Intelligent control method and system for robot cluster response

Abstract

The present invention relates to the field of robot control technology, and discloses an intelligent control method and system for robot cluster response, the method comprising: collecting historical response data of the robot cluster, extracting cluster state features to construct an initial candidate set; analyzing the functional matching between historical tasks and clusters in the initial candidate set, and screening out a second candidate set; analyzing the response behavior pattern of the cluster; analyzing the response adaptability of the cluster and the cluster in the second candidate set, and screening out a third candidate set; obtaining the corresponding cluster information in the third candidate set based on the cluster's real-time response feedback records, and optimizing control instructions. The system comprises a data acquisition and feature extraction module, a functional matching screening module, a behavior pattern analysis module, a response adaptation screening module, and a real-time optimization control module. The invention can improve the accuracy, efficiency, and flexibility of the robot cluster response, and enhance overall performance.

Description

Technical Field

[0001] The present invention relates to the field of robot control technology, and in particular to an intelligent control method and system for robot cluster response. Background Art

[0002] With the rapid development of science and technology, robotics has made significant progress, and robot swarms are increasingly being used in a wide range of fields, such as industrial production, logistics and transportation, environmental monitoring, and disaster relief. However, as the application scenarios of robot swarms become increasingly complex and diverse, the requirements for their response control are also becoming increasingly stringent. Existing robot swarm response control technologies face numerous challenges.

[0003] From a data utilization perspective, previous control methods often lack comprehensiveness and depth when processing historical response data from robot swarms. Most simply extract partial data features, failing to fully explore the underlying cluster state characteristics, such as spatial distribution and task load. This results in inaccurate initial candidate set construction, failing to fully reflect the actual state of the swarm, which in turn affects subsequent screening and control decisions.

[0004] When it comes to matching tasks with cluster functions, traditional methods lack a detailed analysis of historical task characteristics and cluster functional characteristics. Without in-depth analysis and matching calculation of the identification and functional characteristics of historical tasks and cluster functional characteristics, this method fails to accurately select clusters with a high degree of matching with task functions, reducing the efficiency and effectiveness of task execution.

[0005] Existing technologies have significant shortcomings in analyzing cluster response behavior patterns. They fail to fully consider factors such as response duration, execution accuracy, and response intervals. This makes it difficult to accurately determine cluster response behavior patterns, such as execution speed, and thus to obtain cluster response preference scores. This lacks a reliable basis for subsequent response suitability analysis.

[0006] Furthermore, existing technologies fail to effectively utilize the cluster's real-time response feedback when optimizing control during the response process. Unable to accurately determine whether the cluster has a collaborative habit, they are unable to optimize control instructions based on different situations. This results in the robot cluster being unable to adjust in a timely manner during the response process, making it difficult to adapt to complex and changing environments.

[0007] In addition, with the continuous expansion of the scale of robot clusters and the continuous expansion of application scenarios, the requirements for intelligent and automated control methods are becoming increasingly higher. The existing control methods are difficult to meet the needs in terms of intelligence and automation. There is an urgent need for a more intelligent and efficient control method and system to improve the responsiveness and overall performance of robot clusters. Summary of the Invention

[0008] The object of the present invention is to provide an intelligent control method and system for robot cluster response to solve the problems raised in the above background technology.

[0009] To achieve the above objectives, the present invention provides the following technical solution: an intelligent control method for robot cluster response, the method comprising:

[0010] S1. Collect historical response data of the robot cluster, extract cluster state features, and build an initial candidate set of the robot cluster;

[0011] S2. Analyze the historical tasks based on the historical response data of the robot cluster, and then determine the degree of matching between the historical tasks and the functions of each cluster in the initial candidate set based on the analysis results, and select a second candidate set of robot clusters;

[0012] S3. Analyze the cluster’s response behavior pattern based on the robot cluster’s historical response data;

[0013] S4. Analyze the clusters according to the response behavior patterns of the clusters, determine the response adaptability of the clusters in the second candidate set based on the results of the cluster analysis, and screen out a third candidate set of robot clusters;

[0014] S5. Obtain corresponding cluster information in the third candidate set according to the cluster real-time response feedback record, and optimize the control instructions during the response process.

