and the dynamic adjustment operation when the resource state of the target robot does not match the resource requirements of its corresponding task. Specifically, the resource allocation scheme can be expressed as a detailed schedule. For example, for a welding task T1, its resource allocation scheme can be as follows:

[0131] Resource type and quantity, such as welding capacity 2 units (welding points / hour), welding gun 1, metal sheet 5 square meters, electricity 3kWh, work space 4 square meters. Time dimension allocation plan, such as welding capacity (use start time 09:00, duration 2 hours, release time 11:00), welding gun (use start time 09:00, duration 2 hours, release time 11:00), metal sheet (use start time 09:00, duration 2 hours, release time 11:00, consumable resource), electricity (use start time 08:55 preheating time, duration 2 hours 5 minutes, release time 11:00), work space (use start time 08:50 preparation time, duration 2 hours 15 minutes, release time 11:05 cleaning time). Dynamic adjustment operations, for example, if the welding capacity of the target robot R1 is found to drop suddenly during execution (for example, from 2 units to 1 unit), the system can take the following dynamic adjustment operations: extend the task execution time (extend the execution time of T1 from 2 hours to 3 hours), allocate additional resources (call another robot R2 with welding capabilities to assist in completing the task), and reorder (postpone the execution of T1 and perform other tasks that do not require high welding capabilities first). This detailed resource allocation plan ensures the accuracy and flexibility of task execution, and also provides a response strategy for problems that may arise.

[0132] like Figure 6 As shown, the step S50, according to the multi-task collaborative scheduling rule, controls the multiple target robots dynamically determined from all available robots in the smart factory to perform the multiple tasks, specifically including:

[0133] S51: According to the optimized matching relationship in the multi-task collaborative scheduling rule, select the target robot that matches each task from all available robots. For example, suppose there are three tasks T1 (welding), T2 (assembly) and T3 (painting), and five available robots R1 (welding expertise), R2 (multi-function), R3 (assembly expertise), R4 (painting expertise) and R5 (multi-function). According to the optimized matching relationship, the system selects R1 to execute T1, R3 to execute T2, and R4 to execute T3. This selection is based on the expertise and current load conditions of each robot to ensure that the task can be completed efficiently.

[0134] S52: Activate the target robot of each task in turn according to the optimized task execution order in the multi-task collaborative scheduling rule. Continuing the above example, assume that the optimized execution order is T2, T1, T3. The system will first activate R3 (execute T2), then R1 (execute T1), and finally R4 (execute T3). This order can be determined based on the urgency of the task, resource availability, or optimization of the overall production process.

[0135] S53: Send corresponding task instructions to each activated target robot, the task instructions include the task type, the optimized execution start time of the task, and the optimized estimated execution time. For example, the system sends the following instructions to R3: Task type: assembly; Execution start time: 09:00; Estimated execution time: 1 hour and 30 minutes. These detailed instructions ensure that each robot knows exactly what it needs to do, when to start, and when it is expected to be completed.

[0136] S54: According to the resource allocation scheme in the multi-task collaborative scheduling rule, allocate the resources required for the corresponding tasks to be performed to each activated target robot. Taking R3 performing T2 (assembly task) as an example, the system will allocate the following resources: one set of assembly tool kits, 100 electronic components, 1000 screws, 5 square meters of workbench area, and 2kWh of power supply. This precise resource allocation ensures that each robot has all the resources it needs when performing tasks, thereby improving efficiency and reducing interruptions.

[0137] S55: After each task is completed, the resource status of the target robot that performs the task is updated, and the target robot is re-included in the identification list of all available robots. For example, when R3 completes T2, the system updates the status of R3: adds 1 hour and 30 minutes to its working time, updates its power consumption, marks the assembly tool set as available, and re-adds R3 to the list of available robots. This ensures that resource usage is accurately tracked and enables robots to be assigned to new tasks immediately after completing a task.

