Multi-device cooperative control method and system under autonomous cooperation algorithm

By constructing a multi-dimensional state matrix and obstacle avoidance reinforcement learning, the problem of insufficient adaptability in multi-device collaborative control is solved, and efficient and reliable collaborative control of equipment in smart factories is achieved.

CN120704212APending Publication Date: 2025-09-26HARBIN SAISI TECH CO LTD
View PDF 0 Cites 7 Cited by

Patent Information

Application Number
CN202510855257.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-06-25
Publication Date
2025-09-26

AI Technical Summary

Technical Problem

Existing multi-device collaborative control technology lacks the ability to adapt to dynamic operating environments and changes in equipment status, and is unable to effectively handle complex coupling relationships and sudden abnormal situations between devices, resulting in low overall collaborative reliability of equipment systems in smart factories.

Method used

By collecting physical property data, real-time operation data and production task data of smart factory equipment, a multi-dimensional state matrix is ​​constructed, spatial and temporal conflict patterns are identified, and collaborative conflict decoupling is performed. The smart factory scene graph is used for obstacle avoidance reinforcement learning, and a feasible coordination mode and global obstacle avoidance path are generated to build a target collaborative control solution.

Benefits of technology

It improves the reliability of multi-device collaborative control, realizes autonomous obstacle avoidance and collaborative scheduling of equipment in dynamic environments, and improves the overall performance and resource allocation efficiency of the production system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120704212A_ABST
    Figure CN120704212A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of equipment cooperative control, and discloses a multi-equipment cooperative control method and system under an autonomous cooperation algorithm, and the method comprises the steps: collecting the physical attribute data, real-time operation data, production task data and a smart factory scene graph of operation equipment under a multi-equipment cooperative task in a smart factory application; based on the physical attribute data and the real-time operation data, identifying a space conflict mode and a time conflict mode of the operation equipment to obtain collaborative conflict information; according to the production task data and the collaboration conflict information, carrying out collaboration conflict decoupling on the operation equipment to obtain a feasible collaboration mode; performing obstacle avoidance reinforcement learning on the operation equipment by using the smart factory scene graph to construct an obstacle avoidance global path of the operation equipment in the smart factory; and based on the feasible coordination mode and the obstacle avoidance global path, constructing a target coordination control scheme of the operation equipment so as to perform coordination control on the operation equipment. According to the invention, the reliability of multi-device cooperative control under the autonomous cooperation algorithm can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to a multi-device collaborative control method and system under an autonomous collaborative algorithm, and belongs to the technical field of device collaborative control. Background Art

[0002] The rapid development of intelligent manufacturing and industry is placing higher demands on the precision and efficiency of equipment collaboration in smart factory scenarios. Autonomous collaborative algorithms, as the core technology for efficient multi-device collaborative control, are becoming increasingly important. Through autonomous collaborative algorithms, multi-device collaborative control enables precise scheduling and real-time interaction across multiple devices, which is of great significance for improving the overall efficiency of production systems and optimizing resource allocation.

[0003] Currently, multi-device collaborative control is mostly based on traditional centralized or distributed control technologies, such as master-slave control architectures or hierarchical control modes. These technologies achieve collaboration between devices through preset control instructions or fixed communication protocols, and obtain device operating parameters under specific task processes. However, this approach usually relies on pre-set control rules for instruction distribution and lacks the ability to adapt to dynamic operating environments and changes in device status. In addition, when simple task allocation or priority scheduling algorithms are used to achieve collaborative control, it is often difficult to effectively handle the complex coupling relationships and sudden abnormal situations between devices, resulting in insufficient overall collaborative reliability of device systems in smart factory applications. Summary of the Invention

[0004] The present invention provides a multi-device collaborative control method and system under an autonomous collaborative algorithm, the main purpose of which is to improve the reliability of multi-device collaborative control under the autonomous collaborative algorithm.

[0005] To achieve the above objectives, the present invention provides a multi-device collaborative control method under an autonomous collaborative algorithm, comprising:

[0006] Collect physical attribute data, real-time operation data, production task data, and smart factory scenario diagrams of devices running in multi-device collaborative tasks in smart factory applications. The physical attribute data includes device type, physical dimensions, and operating parameter limitations; the real-time operation data includes location, speed, load, and energy consumption; and the production task data includes task type, priority, and delivery time.

[0007] constructing a multidimensional state matrix of the operating equipment based on the physical attribute data and the real-time operating data, and using the multidimensional state matrix to identify spatial conflict patterns and temporal conflict patterns of the operating equipment to obtain collaborative conflict information;

[0008] Decoupling the operating equipment from the collaborative conflicts according to the production task data and the collaborative conflict information to obtain a feasible coordination mode;

[0009] Using the smart factory scene graph, performing obstacle avoidance reinforcement learning on the operating equipment to construct a global obstacle avoidance path for the operating equipment in the smart factory;

[0010] Based on the feasible coordination mode and the obstacle avoidance global path, a target collaborative control scheme for the running equipment is constructed, and based on the target collaborative control scheme, the running equipment is subjected to device collaborative control.

[0011] Optionally, constructing a multidimensional state matrix of the operating device based on the physical property data and the real-time operating data includes:

[0012] performing data standardization processing on the physical attribute data and the real-time operation data to obtain a standardized data set;

[0013] Performing multidimensional feature screening on the standardized data set to obtain multidimensional key features;

[0014] Performing time series alignment processing on the features in the multidimensional key features to obtain multidimensional synchronous features;

[0015] The multi-dimensional synchronization feature is used to construct a multi-dimensional state matrix of the operating device.

[0016] Optionally, constructing a multidimensional state matrix of the operating device using the multidimensional synchronization feature includes:

[0017] Arranging the multi-dimensional synchronization features in time series to obtain sequence features;

[0018] Based on the sequence characteristics, a multidimensional state matrix of the operating device is constructed, wherein the multidimensional state matrix can be expressed in the following form:

[0019]

[0020] Among them, M represents the multidimensional state matrix, ID1 represents the ID of the first running device in the sequence feature, ID n Indicates the id of the nth running device in the sequence feature, x1 indicates the x-axis coordinate of the first running device in the sequence feature, x n Indicates the x-axis coordinate of the nth running device in the sequence feature, y1 indicates the y-axis coordinate of the first running device in the sequence feature, x n Indicates the y-axis coordinate of the nth running device in the sequence feature, v x1 represents the velocity corresponding to x1, v xn Indicates that x n The corresponding speed, v y1 represents the velocity corresponding to y1, v yn Indicates that yn Corresponding speed, E1 represents the load rate of the first running equipment in the sequence feature, E n represents the load rate of the nth running device in the sequence feature, α1 represents the energy consumption of the first running device in the sequence feature, α n represents the energy consumption of the nth running device in the sequence feature, β1 represents the priority of the first running device in the sequence feature, β n Indicates the priority of the nth running device in the sequence feature.

