Scheduling method for information aggregation distribution among multiple robots in industrial scene

Abstract

The invention relates to the technical field of information distribution scheduling, and discloses a scheduling method for information aggregation distribution among multiple robots in an industrial scene, comprising the following steps: S1, performing acquisition and structured processing on task requirements and environmental data in the industrial scene through an information modeling component; s2, constructing a data flow path according to the collected information; s3, state data and task execution data from the multiple robots are aligned and merged according to the time dimension and the semantic dimension; s4, judging a matching relationship between the task set and the robot set on the basis of the global industrial information; s5, packaging the task instruction and information related to the task; and S6, state return in the execution process of the robot is monitored. Through aggregation of time, space and semantics, the system can more accurately understand the space-time semantic relationship of the state of the robot, and task allocation errors caused by inconsistent or misaligned data are reduced, so that the task allocation precision and efficiency are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information distribution scheduling, in particular to a scheduling method for information aggregation and distribution among multiple robots in an industrial scenario. BACKGROUND

[0002] In the industrial field, with the continuous improvement of automation and intelligence level, the collaborative work of multiple robot systems has become the key to improving production efficiency and flexibility. Especially in complex manufacturing and logistics environments, Multi-Robot Systems (MRS) can significantly improve the overall system performance by cooperating to perform tasks. In order to achieve this goal, information sharing and collaborative scheduling among robots are particularly important.

[0003] Information aggregation and distribution is one of the core technologies in multi-robot system collaboration. In the industrial scenario, robots often need to make decisions and actions based on real-time data, which may involve position, state, sensor feedback, environmental changes, etc. In order to enable multiple robots to work efficiently under a unified goal, an effective mechanism is needed to aggregate information from different robots and distribute it to relevant robots, ensuring the consistency and timeliness of information among robots.

[0004] In related technologies, in the task allocation process of multi-robot systems, data is usually processed based on time or space dimensions only, lacking the ability to jointly model time, space and semantics, resulting in the uploaded data of different robots being unable to fully align in semantics. When data lacks cross-dimensional relevance, the system is difficult to identify the relationship between robot state, environmental changes and task demand in a timely manner, thus decision bias caused by inconsistent or misaligned data semantics is likely to occur in the task allocation stage. SUMMARY

[0005] To overcome the deficiencies of the prior art, the present application provides a scheduling method for information aggregation and distribution among multiple robots in an industrial scenario, which solves the problem that the uploaded data of different robots cannot be fully aligned in semantics, and when data lacks cross-dimensional relevance, the system is difficult to identify the relationship between robot state, environmental changes and task demand in a timely manner, thus decision bias caused by inconsistent or misaligned data semantics is likely to occur in the task allocation stage.

[0006] To achieve the above purpose, the present application is implemented by the following technical scheme: a scheduling method for information aggregation and distribution among multiple robots in an industrial scenario, comprising the following steps: S1. Incorporate multiple robots with task capability description, state information interface and environmental perception capabilities into a unified information model, and collect and structure data on task requirements, robot status and environment in industrial scenarios through information modeling components. S2. Construct a data flow path based on the collected information, and define the data transmission from the robot to the edge node and from the edge node to the aggregation scheduling layer to form a continuous data input link; S3. Align and merge state data, environmental data, and task execution data from multiple robots according to time, space, and semantic dimensions to form global industrial information; S4. Based on global industrial information, determine the matching relationship between the task set and the robot set, and generate a task allocation scheme through the scheduling decision component; S5. The task instructions and task-related information packages are split and distributed according to the different needs of the robots, and the instructions are sent to the corresponding robots through the communication path. S6. Monitor the status feedback during the robot's execution process, and detect and handle abnormal situations based on the link status and node operation status; S7. Update the task priority, robot capability parameters, and task execution cost parameters based on the feedback of the robot's execution results, in order to build a closed-loop adjustment of the scheduling model.

[0007] Preferably, S1 includes: The position and orientation of the robot in the industrial setting are obtained through position sensing components; The current task execution status of the robot is obtained through the task status acquisition component; The environmental perception component collects obstacle information and equipment operation information in the scene to obtain a unified information model including position, attitude, environmental features and task status.