[0138] S56: Repeat the above steps from activating the target robot for each task in turn to re-introducing the target robot into the queue of available robots until all tasks have been completed. In the above example, the system will continue to activate R1 to execute T1, and then R4 to execute T3. After each task is completed, the corresponding robot will be updated and re-introduced into the available queue. This cycle will continue until all planned tasks have been completed. This approach ensures the continuity and efficiency of the entire production process, while also maintaining the flexibility of the system to cope with possible changes or unexpected situations.

[0139] In a further embodiment, after allocating the resources required for the corresponding task to be performed to each activated target robot according to the resource allocation scheme in the multi-task collaborative scheduling rule in step S54, the following steps are also included:

[0140] Monitor the current resource status data of each target robot in real time. For example, the system checks the resource status of each robot performing a task every 30 seconds. This includes monitoring the robot's power, tool status, raw material inventory, processing capacity, and available workspace. For robot R1 that is performing a welding task, the system monitors its welding capacity (whether it maintains the expected 2 welds / hour), welding gun status (whether it is working properly), the remaining amount of metal sheet, power consumption, and workspace occupancy.

[0141] When it is monitored that the current resource status data of any target robot does not match the resource requirements of its corresponding task, the dynamic adjustment operation is performed, and the dynamic adjustment operation includes at least one of the following:

[0142] Reassign tasks to other target robots whose current resource status data matches the resource requirements of the task. For example, if it is found that the welding capacity of robot R1 suddenly drops to 1 weld point / hour and cannot meet the needs of the current task, the system will reallocate the remaining welding work to another robot R2 with sufficient welding capacity. This adjustment ensures that the task can be completed on time while also avoiding overall production delays caused by the performance degradation of a single robot.

[0143] Postponing the task until the original target robot has enough resources. For example, if robot R3 finds that there is insufficient inventory of electronic components when performing an assembly task, the system will postpone the task until the next batch of electronic components arrives. During this period, R3 can be assigned to perform other tasks that do not require these specific electronic components, thereby improving overall production efficiency.

[0144] The task is split into multiple subtasks and assigned to multiple target robots with complementary resources to complete collaboratively. For example, if a large painting task is originally assigned to robot R4, but R4's painting capacity is not enough to complete the entire task within the specified time, the system will split the task into two subtasks, one for R4 and the other for robot R5 with remaining painting capacity. This collaborative approach ensures that the task is completed on time while making full use of the capabilities of multiple robots.

[0145] Resources that will be released within the expected waiting time are reserved for the target robot, and are immediately assigned to the target robot after the reserved resources are released. For example, if robot R1 needs to use a specific high-precision welding tool to complete a task, but the tool is currently being used by robot R2, the system will reserve the tool for R1. Once R2 completes its task and releases the tool, the system will immediately assign the tool to R1 so that it can start executing the task. This reservation mechanism can minimize resource idle time and improve overall resource utilization.

[0146] By implementing these dynamic adjustment operations, smart factories can respond more flexibly to various resource status changes and unexpected situations, ensuring the continuity and efficiency of the production process. This real-time monitoring and dynamic adjustment method makes the entire production system more adaptable and robust, and can maintain efficient operation in a complex and changing production environment.

[0147] In a further embodiment, determining the resource requirement of each task according to the task scale information in step S16 specifically includes:

[0148] Model the mapping between task size information and resource requirements. This step involves creating a mathematical model that describes the relationship between task size and required resources. For example, for a welding task, the model can consider size information such as weld length (in meters), weld type (spot weld, continuous weld, etc.), material type (steel, aluminum, etc.) and map this information to required resources such as welding time, power consumption, wire usage, etc. This model can take the form of multiple regression analysis, decision trees, or neural networks, depending on the complexity of the task and the nature of the available data.

[0149] The mapping relationship model is trained using a machine learning algorithm. After the model framework is established, the next step is to train the model using historical data. For example, detailed records of all welding tasks in the past six months can be collected, including task size information and actual resource consumption data. Using this data, machine learning algorithms such as gradient descent, random forests, or deep learning can be applied to optimize model parameters. The training process involves dividing the data set into a training set and a validation set, using techniques such as cross-validation to ensure the generalization ability of the model. For example, for the welding task, the model learns that for each meter of steel spot welding, it takes an average of 2 minutes, 0.5 kilowatt-hours of electricity, and 10 grams of welding wire.