[0021] Optionally, using the multidimensional state matrix to identify the spatial conflict mode and the temporal conflict mode of the operating device to obtain the collaborative conflict information includes:

[0022] Extracting spatiotemporal data from the multidimensional state matrix to obtain a spatiotemporal attribute set of the device;

[0023] Using the device spatiotemporal attribute set, predicting the running time space trajectory of the running device to obtain a spatiotemporal trajectory set;

[0024] Performing a conflict grid mapping operation on the spatiotemporal trajectory set to obtain a spatiotemporal conflict heat map;

[0025] The spatial conflict pattern and the temporal conflict pattern of the operating equipment are analyzed by using the spatiotemporal conflict heat map to obtain collaborative conflict information.

[0026] Optionally, performing collaborative conflict decoupling on the operating equipment according to the production task data and the collaborative conflict information to obtain a feasible coordination mode includes:

[0027] Using the production task data and the collaborative conflict information, constructing a conflict-task association table for the operating equipment;

[0028] Performing feasible screening on the conflict-task association table based on a rule engine to obtain a basic decoupling solution set;

[0029] Performing genetic algorithm optimization on the basic decoupling solution set to obtain a Pareto optimal solution set;

[0030] Performing feasibility verification on the Pareto optimal solution set to extract executable coordination instructions from the Pareto optimal solution set;

[0031] The executable coordination instruction is used to perform device control coding operations on the running device to obtain a feasible coordination mode.

[0032] Optionally, using the smart factory scene graph, performing obstacle avoidance reinforcement learning on the operating equipment to construct a global obstacle avoidance path for the operating equipment in the smart factory, including:

[0033] Performing real-time scene updates on the smart factory scene graph to obtain an updated map;

[0034] Identifying geometric features of an update target corresponding to the update map;

[0035] Performing semantic annotation on the geometric features to obtain semantic annotation data;

[0036] Using the semantic annotation data, marking obstacles on the updated map to form an obstacle-marked map;

[0037] Querying the static state and action space of the running device in the obstacle-annotated map;

[0038] Obstacle avoidance reinforcement learning is performed on the operating equipment using the static state and the action space to construct a global obstacle avoidance path for the operating equipment in the smart factory.

[0039] Optionally, the updating map is annotated with obstacles using the semantic annotation data, and the obstacle annotation map includes:

[0040] identifying dynamic objects in the updated map using the semantically annotated data;

[0041] querying the trajectory data of the dynamic object, and using the trajectory data to mark the dangerous area on the updated map to obtain a dynamic prohibited area;

[0042] Performing probability grid modeling on the dynamic forbidden area, a grid area;

[0043] The posterior probability of obstacle distribution in the grid area is calculated using the following formula:

[0044]

[0045] Among them, prob t (i, j) represents the posterior probability of obstacle distribution, t represents time, (i, j) represents a grid in the grid area, prob t-1 (i, j) represents the prior probability of (i, j) at time t-1, p(z|u) represents the detection likelihood, p(z|-u) represents the false detection probability, z represents the latest observation data, u represents the probability that the sensor detects occupation when the grid is indeed occupied, and -u represents the probability that the sensor falsely reports occupation when the grid is actually empty;

[0046] Determining an obstacle occurrence probability distribution in the grid area based on the posterior probability;

[0047] The obstacle occurrence probability distribution is dynamically and statically fused with the updated map to obtain an obstacle annotation map.

[0048] Optionally, using the static state and the action space to perform obstacle avoidance reinforcement learning on the running device to construct a global obstacle avoidance path for the running device includes:

[0049] Performing state space definition on the static state and the action space to obtain a state vector;

[0050] Performing action discretization processing on the state vector to obtain a discrete action set;

[0051] configuring a reward function for the discrete action set;

[0052] constructing an initial obstacle avoidance path for the operating device using the discrete action set;

[0053] Based on the reward function, reinforcement learning is performed on the initial obstacle avoidance path to obtain an adaptive obstacle avoidance strategy;

[0054] Based on the adaptive obstacle avoidance strategy, a global obstacle avoidance path for the operating equipment in the smart factory is constructed.

[0055] Optionally, constructing a target collaborative control scheme for the operating device based on the feasible coordination mode and the obstacle avoidance global path includes:

[0056] Performing multi-source data fusion on the feasible coordination mode and the obstacle avoidance global path to obtain fused data;

[0057] Using the fused data, constructing task-path coordination instructions for the operating device;

[0058] Performing protocol conversion on the task-path coordination instruction to obtain a device control packet;

[0059] Using the device control package, performing coordinated control simulation on the running device;

[0060] A target coordinated control scheme for the operating equipment is constructed based on a simulation set of the coordinated control simulation.

[0061] In order to solve the above problems, the present invention also provides a multi-device collaborative control system under an autonomous collaborative algorithm, the system comprising:

[0062] An information collection module, configured to collect physical attribute data, real-time operation data, production task data, and smart factory scenario diagrams of devices running in multi-device collaborative tasks in smart factory applications. The physical attribute data includes device type, physical dimensions, and operating parameter limits; the real-time operation data includes location, speed, load, and energy consumption; and the production task data includes task type, priority, and delivery time.

[0063] a conflict information analysis module, configured to construct a multidimensional state matrix of the operating equipment based on the physical property data and the real-time operating data, and identify spatial conflict patterns and temporal conflict patterns of the operating equipment using the multidimensional state matrix to obtain collaborative conflict information;

[0064] a conflict decoupling module, configured to perform collaborative conflict decoupling on the operating equipment according to the production task data and the collaborative conflict information, and obtain a feasible coordination mode;

[0065] An obstacle avoidance path planning module, configured to utilize the smart factory scene graph to perform obstacle avoidance reinforcement learning on the operating equipment, so as to construct a global obstacle avoidance path for the operating equipment in the smart factory;

[0066] A collaborative control module is used to construct a target collaborative control scheme for the running equipment based on the feasible coordination mode and the obstacle avoidance global path, and to perform device collaborative control on the running equipment based on the target collaborative control scheme.