[0008] Preferably, step S3 includes the following steps: Timestamp calibration of multiple time series data to construct a unified time axis; To construct a unified spatial reference, coordinate transformation is performed on pose information in different coordinate systems. Task execution tags and environmental semantic tags are categorized and integrated to construct a semantic hierarchical structure; Among them, time aggregation is performed according to a unified time axis alignment method, spatial aggregation is aligned according to coordinate transformation matrix, and semantic aggregation is grouped based on label classification rules.

[0009] Preferably, S4 includes: Set a set of capability parameters for each robot, and a set of requirement parameters for each task; Construct a matching cost based on the difference between the robot's capability parameters and the task requirement parameters; The task allocation is solved based on the matching cost to obtain the correspondence between the robot and the task; The difference includes a weighted sum of differences in execution time estimation, path resource usage, and energy consumption.

[0010] Preferably, step S4 further includes the following step: Detect the area occupied by the path between robots; Mark the area usage when the robot performs tasks; Configure access control conditions for shared resources; Among them, conflict detection is based on path overlap, area occupancy status and resource access conflict, and is used to form constraints in the scheduling solution phase.

[0011] Preferably, step S5 includes the following steps: Construct a set of information packets that includes basic status packets, extended information packets, and execution instruction packets; The information package is divided into layers based on the robot's task execution requirements; Depending on the data channel type, information packets are sent using point-to-point, multicast, or broadcast methods; The basic state package includes parameters for the robot's next action; the extended information package includes a local environment model; and the execution instruction package includes a set of path nodes.

[0012] Preferably, S6 includes: Set link quality monitoring parameters; Configure node operation status monitoring parameters; Set parameters for monitoring the robot's task execution status; Specifically, a scheduling rollback process is triggered when the link quality is lower than a preset threshold or when a node heartbeat is abnormal, and a scheduling replanning process is triggered when the robot task execution status is abnormal.

[0013] Preferably, step S7 includes the following steps: Define a set of task priority weights; Define the set of robot capability parameters; Define the set of parameters for task execution cost; Update task priority weights based on robot execution deviation; Update the robot's capability parameters based on the task completion quality parameters; Update the task execution cost parameters based on resource consumption.

[0014] Preferably, S7 further includes: The input set for the next round of task scheduling is adjusted based on the updated model with the correct parameters. The constraint boundaries of the scheduling model are modified based on the parameter change trend; Specifically, task priority is adjusted based on task delay deviation, robot capability parameters are corrected based on task execution stability indicators, and task cost parameters are adjusted based on execution resource usage.

[0015] A scheduling system for information aggregation and distribution among multiple robots in an industrial setting, the system comprising: The information modeling module is used to collect task requirements, robot status data, and environmental data, and to build a unified information model. The data acquisition module is used to establish the data flow from the robot to the edge node and from the edge node to the scheduling layer; The information aggregation module is used to perform time aggregation, spatial aggregation, and semantic aggregation on multi-source data; The scheduling decision module is used to generate task allocation schemes based on aggregated information; The information distribution module is used to break down task instructions into layers and send them in a point-to-point, multicast, or broadcast manner; The monitoring module is used to monitor link quality, node status, and robot status. The scheduling optimization module is used to update scheduling parameters and form a closed-loop adjustment based on the robot's execution results.

[0016] This invention provides a scheduling method for information aggregation and distribution among multiple robots in industrial scenarios. It offers the following advantages: 1. By aggregating time, space, and semantics, this invention enables the system to more accurately understand the spatiotemporal semantic relationships of robot state, environmental information, and task requirements, reducing task allocation errors caused by data inconsistency or misalignment, thereby improving the accuracy and efficiency of task allocation.

[0017] 2. By constructing the difference between robot capability parameters and task requirement parameters, the system can accurately calculate the matching cost between the task and the robot, and schedule tasks according to the principle of minimizing the cost, thereby ensuring that the task can be executed by the most suitable robot, improving resource utilization and scheduling efficiency.