[0150] The task scale information is input into the trained mapping relationship model to predict the resource type and quantity required for the task. Once the model is trained and verified, it can be used to predict the resource requirements of new tasks. For example, if there is a new welding task that requires spot welding on a 5-meter-long steel plate, the system will input this scale information (5-meter length, steel, spot welding type) into the trained model. The model output is predicted as follows: the time required is about 10 minutes, the power demand is 2.5 kWh, and the welding wire consumption is 50 grams. These prediction results will be used in the subsequent resource allocation and task scheduling process. In this way, the system can more accurately estimate the resource requirements of each task, thereby optimizing the overall resource allocation and task execution efficiency.

[0151] This machine learning-based approach allows the system to continuously improve its predictive capabilities over time. As more tasks are performed and more data is collected, the model can be retrained periodically to adapt to changes in the production environment and improve predictive accuracy.

[0152] In a further embodiment, after the step of controlling multiple target robots dynamically determined from all available robots in the smart factory to perform the multiple tasks according to the multi-task collaborative scheduling rule, the following steps may also be included:

[0153] Monitor the task execution process in real time to detect whether any abnormalities occur. Abnormalities may include but are not limited to data anomalies, task execution failures, robot failures, etc. This monitoring process can be achieved through real-time data collection, sensor feedback, robot status reports, etc., to ensure that various abnormalities in the execution process can be discovered and responded to in a timely manner.

[0154] When an abnormal situation is detected, exception handling operations are performed. In the case of incomplete task input data or robot status data, the system will re-acquire complete data from the relevant data source, or use preset default values ​​to fill in missing data, or make reasonable estimates based on historical data. In the case of task execution failure, the system will obtain the cause of failure through task execution logs or robot feedback information, and take corresponding measures based on the cause of failure, such as re-planning resource allocation, adjusting the order of task execution, reallocating tasks to more suitable robots, or suspending tasks and notifying manual intervention when environmental factors interfere. In the case of robot failure, the system will try to reallocate tasks to other available robots with corresponding capabilities. If there is no available robot, the relevant tasks will be suspended and marked as pending, and the robot maintenance process will be triggered at the same time.

[0155] Update the multi-task collaborative scheduling rules based on the exception handling results. The update is based on the new task status after the exception handling, the latest resource status of the available robots, the actual execution time of completed and unfinished tasks, changes in task priorities, and analysis results of resource utilization and task completion efficiency. By comprehensively considering these factors, the system can dynamically adjust the scheduling strategy to adapt to the new situation after the exception handling, ensuring the efficiency and reliability of the overall task execution.

[0156] Continue to execute the updated multi-task collaborative scheduling rules until all tasks are completed or the preset termination conditions are reached. In this process, the system will continue to apply the updated scheduling rules, dynamically allocate and adjust tasks to adapt to the ever-changing production environment and needs. At the same time, the system will record abnormal situations, processing processes and results. These data will be used for subsequent analysis and system optimization to continuously improve the operating efficiency and abnormal response capabilities of smart factories.

[0157] Embodiment 2

[0158] like Figure 7 As shown, an embodiment of the present invention provides a multi-task collaborative robot scheduling system for a smart factory, which includes:

[0159] A receiving module, used to receive input data of multiple tasks to be executed by the smart factory, wherein the input data of each task includes: task type, dependencies between tasks, task priority, resource requirements of each task and expected completion time of each task;

[0160] An order determination module, used to determine an initial task execution order of a plurality of tasks according to the task types, dependencies between the tasks and the task priorities;

[0161] The status acquisition module is used to obtain the current resource status data of all available robots in the smart factory;

[0162] A scheduling rule generation module, used to generate multi-task collaborative scheduling rules using a scheduling algorithm according to the resource requirements of each task, the expected completion time of each task, the initial task execution order, and the current resource status data of all available robots;

[0163] The execution control module is used to control multiple target robots dynamically determined from all available robots in the smart factory to execute the multiple tasks according to the multi-task collaborative scheduling rules.