[0067] Compared with the problems described in the background technology, the embodiment of the present invention first connects the IoT sensor with the factory system to collect the physical properties of the equipment (type, size, parameter limit), real-time operation data (position, speed, energy consumption), production tasks (type, priority, delivery time) and scene diagrams, and converts the equipment, tasks and environment of the smart factory into a computable data set to achieve comprehensive digitalization of production factors; further, the present invention integrates physical properties and real-time operation data into a multi-dimensional state matrix, and converts fragmented data into structured conflict information through spatiotemporal trajectory prediction and conflict heat map analysis, making implicit risks explicit and providing precise targets for decoupling optimization; based on production task priority and conflict information, the present invention uses rules to Through screening, genetic algorithm optimization and feasibility verification, a device coordination mode without time and space conflicts is generated, and the conflict information is converted into an executable scheduling strategy to realize the dynamic matching of "task-equipment-resource"; the present invention uses the smart factory scene graph for real-time environment modeling, analyzes the dynamic obstacle distribution through semantic annotation and probabilistic grid, combines reinforcement learning to generate a global obstacle avoidance path, and integrates the static coordination mode with the dynamic environment constraints to enable the equipment to have autonomous obstacle avoidance capabilities; further, the present invention integrates the feasible coordination mode and the obstacle avoidance path, and constructs a comprehensive control solution through data fusion, protocol conversion, and simulation verification to realize the collaborative scheduling and real-time control of equipment, thereby improving the reliability of multi-device collaborative control under the autonomous collaborative algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 A flowchart of a multi-device collaborative control method under an autonomous collaborative algorithm provided by one embodiment of the present invention;

[0069] Figure 2A schematic diagram of a module for implementing a multi-device collaborative control system under the autonomous collaboration algorithm provided in one embodiment of the present invention.

[0070] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0071] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0072] The embodiments of the present application provide a multi-device collaborative control method and system under an autonomous collaborative algorithm. The execution subject of the multi-device collaborative control method under the autonomous collaborative algorithm includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the multi-device collaborative control system under the autonomous collaborative algorithm can be executed by software or hardware installed on a terminal device or a server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0073] Example 1:

[0074] Reference Figure 1 FIG2 is a flow chart of a multi-device collaborative control method under an autonomous collaborative algorithm provided by an embodiment of the present invention. In this embodiment, the multi-device collaborative control method under the autonomous collaborative algorithm includes:

[0075] S1. Collect physical property data, real-time operation data, production task data, and smart factory scenario diagrams of equipment running under multi-device collaborative tasks in smart factory applications. The physical property data includes equipment type, physical size, and operating parameter limitations; the real-time operation data includes location, speed, load, and energy consumption; and the production task data includes task type, priority, and delivery time.

[0076] By collecting physical property data, real-time operation data, production task data, and smart factory scene graphs from devices running under multi-device collaborative tasks in smart factory applications, the embodiments of the present invention can comprehensively collect information from four dimensions: device characteristics, operating status, task requirements, and factory environment, transforming the production factors of smart factories into analyzable data, thereby providing a basis for subsequent collaborative control. For example, in flexible production line logistics scenarios, dynamic obstacle avoidance of AGV clusters and path reconstruction during production line switching are achieved; in intelligent warehousing and sorting scenarios, resource sharing between robots and sorting belts and order priority scheduling are achieved; and in high-risk environment inspection scenarios, autonomous obstacle avoidance of explosion-proof robots and real-time monitoring of equipment status are achieved.

[0077] Among them, the smart factory refers to a modern factory that uses new-generation information technologies such as the Internet of Things, big data, artificial intelligence and 5G to deeply interconnect production equipment, materials, systems and other elements to achieve intelligent perception, analysis, decision-making and control of the production process. The physical attribute data includes equipment type, physical size, and operating parameter restrictions. The real-time operation data includes location, speed, load and energy consumption. The production task data includes task type, priority and delivery time.

[0078] During the specific implementation process, the physical properties (such as rated power) and real-time operating data (such as position and speed) of running equipment can be collected in real time through IoT sensors (such as lidar and RFID), and the factory MES system can be connected to obtain production task data (such as order priority); the SLAM algorithm can be used to combine multi-sensor (camera, lidar) data to build a smart factory scene map.

[0079] S2. Based on the physical property data and the real-time operation data, construct a multidimensional state matrix of the operating device, use the multidimensional state matrix to identify the spatial conflict mode and the temporal conflict mode of the operating device, and obtain collaborative conflict information.

[0080] The embodiment of the present invention constructs a multidimensional state matrix of the operating equipment based on the physical property data and the real-time operation data, which can integrate the equipment physical property data (such as static parameters such as size, load-bearing capacity, and range of motion) and real-time operation data (such as dynamic data such as position, speed, energy consumption, and fault codes) into a unified matrix structure, integrate and associate fragmented data, and improve the efficiency of equipment analysis in smart factories.

[0081] Among them, the multidimensional state matrix refers to a mathematical matrix that structures and quantifies the multidimensional operating state data of the target device, which is used to intuitively describe the device state and collaborative relationship, such as the coordinate position and speed in the physical dimension, the task volume and energy consumption in the load dimension, etc.

[0082] As an embodiment of the present invention, constructing a multidimensional state matrix of the operating device based on the physical property data and the real-time operating data includes:

[0083] performing data standardization processing on the physical attribute data and the real-time operation data to obtain a standardized data set;

[0084] Performing multidimensional feature screening on the standardized data set to obtain multidimensional key features;

[0085] Performing time series alignment processing on the features in the multidimensional key features to obtain multidimensional synchronous features;

[0086] The multi-dimensional synchronization feature is used to construct a multi-dimensional state matrix of the operating device.

[0087] The multidimensional key feature refers to, and the multidimensional synchronization feature refers to, a feature set obtained after time synchronization processing is performed on the multidimensional feature.

[0088] During the specific implementation process, Min-Max normalization or Z-score normalization methods can be used for physical property data (such as equipment size and load upper limit) and real-time operation data (such as speed and energy consumption) to map data of different dimensions to a unified range (such as [0, 1] or an interval with a mean of 0 and a standard deviation of 1) to obtain a standardized data set; based on the collaborative needs of smart factory equipment, through correlation analysis or variance analysis, features that have a significant impact on equipment status characterization and conflict detection (such as equipment position, speed and load rate) are screened out, and then redundant or less influential features are eliminated to obtain multidimensional key features; based on the timestamp of equipment status data collection, the missing values ​​in the time series are filled by interpolation methods (such as linear interpolation) to ensure that all features have corresponding data at the same time point, and then multidimensional synchronous features are obtained.