[0015] Preferably, the S1 includes the following specific steps:

[0016] S101, a data acquisition module collects historical response data of a robot cluster and extracts cluster state features from the historical response data, wherein the cluster state features include spatial distribution features and task load features;

[0017] S102: Filter out clusters that meet the spatial distribution conditions in the cluster management module according to the spatial distribution characteristics and output them in the form of an initial candidate set.

[0018] Preferably, S2 includes the following specific steps:

[0019] S201. Obtain historical response data of a robot cluster, analyze features of historical tasks based on the historical response data, wherein the features of the historical tasks include task identification features and functional features, select historical tasks that meet task load conditions based on the task identification features and the task load features, and simultaneously obtain functional features of clusters in the initial candidate set;

[0020] S202: Analyze the functional matching degree between the filtered historical tasks and the clusters in the initial candidate set based on the functional characteristics of the filtered historical tasks and the functional characteristics of the clusters in the initial candidate set;

[0021] S203 : Sort the function matching degrees in descending order, filter out the clusters corresponding to the set sequence numbers, and output them as a second candidate set.

[0022] Preferably, S3 includes the following specific steps:

[0023] S301. Extract the response duration and execution accuracy from the filtered historical response data, obtain the response time interval between two adjacent historical response data, average the obtained response time intervals to obtain an average response interval, and average the obtained execution accuracies to obtain an average execution accuracy.

[0024] S302, comparing the average response interval with a preset first interval threshold and a second interval threshold and determining the response behavior pattern of the cluster; if the average response interval is less than or equal to the first interval threshold, determining that the cluster response task execution speed is fast; if the average response interval is greater than the first interval threshold and the response interval is less than the second interval threshold, determining that the cluster response task execution speed is medium; if the average response interval is greater than or equal to the second interval threshold, determining that the cluster response task execution speed is slow;

[0025] S303: Obtain a cluster response preference score value according to the cluster response behavior pattern.

[0026] Preferably, the S4 includes the following specific steps:

[0027] S401, obtaining the response characteristics and task difficulty of the clusters in the second candidate set, and analyzing the response compatibility of the cluster with the clusters in the second candidate set based on the cluster response preference score, average execution accuracy, the response characteristics and task difficulty of the clusters in the second candidate set;

[0028] S402 : Sort the response adaptability in descending order, filter out the clusters corresponding to the set sequence number, and output them as a third candidate set.

[0029] Preferably, the S5 includes the following specific steps:

[0030] Obtaining real-time cluster response feedback records to analyze whether the cluster has a collaborative habit. If the cluster does not have a collaborative habit, obtaining the cluster information with the highest response match in the third candidate set and optimizing control instructions during the response process. If the cluster does have a collaborative habit, obtaining the cluster information corresponding to the set sequence number in the third candidate set and optimizing control instructions during the response process.

[0031] The method for analyzing whether the cluster has a collaborative habit is to count the proportion of the number of times multiple tasks are executed simultaneously in the cluster's real-time response feedback records to the total number of responses. If the proportion exceeds a preset collaborative habit threshold, it is determined that the cluster has a collaborative habit.

[0032] Preferably, the spatial distribution characteristics include specific distribution characteristics of coordinate deviation, coverage, and communication quality, and the task load characteristics include load characteristics of task processing volume, energy consumption, and device temperature.

[0033] Preferably, the response adaptability analysis of the cluster and the clusters in the second candidate set includes the matching degree between the cluster response preference score value and the cluster response characteristics, and the matching degree between the average execution accuracy and the task difficulty.

[0034] Preferably, the functional matching degree is analyzed by performing a correspondence analysis on the functional features of the filtered historical tasks and the functional features of the clusters in the initial candidate set, and calculating the ratio of the number of matching functional items to the total number of functional items.