[0018] 3. By monitoring link quality, node status, and robot task execution in real time, this invention can promptly identify anomalies and trigger scheduling rollback or replanning mechanisms. When network problems or robot malfunctions occur, the system can quickly adjust, improving the continuity and stability of task execution. Attached Figure Description

[0019] Figure 1This is a flowchart illustrating the scheduling method for information aggregation and distribution among multiple robots in an industrial setting according to the present invention. Figure 2 This is a system architecture diagram of the scheduling system for information aggregation and distribution among multiple robots in an industrial scenario according to the present invention. Detailed Implementation

[0020] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example: Please see the appendix Figure 1 This invention provides a scheduling method for information aggregation and distribution among multiple robots in an industrial scenario, comprising the following steps: S1. Incorporate multiple robots with task capability description, state information interface and environmental perception capabilities into a unified information model, and collect and structure data on task requirements, robot status and environment in industrial scenarios through information modeling components. S1 includes: The position and orientation of the robot in the industrial setting are obtained through position sensing components; The current task execution status of the robot is obtained through the task status acquisition component; The environmental perception component collects obstacle information and equipment operation information in the scene to obtain a unified information model including position, attitude, environmental features and task status.

[0022] Specifically, in pose estimation, an extended Kalman filter is used to process multi-source sensor data, which includes the following two steps: Prediction step, let the state vector be... It is a combination of the robot's position, velocity, and attitude information, represented as: ; in, It's the robot's location. It's the robot's speed. It is the robot's quaternion representation of its posture; In Kalman filtering, the system's prediction equation is used to predict the current state based on the state estimate from the previous time step and the current control input. Assume... It controls the input, and the prediction formula is: ; in, It is a motion model of the system, describing how the state changes from Time shift time.

[0023] The core of Kalman filtering is the covariance matrix, which describes the uncertainty of the state. The formula for updating the covariance matrix in the prediction step is: ; in: It is the covariance matrix of the predicted state. It is the state transition matrix, which describes how the covariance of the previous time step is propagated to the current time step according to the dynamic model of the system. It is the process noise covariance, representing unpredictable noise in the model; The update steps are as follows: When there are observations... At this time, Kalman gain is used to adjust the predicted state. The formula for calculating Kalman gain is: ; in: It is the observation matrix, which maps the state space to the observation space. It is the observation noise covariance, representing the uncertainty of the observation; based on the observed values... and Kalman gain Update state estimate: ; in It is an observation model that maps the predicted state to the observation space; the updated covariance matrix is: ; Then, through quaternions and rotation transformations, where quaternions are a mathematical tool used to represent rotations, quaternions... With rotation matrix The conversion relationship between them is as follows: ; This means through quaternions A rotation matrix can be calculated. It describes rotation from a local coordinate system to a global coordinate system. In robot systems, coordinate transformations typically take the form of rigid body transformations, which use rotation matrices and translation vectors to transform a point in one coordinate system to another. Sensor coordinate system One point in the middle, From the sensor coordinate system To the global reference coordinate system The rotation matrix, If it is a translation vector, then the position of that point in the global coordinate system... It can be calculated using the following formula: ; in: It is a rotation matrix (3x3) describing the rotation from the sensor coordinate system. To the global reference coordinate system rotation, It is a translation vector (3x1) that describes the sensor coordinate system. and global coordinate system In multi-sensor fusion, we need to propagate uncertainty during the translation between sensors. This is typically achieved using a covariance matrix. To describe uncertainty, Bayesian fusion methods can be used to combine measurements from multiple sensors, for two independent measurements. and and its corresponding covariance matrix and The position after fusion Covariance Matrix It can be calculated using the following formula: ; in: and yes and The inverse of the covariance matrix; and These are measurement results from two different sensors; The above describes in detail the mathematical models of robot pose estimation, covariance propagation, quaternions and rotation transformation, providing a complete technical background and implementation principle. In the process of multi-sensor fusion, the covariance matrix is ​​used to propagate and merge the uncertainties of data from different sensors, ensuring that the final output state estimate has high accuracy and stability.

[0024] S2. Construct a data flow path based on the collected information, and define the data transmission from the robot to the edge node and from the edge node to the aggregation scheduling layer to form a continuous data input link; S3. Align and merge state data, environmental data, and task execution data from multiple robots according to time, space, and semantic dimensions to form global industrial information; S3 includes the following steps: Timestamp calibration of multiple time series data to construct a unified time axis; To construct a unified spatial reference, coordinate transformation is performed on pose information in different coordinate systems. Task execution tags and environmental semantic tags are categorized and integrated to construct a semantic hierarchical structure; Among them, time aggregation is performed according to a unified time axis alignment method, spatial aggregation is aligned according to coordinate transformation matrix, and semantic aggregation is grouped based on label classification rules.