[0089] Exemplarily, the constructing the multidimensional state matrix of the operating device by utilizing the multidimensional synchronization feature includes:

[0090] Arranging the multi-dimensional synchronization features in time series to obtain sequence features;

[0091] Based on the sequence characteristics, a multidimensional state matrix of the operating device is constructed, wherein the multidimensional state matrix can be expressed in the following form:

[0092]

[0093] Among them, M represents the multidimensional state matrix, ID1 represents the ID of the first running device in the sequence feature, ID n Indicates the id of the nth running device in the sequence feature, x1 indicates the x-axis coordinate of the first running device in the sequence feature, x n Indicates the x-axis coordinate of the nth running device in the sequence feature, y1 indicates the y-axis coordinate of the first running device in the sequence feature, x n Indicates the y-axis coordinate of the nth running device in the sequence feature, v x1 represents the velocity corresponding to x1, v xn Indicates that x n The corresponding speed, v y1 represents the velocity corresponding to y1, v yn Indicates that y n Corresponding speed, E1 represents the load rate of the first running equipment in the sequence feature, E nrepresents the load rate of the nth running device in the sequence feature, α1 represents the energy consumption of the first running device in the sequence feature, α n represents the energy consumption of the nth running device in the sequence feature, β1 represents the priority of the first running device in the sequence feature, β n Indicates the priority of the nth running device in the sequence feature.

[0094] It should be further explained that the multidimensional state matrix can intuitively display the relevant operating parameters, energy consumption, and task execution priorities of the equipment running in the smart factory, facilitating data analysis. For example, the test data of a robot sorting task in a smart factory can be represented by the following matrix:

[0095]

[0096] Furthermore, when constructing the multi-dimensional state matrix, the matrix needs to be constructed in combination with the values ​​and dimensions of the actual test data.

[0097] Furthermore, the embodiment of the present invention utilizes the multidimensional state matrix to identify the spatial conflict mode and temporal conflict mode of the operating equipment, and obtains collaborative conflict information, which can transform implicit conflicts into explicit space-time trajectory intersections or time window overlaps, realize quantitative assessment of conflict risks, and thus provide data support for equipment scheduling optimization.

[0098] Among them, the spatial conflict mode refers to the potential collision or path overlap risk mode of running equipment in spatial position, which is manifested as the intersection, overlap or close approach of the motion trajectories of different equipment in spatial coordinates. The temporal conflict mode refers to the timing conflict risk mode of running equipment in the time dimension, which is manifested as the time interval between different equipment arriving at the same spatial position is less than the safety time threshold, resulting in potential collision or operation interference.

[0099] As an embodiment of the present invention, using the multidimensional state matrix to identify the spatial conflict mode and the temporal conflict mode of the operating device to obtain the collaborative conflict information includes:

[0100] Extracting spatiotemporal data from the multidimensional state matrix to obtain a spatiotemporal attribute set of the device;

[0101] Using the device spatiotemporal attribute set, predicting the running time space trajectory of the running device to obtain a spatiotemporal trajectory set;

[0102] Performing a conflict grid mapping operation on the spatiotemporal trajectory set to obtain a spatiotemporal conflict heat map;

[0103] The spatial conflict pattern and the temporal conflict pattern of the operating equipment are analyzed by using the spatiotemporal conflict heat map to obtain collaborative conflict information.

[0104] Among them, the device spatiotemporal attribute set refers to a structured data set extracted from a multidimensional state matrix that characterizes the relationship between the spatial position and time of the device, including core attributes such as the spatial coordinates, motion parameters and timestamps of the device at different time points. The spatiotemporal trajectory set refers to a set of spatial motion paths that change over time based on the device spatiotemporal attribute set. The spatiotemporal conflict heat map refers to a conflict risk visualization map formed by probabilistic superposition after mapping the spatiotemporal trajectories of all devices to a three-dimensional grid space.

[0105] During the specific implementation process, spatially related data columns such as device location coordinates (x, y), timestamps, and velocity vectors can be extracted from the multidimensional state matrix. Then, the time information corresponding to the spatial correlation can be queried to construct a four-dimensional spatiotemporal attribute table containing the device ID, time point, position, and velocity. Based on the current position, velocity, and acceleration of the device, the Kalman filter algorithm can be used to predict the motion trajectory in the short term in the future, and the complete spatiotemporal trajectory can be generated by combining the device task path planning information (such as the preset navigation path of the AGV). The continuous spatiotemporal trajectory is discretized into a three-dimensional grid space (x, y, t), and each grid records the device occupancy probability. A spatiotemporal conflict heat map is generated by superimposing the grid probabilities of all devices. High-probability conflict areas (such as grid probability >80%) are identified in the heat map, and the DBSCAN clustering algorithm is used to distinguish between spatial static conflicts (such as path intersections) and temporal dynamic conflicts (such as the time difference between devices arriving at the same location is less than a safety threshold).

[0106] S3. Decoupling the operating equipment from collaborative conflicts based on the production task data and the collaborative conflict information to obtain a feasible coordination mode.

[0107] The embodiment of the present invention decouples the operating equipment from collaborative conflicts based on the production task data and the collaborative conflict information to obtain a feasible coordination mode, which can accurately resolve the conflicts of equipment in space and time to obtain a feasible coordination mode, thereby improving production efficiency, reducing costs, and enhancing production line flexibility and fault tolerance, which is a key link in realizing efficient collaborative production in smart factories.

[0108] The feasible coordination mode refers to a collaborative operation plan that is generated for the operating equipment without time and space conflicts and meets production constraints by analyzing and decoupling the production task data and collaborative conflict information.

[0109] As an embodiment of the present invention, based on the production task data and the collaborative conflict information, collaborative conflict decoupling is performed on the operating equipment to obtain a feasible coordination mode, including:

[0110] Using the production task data and the collaborative conflict information, constructing a conflict-task association table for the operating equipment;

[0111] Performing feasible screening on the conflict-task association table based on a rule engine to obtain a basic decoupling solution set;

[0112] Performing genetic algorithm optimization on the basic decoupling solution set to obtain a Pareto optimal solution set;

[0113] Performing feasibility verification on the Pareto optimal solution set to extract executable coordination instructions from the Pareto optimal solution set;

[0114] The executable coordination instruction is used to perform device control coding operations on the running device to obtain a feasible coordination mode.