[0164] In some embodiments, the receiving module specifically includes:

[0165] A data receiving unit, used to receive a data packet containing key information of multiple tasks from a production management system of a smart factory;

[0166] A data parsing unit, used to parse the data packet and extract key information of each task from the data packet, wherein the key information includes: task identification, task type, task description text, expected completion time of each task, task scale information, and task urgency;

[0167] An information extraction unit, used to directly extract the task type and the expected completion time of each task from the key information;

[0168] A dependency determination unit, configured to determine the dependency between tasks using a natural language processing algorithm based on the task description text;

[0169] A priority determination unit, used to determine the task priority according to the task urgency and the task type;

[0170] The resource requirement determination unit is used to determine the resource requirement of each task according to the task scale information.

[0171] In some embodiments, the sequence determination module specifically includes:

[0172] A graph construction unit, used to construct a directed acyclic graph based on the dependency relationship between tasks, wherein each node in the directed acyclic graph represents a task, and the edge represents the dependency relationship between tasks;

[0173] A priority conversion unit, used for converting the task priority of each task into a numerical form to obtain a priority value of each task;

[0174] A topological sorting unit, used for topologically sorting the directed acyclic graph to obtain a preliminary task sequence that satisfies the dependency relationship between tasks;

[0175] A sequence processing unit is used to perform the following processing based on the preliminary task sequence and in combination with the priority value and the task type of each task to obtain a processed task sequence: arranging tasks with high priority values ​​at the front of the preliminary task sequence while maintaining the dependency relationship between tasks; adjusting the order of tasks with the same priority value so that tasks of different task types appear alternately to achieve a balanced distribution of task types;

[0176] The sequence determination unit is used to use the processed task sequence as the initial task execution sequence of multiple tasks.

[0177] In some embodiments, the status acquisition module specifically includes:

[0178] A list query unit is used to send a robot list query request to the production management system of the smart factory to obtain a list of identifiers of all available robots in the current smart factory;

[0179] A status query unit is used to traverse the identification list of all available robots and perform the following sub-steps for each available robot: send a resource status query request for the available robot to the production management system; receive and parse the current resource status data of the available robot returned by the production management system based on the resource status query request;

[0180] The data integration unit is used to integrate the current resource status data of all available robots after completing the resource status query of all available robots to form a data set containing the current resource status data of all available robots.

[0181] In some embodiments, the scheduling rule generation module specifically includes:

[0182] A matrix creation unit, used to create a task resource demand matrix and a robot resource status matrix, wherein the rows of the task resource demand matrix represent task identifiers, the columns represent resource types, and the matrix elements represent the demand for the resource type corresponding to each task identifier; the rows of the robot resource status matrix represent robot identifiers, the columns represent resource types, and the matrix elements represent the available amount of the resource type corresponding to each robot identifier;

[0183] A preliminary matching unit, used to determine a preliminary matching relationship between a plurality of tasks and a plurality of target robots according to a corresponding relationship between the task resource requirement matrix and the robot resource state matrix, and to determine a preliminary execution start time and a preliminary estimated execution duration of each task according to the initial task execution sequence, and to use the preliminary matching relationship, the preliminary execution start time of each task and the preliminary estimated execution duration of each task as a preliminary task allocation result;

[0184] An optimization and adjustment unit, configured to adjust the preliminary matching relationship, the initial task execution sequence and the preliminary execution start time of each task through an optimization algorithm based on the preliminary task allocation result, in combination with the expected completion time of each task and the resource requirement of each task, to determine the optimized matching relationship between the multiple tasks and the multiple target robots, the optimized task execution sequence of the multiple tasks, the optimized execution start time of each task and the optimized expected execution time of each task;

[0185] A resource allocation unit, configured to obtain a resource allocation scheme for multiple tasks and multiple target robots by using a hybrid meta-heuristic algorithm based on the resource requirements of each task and the current resource status data of all available robots;

[0186] A rule integration unit is used to integrate the optimized matching relationship between the multiple tasks and the multiple target robots, the optimized execution start time of each task and the optimized estimated execution time of each task, the optimized task execution order of the multiple tasks, and the resource allocation plan for the multiple tasks and the multiple target robots to form a multi-task collaborative scheduling rule.