[0025] Specifically, the purpose of time alignment is to align data from different sources onto the same unified timeline. This process involves timestamp calibration of multiple time series data sets, ensuring that each data source has corresponding data values ​​at the same point in time. The specific implementation is as follows: Given a certain point in time lie in and Between (i.e.) The linear interpolation formula is: ; in: and These represent the sensors. At any moment and Data values; The sensor is obtained through interpolation on a unified time axis. At the point of time Data values; For the time aggregation of data from multiple sensors, a weighted average method is used: ; in: It is a sensor The weight. It is the aggregated result after time alignment, so that all data sources can be accurately aligned to a unified timeline; The purpose of spatial alignment is to transform data points from different coordinate systems to the same unified spatial reference. This typically requires coordinate transformation of the pose data from each sensor to ensure that all data are in the same reference coordinate system. Assuming there are two coordinate systems... and data points In coordinate system The following is represented as The goal is to transform it to a coordinate system. In this process, the pose is transformed through rigid body transformation, and the transformation formula is as follows: ; in: From coordinate system Transform to rotation matrix; It is the translation vector between the two coordinate systems; It is a point In coordinate system The following representation; if the pose is represented by quaternions, then the rotation relation is implemented through quaternion multiplication: ; in: and These represent the orientations of the two coordinate systems, respectively. It is a quaternion The inverse quaternion; From coordinate system arrive Rotational relationship; The spatial aggregation of data points from multiple sensors in a unified coordinate system can be calculated using the following formula: ; in: It is a sensor The original spatial point.

[0026] and These are the rotation matrix and translation vector between the sensor coordinate system and the unified coordinate system, respectively.

[0027] It refers to the unified location after spatial alignment and aggregation. Semantic alignment involves classifying and integrating task execution labels and environmental semantic labels to form a unified semantic structure. This step involves classifying labels according to semantic rules and constructing a semantic hierarchical structure. Assume the label set is... A classification function maps labels to different categories. The expression for a classification function is as follows: ; in: It is a classification rule that determines the category based on the content of the label. It is a tag The semantic categories to which they are mapped are weighted and averaged during the semantic aggregation process to obtain the final semantic aggregation result for that category, as shown in the following formula: ; in, It is classified as the first A collection of class tags; It is a tag The weights; It is a category The final semantic aggregation result; Through the above steps, the temporal, spatial, and semantic information of the data is effectively aligned, ultimately forming a unified spatiotemporal semantic data structure. The specific process includes: Multiple time series are aligned using linear interpolation and time aggregation is performed using weighted averaging. Pose transformation between different coordinate systems is achieved through rigid body transformation and quaternion rotation, thus completing spatial alignment and aggregation. Task execution labels and environmental semantic labels are classified and integrated using label classification and clustering methods to obtain semantic aggregation results. Through multi-source sensor data, accurate alignment can be achieved in time, space, and semantics, forming a unified spatiotemporal semantic data structure that can be directly applied to downstream analysis, prediction, and decision-making tasks.

[0028] S4. Based on global industrial information, determine the matching relationship between the task set and the robot set, and generate a task allocation scheme through the scheduling decision component; S4 includes: Set a set of capability parameters for each robot, and a set of requirement parameters for each task; Construct a matching cost based on the difference between the robot's capability parameters and the task requirement parameters; The task allocation is solved based on the matching cost to obtain the correspondence between the robot and the task; The difference includes a weighted sum of differences in execution time estimation, path resource usage, and energy consumption.