[0115] Among them, the conflict-task association table refers to a two-dimensional table that structuredly integrates production task data and collaborative conflict information; the basic decoupling solution set refers to a set of basic solutions obtained after preliminary screening of the conflict-task association table through a rule engine; the Pareto optimal solution set refers to a non-dominated solution set obtained after multi-objective optimization of the basic decoupling solution set through a genetic algorithm; the executable coordination instruction refers to a control code or parameter set that is screened out from the Pareto optimal solution set and can directly drive the operation of the equipment.

[0116] During the specific implementation process, information such as task ID, priority and equipment requirements can be extracted from the production task data, and the conflict type, involved equipment and conflict time and space points can be obtained from the collaborative conflict information. The two are then associated according to the equipment ID, and the tasks and priorities corresponding to each conflict are recorded in a table to obtain a conflict-task association table; conflict decoupling rules can be set in advance (such as high-priority tasks first, spatial conflicts first bypass), and the conflict-task association table data can be matched in the rule engine through the rete algorithm, and then decoupling solutions that meet the basic rules can be screened out, such as adjusting the time window or equipment path of low-priority tasks to form a basic decoupling solution set; the basic decoupling solution is encoded as a "chromosome" in the genetic algorithm, with the goal of minimizing task delay, minimizing energy consumption, and eliminating conflicts. The multi-objective function with the highest solution rate is then iterated through selection, crossover, and mutation operations, and a genetic algorithm is implemented using the Scikit-opt library to ultimately obtain a Pareto optimal solution set that balances different objectives. For each Pareto solution, the equipment operation process is simulated to check whether there are new conflicts and whether the equipment's physical constraints (such as speed limits and turning radius) are met. Conflict-free solutions that meet the equipment's capabilities are screened out, and executable instructions such as the equipment path and start and stop times are extracted. The executable instructions are converted into control codes recognizable by the equipment (such as PLC control instructions and AGV navigation path coordinates). Communication with the equipment is carried out via the Modbus protocol, and the encoded instructions are sent to the equipment controller to achieve coordinated operation of the equipment and form a final feasible coordination model.

[0117] S4. Using the smart factory scene graph, perform obstacle avoidance reinforcement learning on the running equipment to construct a global obstacle avoidance path for the running equipment in the smart factory.

[0118] The embodiment of the present invention utilizes the smart factory scene graph to perform obstacle avoidance reinforcement learning on the operating equipment to construct a global obstacle avoidance path for the operating equipment in the smart factory. This allows the target equipment to autonomously learn the optimal obstacle avoidance strategy while considering its own physical constraints and environmental limitations, balance safety and task efficiency, and realize intelligent path planning in the smart factory.

[0119] The obstacle avoidance global path refers to a complete moving route from the starting position to the target position planned for the running equipment through obstacle avoidance reinforcement learning in the global environment of the smart factory.

[0120] As an embodiment of the present invention, using the smart factory scene graph, obstacle avoidance reinforcement learning is performed on the operating equipment to construct a global obstacle avoidance path for the operating equipment in the smart factory, including:

[0121] Performing real-time scene updates on the smart factory scene graph to obtain an updated map;

[0122] Identifying geometric features of an update target corresponding to the update map;

[0123] Performing semantic annotation on the geometric features to obtain semantic annotation data;

[0124] Using the semantic annotation data, marking obstacles on the updated map to form an obstacle-marked map;

[0125] Querying the static state and action space of the running device in the obstacle-annotated map;

[0126] Obstacle avoidance reinforcement learning is performed on the operating equipment using the static state and the action space to construct a global obstacle avoidance path for the operating equipment in the smart factory.

[0127] Among them, the updated map refers to the map obtained after the scene target of the smart factory scene map is updated, the semantic annotation data refers to the structured data that assigns semantic categories, attributes and logical relationships to geometric features, and the obstacle annotation map refers to the map in which obstacles are annotated in the updated map. For example, when the running equipment (forklift) fails after running for a period of time and stays in a certain path, or when the forklift is used to transport goods, the goods fall into the movement path of the running equipment, which is also regarded as an obstacle. The static state refers to the inherent properties and fixed parameters of the running equipment, including the current position, size specifications, etc. The action space refers to the effective range of the running equipment that can perform actions in the updated map.

[0128] During the specific implementation process, environmental data can be collected in real time through sensors such as lidar and cameras in the factory. The SLAM algorithm can be used to integrate the newly collected data into the original scene map, correct the positions of equipment and obstacles in the map, and generate an updated map; the edge detection algorithm (such as the Canny algorithm) and shape recognition algorithm in computer vision can be used to extract the contours and analyze the shapes of objects in the updated map, and obtain the size, shape and other geometric feature data of the updated targets (such as material carts, shelves, etc.); the semantic segmentation model in deep learning (such as U-Net, DeepLab) is used to match geometric features with preset object categories (such as "obstacles", "passable areas", and "target workstations"), assign semantic labels to each geometric feature, and output semantic annotation data; the static state information such as the current position and orientation of the running equipment is extracted from the obstacle annotation map, and then combined with the equipment's motion capability parameters (such as maximum speed and turning radius) to determine the feasible movement direction and distance of the equipment in the current environment and delineate the action space.

[0129] Preferably, the updating map is annotated with obstacles using the semantic annotation data, and the obstacle annotation map includes:

[0130] identifying dynamic objects in the updated map using the semantically annotated data;

[0131] querying the trajectory data of the dynamic object, and using the trajectory data to mark the dangerous area on the updated map to obtain a dynamic prohibited area;

[0132] Performing probability grid modeling on the dynamic forbidden area, a grid area;

[0133] The posterior probability of obstacle distribution in the grid area is calculated using the following formula:

[0134]

[0135] Among them, prob t (i, j) represents the posterior probability of obstacle distribution, t represents time, (i, j) represents a grid in the grid area, prob t-1 (i, j) represents the prior probability of (i, j) at time t-1, p(z|u) represents the detection likelihood, p(z|-u) represents the false detection probability, z represents the latest observation data, u represents the probability that the sensor detects occupation when the grid is indeed occupied, and -u represents the probability that the sensor falsely reports occupation when the grid is actually empty;

[0136] Determining an obstacle occurrence probability distribution in the grid area based on the posterior probability;

[0137] The obstacle occurrence probability distribution is dynamically and statically fused with the updated map to obtain an obstacle annotation map.