[0187] In some embodiments, the resource types include: the processing power of the robot, the tools required for the task, the raw materials consumed during the execution of the task, the electrical energy required for the operation of the robot, and the physical workspace required for the robot to perform the task;

[0188] The resource allocation scheme includes: the resource type and quantity required for each task; the time dimension allocation plan of each type of resource during the task execution process, which includes the start time of resource use, the duration of resource use and the release time of the resource; and dynamic adjustment operations when the resource status of the target robot does not match the resource requirements of its corresponding task.

[0189] In some embodiments, the execution control module specifically includes:

[0190] A robot selection unit, configured to select a target robot matching each task from all available robots according to the optimized matching relationship in the multi-task collaborative scheduling rule;

[0191] A robot activation unit, used to activate the target robot of each task in turn according to the optimized task execution order in the multi-task collaborative scheduling rule;

[0192] An instruction sending unit, used to send a corresponding task instruction to each activated target robot, wherein the task instruction includes a task type, an optimized execution start time of the task, and an optimized estimated execution time;

[0193] A resource allocation unit, configured to allocate resources required for the corresponding task to be performed to each activated target robot according to the resource allocation scheme in the multi-task collaborative scheduling rule;

[0194] A status updating unit, used to update the resource status of the target robot corresponding to the task after each task is completed, and to add the target robot back into the identification list of all available robots;

[0195] The loop control unit is used to repeatedly execute the above steps from sequentially activating the target robot of each task to re-introducing the target robot into the queue of available robots until all tasks have been completed.

[0196] In some embodiments, the execution control module further includes:

[0197] A monitoring unit, used to monitor the current resource status data of each target robot in real time;

[0198] The dynamic adjustment unit is used to perform the dynamic adjustment operation when it is monitored that the current resource state data of any target robot does not match the resource requirements of its corresponding task, and the dynamic adjustment operation includes at least one of the following:

[0199] Reassign the task to other target robots whose current resource state data matches the resource requirements of the task;

[0200] Postpone the task until the original target robot obtains sufficient resources;

[0201] Split the task into multiple subtasks and assign them to multiple target robots with complementary resources to complete them collaboratively;

[0202] Resources that will be released within the expected waiting time are reserved for the target robot, and the reserved resources are immediately allocated to the target robot after being released.

[0203] In some embodiments, the resource requirement determination unit specifically includes:

[0204] A model building subunit is used to build a mapping relationship model between task scale information and resource requirements;

[0205] A model training subunit, used for training the mapping relationship model using a machine learning algorithm;

[0206] The resource prediction subunit is used to input the task scale information into the trained mapping relationship model to predict the resource type and resource quantity required for the task.

[0207] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0208] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, any one of the above methods is implemented.

[0209] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. Of course, there are other ways of readable storage media, such as quantum memory, graphene memory, etc.

[0210] The present invention also provides an electronic device. The electronic device of an embodiment of the present invention comprises: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method provided by the present invention.

[0211] Reference below Figure 8, which shows a schematic diagram of the structure of a computer system 800 of an electronic device suitable for implementing an embodiment of the present invention. Figure 8 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0212] like Figure 8 As shown, the computer system 800 includes a central processing unit (CPU) 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage part 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the computer system 800 are also stored. The CPU 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0213] The following components are connected to the I / O interface 805: an input section 806 including a keyboard, a mouse, etc.; an output section 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, a modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the I / O interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed so that a computer program read therefrom is installed into the storage section 808 as needed.

[0214] The embodiment of the present invention also includes a computer program product, which includes a computer program carried on a computer readable medium, and the computer program contains program code for executing the method shown in the main step diagram. In the above embodiment, the computer program can be downloaded and installed from the network through the communication part 809, and / or installed from the removable medium 811. When the computer program is executed by the central processing unit 801, the above functions defined in the system of the present invention are executed.