[0029] Specifically, the first step is to define a set of capability parameters for each robot. These capability parameters describe the robot's basic performance under typical tasks. Capability parameters include execution time prediction capability Path resource utilization capacity Energy consumption capacity The three parameters correspond to the main performance dimensions of the robot during task execution, and are used for subsequent evaluation of the degree of adaptation; At the same time, a set of requirement parameters is set for each task. The required parameters include execution time requirements. Path resource requirements Energy consumption demand These requirements reflect the consumption of robot capabilities during task completion, enabling the scheduling system to accurately determine the extent to which tasks consume robot resources. After the capabilities and requirements are determined, the difference between the two is calculated to measure four scenarios between the robot and the task. The difference is calculated as the deviation of the corresponding parameters: Execution time prediction error is ; Path resource usage difference

[0030] The difference in energy consumption is Three discrepancy metrics are used to construct the basic indicators for matching calculations. Then, a weighted summation method is used to build the matching cost. The matching cost reflects the robot's suitability for a particular task, and the formula is as follows: ; in , , These three factors serve as weights, used to adjust the proportions of time difference, path resource difference, and energy consumption difference in cost calculation. The sum of these weights is 1, and none are negative, ensuring the cost calculation has physical meaning and is controllable. Based on the constructed matching cost, this embodiment solves for task allocation. The goal is to minimize the total cost to obtain the correspondence between the robot and the task. Whether a task is executed by the robot is determined by… This indicates that matching can be achieved using linear programming, the Hungarian algorithm, or other minimum-cost matching methods, provided that the constraint that each task can only be assigned to one robot is satisfied. And each robot can perform at most one task, that is After solving, the optimal task-robot assignment result can be obtained.

[0031] S4 also includes the following steps: Detect the area occupied by the path between robots; Mark the area usage when the robot performs tasks; Configure access control conditions for shared resources; Among them, conflict detection is based on path overlap, area occupancy status and resource access conflict, and is used to form constraints in the scheduling solution phase.

[0032] Specifically, the path occupancy areas between robots are detected. Detection methods can be based on the robot's planned path trajectory, extracting path segments, occupancy blocks, or grid regions. Simple discretization can be performed in the time dimension, marking the path position corresponding to each time slice to obtain the path occupancy distribution. This data serves as the basis for subsequent conflict judgment, used to determine whether robots are likely to enter the same area within a future period. Simultaneously, the area usage during robot task execution is identified. Area usage can be refined to stations, work blocks, and channel resources. The identification method is relatively straightforward, recording the location points, operating range, and equipment occupancy range involved in the task steps. Each task corresponds to an area usage set, indicating which spaces or facilities are occupied during task execution. Access control conditions for shared resources also need to be set. Shared resources can be workbenches, material racks, charging stations, or critical equipment that robots must access together. Access control conditions typically include information such as the maximum concurrent access count, access order restrictions, and access time windows. These conditions constitute the source of constraints during scheduling, enabling scheduling to avoid resource contention. In this embodiment, conflict detection is jointly determined by path overlap, area occupancy status, and resource access conflict quantity. Path overlap is used to determine whether different robots simultaneously occupy a path; area occupancy status is used to determine whether they will simultaneously enter the same work area when executing tasks; and resource access conflict quantity is used to identify whether shared resources will experience access exceeding limits or access order conflicts. These three types of detection results together constitute a conflict index, which can be used to determine whether a matching allocation scheme is feasible. The conflict detection results are added to the solution model as constraints in the scheduling solution phase. For example, in the matching solution, constraints are set to prevent two robots from simultaneously executing tasks that would lead to path overlap, or to prevent them from accessing the same restricted resource within the same time slice.

[0033] S5. The task instructions and task-related information packages are split and distributed according to the different needs of the robots, and the instructions are sent to the corresponding robots through the communication path. S5 includes the following steps: Construct a set of information packets that includes basic status packets, extended information packets, and execution instruction packets; The information package is divided into layers based on the robot's task execution requirements; Depending on the data channel type, information packets are sent using point-to-point, multicast, or broadcast methods; The basic state package includes parameters for the robot's next action; the extended information package includes a local environment model; and the execution instruction package includes a set of path nodes.

[0034] Specifically, the system first constructs an information package set, which consists of a basic state package, an extended information package, and an execution instruction package. The basic state package contains parameters for the robot's next action, such as the next motion point or local control instructions. The extended information package carries a local environment model, including local occupancy states, dynamic object distribution, etc. The execution instruction package contains a set of path nodes, which are used to describe the complete path sequence when the robot performs a task. Subsequently, the information packets are layered and split according to the robot's task execution requirements. The split data is sent as needed: basic status packets are sent when only synchronous actions are required, extended information packets are sent when local environmental support is needed, and execution instruction packets are sent when path adjustments are needed. This splitting makes the information volume more concise, the communication load easier to control, and the transmission method is selected according to the data channel type. In the point-to-point channel, the information is directly transmitted to the target robot; in the multicast channel, extended information packets are sent to a specific robot group; and in the broadcast channel, globally updated data can be synchronized to all robots. These three modes make the communication structure more flexible in different task scenarios. Therefore, by using the above method, the structure, splitting granularity, and transmission path of the information packets can remain controllable, enabling robots to obtain the required data at a lower communication cost when performing tasks. This is suitable for high-concurrency, multi-tasking industrial scenarios.