[0138] Among them, the dynamic prohibited area refers to the impassable area temporarily demarcated in the global topological map due to the presence of dynamic obstacles (such as mobile robots, pedestrians) or trajectory prediction. The grid area refers to a set of regular grid units divided into a continuous space by a discretization method. The posterior probability refers to the updated estimate of the probability of obstacle existence in the grid area after combining prior knowledge with the latest observation data. The obstacle occurrence probability distribution refers to the spatial distribution of the obstacle posterior probability values ​​of all grids in the global scene map.

[0139] During the specific implementation process, the pixel-level labels output by the semantic segmentation model (such as DeepLab) can be combined with the optical flow method or background difference method to detect pixel motion, and objects labeled as "AGV", "robot" and other categories with pixel displacement can be identified as dynamic objects; the historical coordinates of the dynamic object in the past 5-10 seconds (such as the (x, y, t) sequence of AGV) are obtained from the factory positioning system, and the Kalman filter is used to predict the trajectory of the next 10 seconds. The 1-meter-wide dangerous area is expanded with the trajectory as the center to realize trajectory prediction and dynamic prohibited area generation; the grid area can discretize the dynamic prohibited area into a grid with a resolution of 0.1m.

[0140] It should be further explained that the posterior probability calculation formula is based on Bayes' theorem. By integrating "historical experience" (prior probability) and "latest observations" (sensor detection data), the probability of the existence of obstacles in the grid is dynamically updated. This can reflect the location and risk level of dynamic obstacles in real time, avoid false detection or missed detection, provide a quantitative environmental risk distribution for multiple devices, and support path planning and collaborative decision-making.

[0141] Furthermore, as another optional embodiment of the present invention, using the static state and the action space to perform obstacle avoidance reinforcement learning on the running device to construct a global obstacle avoidance path for the running device includes:

[0142] Performing state space definition on the static state and the action space to obtain a state vector;

[0143] Performing action discretization processing on the state vector to obtain a discrete action set;

[0144] configuring a reward function for the discrete action set;

[0145] constructing an initial obstacle avoidance path for the operating device using the discrete action set;

[0146] Based on the reward function, reinforcement learning is performed on the initial obstacle avoidance path to obtain an adaptive obstacle avoidance strategy;

[0147] Based on the adaptive obstacle avoidance strategy, a global obstacle avoidance path for the operating equipment in the smart factory is constructed.

[0148] Among them, the state vector refers to a multidimensional vector composed of the current environmental information of the target device and its own state parameters; the discrete action set refers to a finite action set obtained by discretizing the continuous action space of the target device (such as steering angle and speed); the reward function refers to a scalar function used to evaluate the pros and cons of the state transition after the target device performs an action, and guides the reinforcement learning model to learn the obstacle avoidance strategy; the adaptive obstacle avoidance strategy refers to the mapping relationship obtained through reinforcement learning training, that is, the probability distribution of selecting action a when a given state vector s is given.

[0149] Optionally, the state vector can fuse the device size, current position, obstacle distance (such as the 8-direction distance collected by the lidar) and grid probability to construct a multidimensional state vector containing parameters such as position (x, y), orientation, speed V, etc. The discrete action set can discretize the continuous action space (such as steering angle ±30°, speed 0-2m / s) into 9 actions (such as turning left 15°+deceleration, keeping direction+acceleration, etc.), ensuring that the action meets the maximum steering angle and acceleration limits of the device. The reward function can design positive rewards (such as +10 points for approaching the target) and negative penalties. Penalty (such as collision risk -20 points, entering a high-risk area -15 points), and the relationship between distance and reward can be smoothed in combination with a Gaussian function. The initial obstacle avoidance path can be obtained by planning a collision-free initial path in the global map through the A algorithm. The adaptive obstacle avoidance strategy can use the PPO algorithm to train the strategy network, with the state vector as input and the discrete action set as output, and is obtained by interacting with the environment (such as dynamic obstacle movement in a simulator). The global obstacle avoidance path can be obtained by applying the trained strategy network to a real-time environment, selecting actions according to the current state, and dynamically adjusting the sub-target path.

[0150] S5. Based on the feasible coordination mode and the obstacle avoidance global path, a target collaborative control scheme for the running equipment is constructed, and based on the target collaborative control scheme, the running equipment is subjected to device collaborative control.

[0151] The embodiment of the present invention can realize the integrated dynamic scheduling of "task-path" by constructing the target collaborative control scheme of the operating equipment based on the feasible coordination mode and the obstacle avoidance global path, thereby optimizing the resource allocation of operating equipment in the smart factory and avoiding production line stagnation caused by single machine failure or path congestion.

[0152] Among them, the target collaborative control scheme refers to a comprehensive control strategy constructed by integrating feasible coordination modes (such as equipment collaboration logic, task allocation rules) and global obstacle avoidance paths (global collision-free movement routes of equipment) in the scenario of multi-equipment collaborative tasks in smart factories, with the goal of achieving safe and efficient operation of equipment.

[0153] As an embodiment of the present invention, based on the feasible coordination mode and the obstacle avoidance global path, a target collaborative control scheme for the operating device is constructed, including:

[0154] Performing multi-source data fusion on the feasible coordination mode and the obstacle avoidance global path to obtain fused data;

[0155] Using the fused data, constructing task-path coordination instructions for the operating device;

[0156] Performing protocol conversion on the task-path coordination instruction to obtain a device control packet;

[0157] Using the device control package, performing coordinated control simulation on the running device;

[0158] A target coordinated control scheme for the operating equipment is constructed based on a simulation set of the coordinated control simulation.

[0159] Among them, the task-path coordination instruction refers to the device execution-level instruction set generated after coupling the task logic in the feasible coordination mode with the spatial constraints of the obstacle avoidance global path, and the device control package refers to the instruction data packet encapsulated after converting the task-path coordination instruction into the underlying control protocol recognizable by the device hardware.

[0160] During the specific implementation process, data such as task allocation and timing priority are extracted from the feasible coordination mode, and information such as trajectory coordinates and safety distance are obtained from the global obstacle avoidance path. Then, Python's Pandas library is used to clean and unify the data format, and then integrated into structured fusion data through data association operations. Based on the task priority and obstacle avoidance path constraints in the fusion data, a heuristic algorithm (such as the A* algorithm) is used to generate the order in which the equipment executes the tasks and the corresponding path instructions. The instructions are structured and arranged in the format of "task ID-starting point-passing point-destination point-execution time" to form a task-path collaborative instruction set. For different types of equipment (such as AGV, robotic arm), industrial communication protocol conversion tools (such as Mo dbus gateway), converts task-path collaborative instructions into control protocol data recognizable by the device, and encapsulates the converted data into a device control package containing instruction parameters, checksums and other information; builds a virtual scene of a smart factory in the simulation platform, imports the 3D model of the equipment and the environmental map, and then inputs the device control package into the simulation system to drive the virtual equipment to perform collaborative tasks, monitor the equipment operating status and potential conflicts in real time, and realize the coordinated control simulation of the running equipment; analyzes the equipment's task completion status, number of conflicts, energy consumption and other indicators during the simulation process to evaluate the effectiveness of the control package, and then optimizes and adjusts the equipment control package based on the simulation results, ultimately forming a target collaborative control solution that includes task scheduling, path planning, and conflict resolution strategies.