[0215] It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-task collaborative robot scheduling method for smart factories, characterized in that: The following steps are involved: Receiving input data of a plurality of tasks to be performed by the smart factory, wherein the input data of each task includes: task type, dependencies between tasks, task priority, resource requirements of each task, and expected completion time of each task; Determining an initial task execution order of the plurality of tasks according to the task types, dependencies between the tasks and the task priorities; Get the current resource status data of all available robots in the smart factory; Generate a multi-task collaborative scheduling rule using a scheduling algorithm according to the resource requirements of each task, the expected completion time of each task, the initial task execution order, and the current resource status data of all available robots; According to the multi-task collaborative scheduling rule, multiple target robots dynamically determined from all available robots in the smart factory are controlled to perform the multiple tasks.

2. The method according to claim 1, characterized in that: The receiving of input data of multiple tasks to be performed by the smart factory specifically includes: Receive data packets containing critical information for multiple tasks from the production management system of the smart factory; Parsing the data packet, extracting key information of each task from the data packet, the key information including: task identification, task type, task description text, expected completion time of each task, task scale information, and task urgency; Directly extracting the task type and the expected completion time of each task from the key information; Based on the task description text, a natural language processing algorithm is used to determine the dependency relationship between tasks; Determine the task priority according to the task urgency and the task type; The resource requirement of each task is determined according to the task scale information.

3. The method according to claim 1, characterized in that: The determining of the initial task execution order of the plurality of tasks according to the task types, the dependencies between the tasks and the task priorities specifically includes: Based on the dependency relationship between the tasks, a directed acyclic graph is constructed, where each node in the directed acyclic graph represents a task, and the edges represent the dependency relationship between the tasks; Convert the task priority of each task into a numerical form to obtain the priority value of each task; Topologically sorting the directed acyclic graph to obtain a preliminary task sequence that satisfies the dependency relationship between tasks; Based on the preliminary task sequence, combined with the priority value and the task type of each task, the following processing is performed to obtain a processed task sequence: on the premise of maintaining the dependency relationship between tasks, tasks with high priority values ​​are arranged at the front of the preliminary task sequence; the order of tasks with the same priority value is adjusted so that tasks of different task types appear alternately to achieve a balanced distribution of task types; The processed task sequence is used as the initial task execution order of multiple tasks.

4. The method according to claim 1, characterized in that The acquisition of current resource status data of all available robots in the smart factory specifically includes: Send a robot list query request to the production management system of the smart factory to obtain the identification list of all available robots in the current smart factory; Traversing the identification list of all available robots, and executing the following sub-steps for each available robot: sending a resource status query request for the available robot to the production management system; receiving and parsing the current resource status data of the available robot returned by the production management system based on the resource status query request; After completing the resource status query of all available robots, the acquired current resource status data of all available robots are integrated to form a data set containing the current resource status data of all available robots.

5. The method according to claim 1, characterized in that The multi-task collaborative scheduling rules are generated by using a scheduling algorithm according to the resource requirements of each task, the expected completion time of each task, the initial task execution order, and the current resource status data of all available robots, specifically including: Creating a task resource demand matrix and a robot resource status matrix, wherein the rows of the task resource demand matrix represent task identifiers, the columns represent resource types, and the matrix elements represent the demand for the resource type corresponding to each task identifier; the rows of the robot resource status matrix represent robot identifiers, the columns represent resource types, and the matrix elements represent the available quantity of the resource type corresponding to each robot identifier; Determine a preliminary matching relationship between a plurality of tasks and a plurality of target robots according to the corresponding relationship between the task resource requirement matrix and the robot resource state matrix, and determine a preliminary execution start time and a preliminary estimated execution duration of each task according to the initial task execution sequence, and use the preliminary matching relationship, the preliminary execution start time of each task and the preliminary estimated execution duration of each task as a preliminary task allocation result; Based on the preliminary task allocation result, combined with the expected completion time of each task and the resource requirements of each task, the preliminary matching relationship, the initial task execution order and the preliminary execution start time of each task are adjusted through an optimization algorithm to determine the optimized matching relationship between multiple tasks and multiple target robots, the optimized task execution order of multiple tasks, the optimized execution start time of each task and the optimized expected execution time of each task; Based on the resource requirements of each task and the current resource status data of all available robots, a hybrid meta-heuristic algorithm is used to obtain a resource allocation scheme for multiple tasks and multiple target robots; The optimized matching relationship between the multiple tasks and the multiple target robots, the optimized execution start time of each task and the optimized estimated execution time of each task, the optimized task execution order of the multiple tasks, and the resource allocation plan for the multiple tasks and the multiple target robots are integrated to form a multi-task collaborative scheduling rule.