[0035] S6. Monitor the status feedback during the robot's execution process, and detect and handle abnormal situations based on the link status and node operation status; S6 includes: Set link quality monitoring parameters; Configure node operation status monitoring parameters; Set parameters for monitoring the robot's task execution status; Specifically, a scheduling rollback process is triggered when the link quality is lower than a preset threshold or when a node heartbeat is abnormal, and a scheduling replanning process is triggered when the robot task execution status is abnormal.

[0036] Specifically, firstly, link quality monitoring parameters are set to detect the quality of the communication link between robots in real time, and to monitor indicators such as signal strength, latency and packet loss rate. When the link quality is lower than the preset threshold, the system will determine that the link is unstable, trigger the scheduling rollback process, and re-evaluate the task allocation. Secondly, node running status monitoring parameters are set to monitor the working status of each node. The node status includes CPU utilization, memory usage, task execution time, etc. If the node's heartbeat signal is abnormal or the running status is abnormal, the scheduling rollback process will be triggered to ensure the correct execution of the task. Parameters for monitoring robot task execution status are set to detect abnormal states of the robot during task execution, including whether the robot is stuck, malfunctioning, or has timed out. When a robot task execution is abnormal, the system will initiate a scheduling replanning process to readjust task allocation and execution order to ensure task completion. By setting these monitoring parameters, the system can dynamically respond to changes in links, nodes, and robot states, improving its robustness and adaptability.

[0037] S7. Update the task priority, robot capability parameters, and task execution cost parameters based on the feedback of the robot's execution results, in order to build a closed-loop adjustment of the scheduling model.

[0038] S7 includes the following steps: Define a set of task priority weights; Define the set of robot capability parameters; Define the set of parameters for task execution cost; Update task priority weights based on robot execution deviation; Update the robot's capability parameters based on the task completion quality parameters; Update the task execution cost parameters based on resource consumption.

[0039] Specifically, firstly, a set of task priority weights is defined to describe the importance of each task in scheduling. Weights can be preset based on task urgency, timeliness, or business needs. Secondly, a set of robot capability parameters is defined. Capability parameters characterize the robot's overall performance in task execution, including speed, path-passing ability, and energy efficiency. Thirdly, a set of task execution cost parameters is defined to reflect the resource consumption required to complete the task, including path length, time overhead, and energy consumption. During task execution, the system updates task priority weights based on robot execution deviation. Execution deviation can be determined by task completion time deviation, trajectory deviation, or energy consumption deviation. The larger the deviation, the higher or lower the priority of the relevant task is prioritized to enhance scheduling adaptability. The system also updates robot capability parameters based on task completion quality parameters. Task completion quality can include accuracy, stability, and time consumption. If a robot consistently performs well or poorly, its capability parameters will be adjusted accordingly to more accurately reflect its available performance. Finally, the system updates task execution cost parameters based on resource consumption. Resource consumption can come from energy consumption, path congestion, or equipment usage time recorded during actual execution, and is used to adjust the cost assessment of the task in subsequent scheduling, making the cost model closer to real-world execution conditions.

[0040] S7 also includes: The input set for the next round of task scheduling is adjusted based on the updated model with the correct parameters. The constraint boundaries of the scheduling model are modified based on the parameter change trend; Specifically, task priority is adjusted based on task delay deviation, robot capability parameters are corrected based on task execution stability indicators, and task cost parameters are adjusted based on execution resource usage.

[0041] Specifically, task priority is adjusted based on task delay deviation. When the task completion time delay is large, the system will increase its priority to ensure that the task is scheduled first and reduce subsequent delays. The robot capability parameters are corrected based on task execution stability indicators. Stability analysis after task completion, including stagnation time and path deviation, will affect the update of robot capabilities, improve the capability value of well-performing robots, and reduce the capability value of poorly performing robots. The task cost parameters are adjusted based on the resource consumption recorded during execution, reflecting the actual resource consumption of task execution.