[0161] Furthermore, the embodiment of the present invention can perform equipment collaborative control on the running equipment based on the target collaborative control scheme, and can use the preferred control scheme obtained after step-by-step analysis of the relevant information of the real-time operation scenario of the smart factory to perform intelligent control on the equipment running in the smart factory, so as to improve the reliability of collaborative control of multiple devices in the smart factory scenario.

[0162] In the specific process, the target collaborative control plan can be converted into operating parameters that can be executed by the equipment control system of the smart factory, and then the control system drives the equipment in combination with the operating parameters.

[0163] Example 2:

[0164] like Figure 2 , which is a functional module diagram of a multi-device collaborative control system under the autonomous collaborative algorithm of the present invention.

[0165] The multi-device collaborative control system 200 using the autonomous collaborative algorithm described in the present invention can be installed in an electronic device. Depending on the functionality implemented, the multi-device collaborative control system using the autonomous collaborative algorithm can include an information acquisition module 201, a conflict information analysis module 202, a conflict decoupling module 203, an obstacle avoidance path planning module 204, and a collaborative control module 205. The modules described in the present invention, also referred to as units, refer to a series of computer program segments that can be executed by an electronic device processor and can perform fixed functions, and are stored in the memory of the electronic device.

[0166] In the embodiment of the present invention, the functions of each module / unit are as follows:

[0167] The information collection module 201 is used to collect physical attribute data, real-time operation data, production task data, and smart factory scene diagrams of devices running in multi-device collaborative tasks in smart factory applications, wherein the physical attribute data includes device type, physical size, and operating parameter limitations; the real-time operation data includes location, speed, load, and energy consumption; and the production task data includes task type, priority, and delivery time;

[0168] The conflict information analysis module 202 is configured to construct a multidimensional state matrix of the operating devices based on the physical attribute data and the real-time operating data, and identify spatial and temporal conflict patterns of the operating devices using the multidimensional state matrix to obtain collaborative conflict information;

[0169] The conflict decoupling module 203 is configured to perform collaborative conflict decoupling on the operating equipment according to the production task data and the collaborative conflict information to obtain a feasible coordination mode;

[0170] The obstacle avoidance path planning module 204 is configured to utilize the smart factory scene graph to perform obstacle avoidance reinforcement learning on the operating equipment to construct a global obstacle avoidance path for the operating equipment in the smart factory;

[0171] The collaborative control module 205 is used to construct a target collaborative control scheme for the running equipment based on the feasible coordination mode and the global obstacle avoidance path, and perform device collaborative control on the running equipment based on the target collaborative control scheme.

[0172] In detail, each module in the multi-device collaborative control system 200 under the autonomous collaborative algorithm in the embodiment of the present invention adopts the same Figure 1 The multi-device collaborative control method under the autonomous collaborative algorithm described in the present invention has the same technical means and can produce the same technical effects, so it will not be repeated here.

[0173] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multi-device collaborative control method and system based on an autonomous collaborative algorithm, characterized in that: The method comprises: Collect physical attribute data, real-time operation data, production task data, and smart factory scenario diagrams of devices running in multi-device collaborative tasks in smart factory applications. The physical attribute data includes device type, physical dimensions, and operating parameter limitations; the real-time operation data includes location, speed, load, and energy consumption; and the production task data includes task type, priority, and delivery time. constructing a multidimensional state matrix of the operating equipment based on the physical attribute data and the real-time operating data, and using the multidimensional state matrix to identify spatial conflict patterns and temporal conflict patterns of the operating equipment to obtain collaborative conflict information; Decoupling the operating equipment from the collaborative conflicts according to the production task data and the collaborative conflict information to obtain a feasible coordination mode; Using the smart factory scene graph, performing obstacle avoidance reinforcement learning on the operating equipment to construct a global obstacle avoidance path for the operating equipment in the smart factory; Based on the feasible coordination mode and the obstacle avoidance global path, a target collaborative control scheme for the running equipment is constructed, and based on the target collaborative control scheme, the running equipment is subjected to device collaborative control.

2. The multi-device collaborative control method under the autonomous collaborative algorithm according to claim 1, characterized in that: Constructing a multidimensional state matrix of the operating device based on the physical property data and the real-time operating data, including: performing data standardization processing on the physical attribute data and the real-time operation data to obtain a standardized data set; Performing multidimensional feature screening on the standardized data set to obtain multidimensional key features; Performing time series alignment processing on the features in the multidimensional key features to obtain multidimensional synchronous features; The multi-dimensional synchronization feature is used to construct a multi-dimensional state matrix of the operating device.

3. The multi-device collaborative control method under the autonomous collaborative algorithm according to claim 2, characterized in that: Utilizing the multidimensional synchronization feature, constructing a multidimensional state matrix of the operating device, including: Arranging the multi-dimensional synchronization features in time series to obtain sequence features; Based on the sequence characteristics, a multidimensional state matrix of the operating device is constructed, wherein the multidimensional state matrix can be expressed in the following form: Among them, M represents the multidimensional state matrix, ID1 represents the ID of the first running device in the sequence feature, ID n Indicates the id of the nth running device in the sequence feature, x1 indicates the x-axis coordinate of the first running device in the sequence feature, x n Indicates the x-axis coordinate of the nth running device in the sequence feature, y1 indicates the y-axis coordinate of the first running device in the sequence feature, x n Indicates the y-axis coordinate of the nth running device in the sequence feature, v x1 represents the velocity corresponding to x1, v xn Indicates that x n The corresponding speed, v y1 represents the velocity corresponding to y1, v yn Indicates that y n Corresponding speed, E1 represents the load rate of the first running equipment in the sequence feature, E n represents the load rate of the nth running device in the sequence feature, α1 represents the energy consumption of the first running device in the sequence feature, α n represents the energy consumption of the nth running device in the sequence feature, β1 represents the priority of the first running device in the sequence feature, β n Indicates the priority of the nth running device in the sequence feature.