6. The method according to claim 5, characterized in that The resource types include: the processing power of the robot, the tools required for the task, the raw materials consumed during the task execution, the electrical energy required for the robot to operate, and the physical workspace required for the robot to perform the task; The resource allocation scheme includes: the resource type and quantity required for each task; the time dimension allocation plan of each type of resource during the task execution process, which includes the start time of resource use, the duration of resource use and the release time of the resource; and dynamic adjustment operations when the resource status of the target robot does not match the resource requirements of its corresponding task.

7. The method according to claim 6, characterized in that The controlling, according to the multi-task collaborative scheduling rule, of multiple target robots dynamically determined from all available robots in the smart factory to perform the multiple tasks specifically includes: According to the optimized matching relationship in the multi-task collaborative scheduling rule, a target robot matching each task is selected from all available robots; Activate the target robot of each task in turn according to the optimized task execution order in the multi-task collaborative scheduling rule; Sending a corresponding task instruction to each activated target robot, wherein the task instruction includes a task type, an optimized execution start time of the task, and an optimized estimated execution duration; According to the resource allocation scheme in the multi-task collaborative scheduling rule, resources required for the corresponding task to be executed are allocated to each activated target robot; After each task is completed, the resource status of the target robot corresponding to the task is updated, and the target robot is re-included in the identification list of all available robots; The above steps of activating the target robot of each task in turn and adding the target robot back into the queue of available robots are repeated until all tasks have been executed.

8. The method according to claim 7, characterized in that After allocating the resources required for the corresponding tasks to be performed to each activated target robot according to the resource allocation scheme in the multi-task collaborative scheduling rule, the following steps are also included: Monitor the current resource status data of each target robot in real time; When it is monitored that the current resource status data of any target robot does not match the resource requirements of its corresponding task, the dynamic adjustment operation is performed, and the dynamic adjustment operation includes at least one of the following: Reassign the task to other target robots whose current resource state data matches the resource requirements of the task; Postpone the task until the original target robot obtains sufficient resources; Split the task into multiple subtasks and assign them to multiple target robots with complementary resources to complete them collaboratively; Resources that will be released within the expected waiting time are reserved for the target robot, and the reserved resources are immediately allocated to the target robot after being released.

9. The method according to claim 2, characterized in that: Determining the resource requirements of each task according to the task scale information specifically includes: Establish a mapping relationship model between task scale information and resource requirements; Using a machine learning algorithm to train the mapping relationship model; The task scale information is input into the trained mapping relationship model to predict the resource type and resource quantity required for the task.

10. A multi-task collaborative robot scheduling system for smart factories, characterized in that: The system comprises: A task input module is used to receive input data of multiple tasks to be executed by the smart factory, wherein the input data of each task includes: task type, dependencies between tasks, task priority, resource requirements of each task and expected completion time of each task; An initial order determination module, used to determine an initial task execution order of a plurality of tasks according to the task types, dependencies between the tasks and the task priorities; Resource status acquisition module, used to obtain the current resource status data of all available robots in the smart factory; A scheduling rule generation module, used to generate multi-task collaborative scheduling rules using a scheduling algorithm according to the resource requirements of each task, the expected completion time of each task, the initial task execution order, and the current resource status data of all available robots; The task execution control module is used to control multiple target robots dynamically determined from all available robots in the smart factory to execute the multiple tasks according to the multi-task collaborative scheduling rules.

Images