[0042] Please see the appendix Figure 2 A scheduling system for information aggregation and distribution among multiple robots in industrial scenarios, comprising: The information modeling module is used to collect task requirements, robot status data, and environmental data, and to build a unified information model. The data acquisition module is used to establish the data flow from the robot to the edge node and from the edge node to the scheduling layer; The information aggregation module is used to perform time aggregation, spatial aggregation, and semantic aggregation on multi-source data; The scheduling decision module is used to generate task allocation schemes based on aggregated information; The information distribution module is used to break down task instructions into layers and send them in a point-to-point, multicast, or broadcast manner; The monitoring module is used to monitor link quality, node status, and robot status. The scheduling optimization module is used to update scheduling parameters and form a closed-loop adjustment based on the robot's execution results.

[0043] Specifically, the information modeling module is responsible for collecting task requirements, robot status data, and environmental data to build a unified information model. Task requirements include execution time, path resource usage, and energy consumption; robot status data includes position, speed, and execution progress; and environmental data includes the occupancy status of the work area and the distribution of obstacles. The data acquisition module is used to establish the data flow from the robot to the edge node and from the edge node to the scheduling layer. It sends its own status data, sensor information and other data to the edge node through the communication link. The edge node then forwards the aggregated information to the scheduling layer. The information aggregation module is responsible for multi-dimensional aggregation of data from different data sources, including: Time aggregation: It summarizes data from different points in time to provide time-series information; Spatial aggregation: Merging status information from different locations to provide spatial distribution information; Semantic aggregation: Analyzing and fusing the meaning of data to ensure that scheduling decisions can be based on more precise contextual information; The scheduling decision module generates a task allocation scheme based on aggregated information, and determines the allocation strategy for each task by taking into account factors such as robot capabilities, task requirements, and environmental changes. The information distribution module is used to split task instructions into layers and to transmit information using different data transmission methods according to the needs of different tasks and the communication network conditions. Based on the importance and urgency of the task instructions, it selects point-to-point, multicast, or broadcast methods for task allocation. The monitoring module is responsible for real-time monitoring of link quality, node status and robot execution status. Link quality monitoring can be judged by indicators such as network latency and packet loss rate. Node status monitoring can check the health status of control center or edge nodes. Robot status monitoring ensures that the robot does not experience any abnormalities such as failure, stagnation or timeout during execution. The scheduling optimization module dynamically updates the scheduling parameters based on the feedback information from the robot's execution and forms a closed-loop adjustment. The scheduling optimization module continuously monitors the task execution progress, resource consumption, and execution quality, and adjusts the task priority, robot capability parameters, and execution costs when necessary to ensure that the scheduling system always operates in the optimal state.

[0044] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A scheduling method for information aggregation and distribution among multiple robots in an industrial setting, characterized in that, Includes the following steps: S1. Incorporate multiple robots with task capability description, state information interface and environmental perception capabilities into a unified information model, and collect and structure data on task requirements, robot status and environment in industrial scenarios through information modeling components. S2. Construct a data flow path based on the collected information, and define the data transmission from the robot to the edge node and from the edge node to the aggregation scheduling layer to form a continuous data input link; S3. Align and merge state data, environmental data, and task execution data from multiple robots according to time, space, and semantic dimensions to form global industrial information; S4. Based on global industrial information, determine the matching relationship between the task set and the robot set, and generate a task allocation scheme through the scheduling decision component; S5. The task instructions and task-related information packages are split and distributed according to the different needs of the robots, and the instructions are sent to the corresponding robots through the communication path. S6. Monitor the status feedback during the robot's execution process, and detect and handle abnormal situations based on the link status and node operation status; S7. Update the task priority, robot capability parameters, and task execution cost parameters based on the feedback of the robot's execution results, in order to build a closed-loop adjustment of the scheduling model.