4. The multi-device collaborative control method under the autonomous collaborative algorithm according to claim 1, characterized in that: Using the multidimensional state matrix, identifying the spatial conflict mode and the temporal conflict mode of the operating device to obtain collaborative conflict information includes: Extracting spatiotemporal data from the multidimensional state matrix to obtain a spatiotemporal attribute set of the device; Using the device spatiotemporal attribute set, predicting the running time space trajectory of the running device to obtain a spatiotemporal trajectory set; Performing a conflict grid mapping operation on the spatiotemporal trajectory set to obtain a spatiotemporal conflict heat map; The spatial conflict pattern and the temporal conflict pattern of the operating equipment are analyzed by using the spatiotemporal conflict heat map to obtain collaborative conflict information.

5. The multi-device collaborative control method under the autonomous collaborative algorithm according to claim 1, characterized in that: Decoupling the operating equipment from the collaborative conflicts based on the production task data and the collaborative conflict information to obtain a feasible coordination mode includes: Using the production task data and the collaborative conflict information, constructing a conflict-task association table for the operating equipment; Performing feasible screening on the conflict-task association table based on a rule engine to obtain a basic decoupling solution set; Performing genetic algorithm optimization on the basic decoupling solution set to obtain a Pareto optimal solution set; Performing feasibility verification on the Pareto optimal solution set to extract executable coordination instructions from the Pareto optimal solution set; The executable coordination instruction is used to perform device control coding operations on the running device to obtain a feasible coordination mode.

6. The multi-device collaborative control method under the autonomous collaborative algorithm according to claim 1, characterized in that: Using the smart factory scene graph, performing obstacle avoidance reinforcement learning on the operating equipment to construct a global obstacle avoidance path for the operating equipment in the smart factory, including: Performing real-time scene updates on the smart factory scene graph to obtain an updated map; Identifying geometric features of an update target corresponding to the update map; Performing semantic annotation on the geometric features to obtain semantic annotation data; Using the semantic annotation data, marking obstacles on the updated map to form an obstacle-marked map; Querying the static state and action space of the running device in the obstacle-annotated map; Obstacle avoidance reinforcement learning is performed on the operating equipment using the static state and the action space to construct a global obstacle avoidance path for the operating equipment in the smart factory.

7. The multi-device collaborative control method under the autonomous collaborative algorithm according to claim 6, characterized in that: The step of using the semantic annotation data to mark obstacles on the updated map, wherein the obstacle-marked map includes: identifying dynamic objects in the updated map using the semantically annotated data; querying the trajectory data of the dynamic object, and using the trajectory data to mark the dangerous area on the updated map to obtain a dynamic prohibited area; Performing probability grid modeling on the dynamic forbidden area, a grid area; The posterior probability of obstacle distribution in the grid area is calculated using the following formula: Among them, prob t (i, j) represents the posterior probability of obstacle distribution, t represents time, (i, j) represents a grid in the grid area, prob t-1 (i, j) represents the prior probability of (i, j) at time t-1, p(z|u) represents the detection likelihood, p(z|-u) represents the false detection probability, z represents the latest observation data, u represents the probability that the sensor detects occupation when the grid is indeed occupied, and -u represents the probability that the sensor falsely reports occupation when the grid is actually empty; Determining an obstacle occurrence probability distribution in the grid area based on the posterior probability; The obstacle occurrence probability distribution is dynamically and statically fused with the updated map to obtain an obstacle annotation map.

8. The multi-device collaborative control method under the autonomous collaborative algorithm according to claim 6, characterized in that: Performing obstacle avoidance reinforcement learning on the running device using the static state and the action space to construct a global obstacle avoidance path for the running device, including: Performing state space definition on the static state and the action space to obtain a state vector; Performing action discretization processing on the state vector to obtain a discrete action set; configuring a reward function for the discrete action set; constructing an initial obstacle avoidance path for the operating device using the discrete action set; Based on the reward function, reinforcement learning is performed on the initial obstacle avoidance path to obtain an adaptive obstacle avoidance strategy; Based on the adaptive obstacle avoidance strategy, a global obstacle avoidance path for the operating equipment in the smart factory is constructed.

9. The multi-device collaborative control method under the autonomous collaborative algorithm according to claim 1, characterized in that: Based on the feasible coordination mode and the obstacle avoidance global path, a target collaborative control scheme for the operating device is constructed, including: Performing multi-source data fusion on the feasible coordination mode and the obstacle avoidance global path to obtain fused data; Using the fused data, constructing task-path coordination instructions for the operating device; Performing protocol conversion on the task-path coordination instruction to obtain a device control packet; Using the device control package, performing coordinated control simulation on the running device; A target coordinated control scheme for the operating equipment is constructed based on a simulation set of the coordinated control simulation.

10. A multi-device collaborative control system based on an autonomous collaborative algorithm, characterized in that: The system comprises: An information collection module, configured to collect physical attribute data, real-time operation data, production task data, and smart factory scenario diagrams of devices running in multi-device collaborative tasks in smart factory applications. The physical attribute data includes device type, physical dimensions, and operating parameter limits; the real-time operation data includes location, speed, load, and energy consumption; and the production task data includes task type, priority, and delivery time. a conflict information analysis module, configured to construct a multidimensional state matrix of the operating equipment based on the physical property data and the real-time operating data, and identify spatial conflict patterns and temporal conflict patterns of the operating equipment using the multidimensional state matrix to obtain collaborative conflict information; a conflict decoupling module, configured to perform collaborative conflict decoupling on the operating equipment according to the production task data and the collaborative conflict information, and obtain a feasible coordination mode; An obstacle avoidance path planning module, configured to utilize the smart factory scene graph to perform obstacle avoidance reinforcement learning on the operating equipment, so as to construct a global obstacle avoidance path for the operating equipment in the smart factory; A collaborative control module is used to construct a target collaborative control scheme for the running equipment based on the feasible coordination mode and the obstacle avoidance global path, and to perform device collaborative control on the running equipment based on the target collaborative control scheme.

Citation Information

Cited By

  • Intelligent stacking method and device

    CN120931205A

  • Intelligent stacking method and device

    CN120931205B

  • SCADA (supervisory control and data acquisition)-based multi-specification flexible subpackage control method and system for medicines

    CN121069937A

  • Synchronous control method and system for electrification of lifting system for maintenance

    CN121763822A

  • Method and system for the synchronized control of the motorization of a lifting system for maintenance

    CN121763822B