2. The scheduling method for information aggregation and distribution among multiple robots in an industrial scenario according to claim 1, characterized in that: S1 includes: The position and orientation of the robot in the industrial setting are obtained through position sensing components; The current task execution status of the robot is obtained through the task status acquisition component; The environmental perception component collects obstacle information and equipment operation information in the scene to obtain a unified information model including position, attitude, environmental features and task status.

3. The scheduling method for information aggregation and distribution among multiple robots in an industrial scenario according to claim 1, characterized in that: S3 includes the following steps: Timestamp calibration of multiple time series data to construct a unified time axis; To construct a unified spatial reference, coordinate transformation is performed on pose information in different coordinate systems. Task execution tags and environmental semantic tags are categorized and integrated to construct a semantic hierarchical structure; Among them, time aggregation is performed according to a unified time axis alignment method, spatial aggregation is aligned according to coordinate transformation matrix, and semantic aggregation is grouped based on label classification rules.

4. The scheduling method for information aggregation and distribution among multiple robots in an industrial scenario according to claim 1, characterized in that: S4 includes: Set a set of capability parameters for each robot, and a set of requirement parameters for each task; Construct a matching cost based on the difference between the robot's capability parameters and the task requirement parameters; The task allocation is solved based on the matching cost to obtain the correspondence between the robot and the task; The difference includes a weighted sum of differences in execution time estimation, path resource usage, and energy consumption.

5. The scheduling method for information aggregation and distribution among multiple robots in an industrial scenario according to claim 1, characterized in that: S4 further includes the following steps: Detect the area occupied by the path between robots; Mark the area usage when the robot performs tasks; Configure access control conditions for shared resources; Among them, conflict detection is based on path overlap, area occupancy status and resource access conflict, and is used to form constraints in the scheduling solution phase.

6. The scheduling method for information aggregation and distribution among multiple robots in an industrial scenario according to claim 1, characterized in that: S5 includes the following steps: Construct a set of information packets that includes basic status packets, extended information packets, and execution instruction packets; The information package is divided into layers based on the robot's task execution requirements; Depending on the data channel type, information packets are sent using point-to-point, multicast, or broadcast methods; The basic state package includes parameters for the robot's next action; the extended information package includes a local environment model; and the execution instruction package includes a set of path nodes.

7. The scheduling method for information aggregation and distribution among multiple robots in an industrial scenario according to claim 1, characterized in that: S6 includes: Set link quality monitoring parameters; Configure node operation status monitoring parameters; Set parameters for monitoring the robot's task execution status; Specifically, a scheduling rollback process is triggered when the link quality is lower than a preset threshold or when a node heartbeat is abnormal, and a scheduling replanning process is triggered when the robot task execution status is abnormal.

8. The scheduling method for information aggregation and distribution among multiple robots in an industrial scenario according to claim 1, characterized in that: S7 includes the following steps: Define a set of task priority weights; Define the set of robot capability parameters; Define the set of parameters for task execution cost; Update task priority weights based on robot execution deviation; Update the robot's capability parameters based on the task completion quality parameters; Update the task execution cost parameters based on resource consumption.

9. The scheduling method for information aggregation and distribution among multiple robots in an industrial scenario according to claim 1, characterized in that: The S7 also includes: The input set for the next round of task scheduling is adjusted based on the updated model with the correct parameters. The constraint boundaries of the scheduling model are modified based on the parameter change trend; Specifically, task priority is adjusted based on task delay deviation, robot capability parameters are corrected based on task execution stability indicators, and task cost parameters are adjusted based on execution resource usage.

10. A scheduling system for information aggregation and distribution among multiple robots in an industrial setting, applied to the scheduling method for information aggregation and distribution among multiple robots in an industrial setting as described in any one of claims 1-9, characterized in that, The system includes: The information modeling module is used to collect task requirements, robot status data, and environmental data, and to build a unified information model. The data acquisition module is used to establish the data flow from the robot to the edge node and from the edge node to the scheduling layer; The information aggregation module is used to perform time aggregation, spatial aggregation, and semantic aggregation on multi-source data; The scheduling decision module is used to generate task allocation schemes based on aggregated information; The information distribution module is used to break down task instructions into layers and send them in a point-to-point, multicast, or broadcast manner; The monitoring module is used to monitor link quality, node status, and robot status. The scheduling optimization module is used to update scheduling parameters and form a closed-loop adjustment based on the robot's execution results.

Images