[0035] Preferably, the present invention further includes an intelligent control system for robot cluster response, the system comprising:

[0036] The data collection and feature extraction module is used to collect historical response data of the robot cluster, extract cluster state features, and build an initial candidate set of the robot cluster;

[0037] a function matching screening module, connected to the data acquisition and feature extraction module, for analyzing the function matching between historical tasks and clusters in the initial candidate set based on the historical response data of the robot cluster, and screening out a second candidate set of robot clusters;

[0038] A behavior pattern analysis module, connected to the data acquisition and feature extraction module, for analyzing the response behavior pattern of the cluster based on the historical response data of the robot cluster;

[0039] a response adaptation screening module, connected to the behavior pattern analysis module and the function matching screening module, configured to analyze the response adaptability of the cluster with the clusters in the second candidate set according to the response behavior pattern of the cluster, and screen out a third candidate set of robot clusters;

[0040] The real-time optimization control module is connected to the response adaptation screening module and is used to obtain corresponding cluster information in the third candidate set according to the cluster real-time response feedback record, and optimize the control instructions during the response process.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] The intelligent control method and system for robot cluster response provided by this invention offer significant advantages in many aspects. During data acquisition and processing, by comprehensively collecting historical robot cluster response data and extracting cluster status characteristics, including spatial distribution characteristics (such as coordinate deviation, coverage, and specific distribution characteristics of communication quality) and task load characteristics (such as task processing volume, energy consumption, and device temperature load characteristics), the system can accurately construct an initial candidate set, laying a solid foundation for subsequent screening and ensuring that the initial candidate set fully and accurately reflects the actual state of the cluster.

[0043] During the function matching screening process, the characteristics of historical tasks (including task identification characteristics and function characteristics) are deeply analyzed, and the corresponding analysis is carried out item by item with the function characteristics of the clusters in the initial candidate set. The function matching degree is determined by counting the proportion of the number of matching function items to the total number of function items. The function matching degree is then sorted in descending order, and the clusters within the set sequence number are screened out to form the second candidate set. In this way, clusters with high matching degree with historical task functions can be accurately screened out, which greatly improves the accuracy of task and cluster function matching and provides a strong guarantee for the efficient execution of tasks.

[0044] For the analysis of cluster response behavior patterns, the response duration and execution accuracy in the filtered historical response data are extracted, the response time interval between two adjacent historical response data is obtained, and the average response interval and average execution accuracy are calculated. The average response interval is compared with the preset first interval threshold and second interval threshold to accurately judge the cluster's response behavior pattern (fast, medium or slow execution speed), and then obtain the cluster response preference score value, providing a reliable basis for subsequent response adaptability analysis, making the understanding of cluster response behavior more in-depth and accurate.

[0045] In the response adaptation screening stage, comprehensive consideration is given to factors such as the cluster response preference score, average execution accuracy, response characteristics of the clusters in the second candidate set, and task difficulty. An in-depth analysis is conducted on the response adaptability of the cluster with the clusters in the second candidate set, including the degree of matching between the cluster response preference score and the cluster response characteristics, as well as the degree of fit between the average execution accuracy and the task difficulty. The response adaptability is then sorted in descending order, and clusters within the set sequence number are screened out to form a third candidate set. This can accurately screen out clusters with high cluster response adaptability, further improving the accuracy and efficiency of the robot cluster response.

[0046] In terms of real-time optimized control, the system obtains the cluster's real-time response feedback records and determines whether the cluster has a collaborative habit by counting the proportion of times multiple tasks are executed simultaneously in the cluster's total response times. If a collaborative habit does not exist, the system obtains the cluster information optimization control instructions with the highest response matching in the third candidate set. If a collaborative habit exists, the system obtains the cluster information optimization control instructions within a set sequence number in the third candidate set. This allows for real-time optimization of control instructions based on the cluster's actual situation during the response process, enabling the robot cluster to better adapt to complex and changing environments and improving its response flexibility and adaptability.

[0047] From the perspective of the overall system, the system includes data acquisition and feature extraction module, function matching and screening module, behavior pattern analysis module, response adaptation and screening module and real-time optimization control module. The modules cooperate with each other and work closely together to form a complete intelligent control system, which realizes intelligent and automated control of the response of robot clusters, greatly improves the response capability and overall performance of robot clusters, and has broad application prospects and important practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a working principle diagram of the intelligent control method for robot cluster response according to the present invention;

[0049] Figure 2 Design diagram for analyzing the response behavior patterns of robot swarms;

[0050] Figure 3 Design diagrams for real-time optimal control of robot swarms;

[0051] Figure 4 Design diagram for the robot swarm intelligent control system. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0053] See also Figure 1-Figure 4 The intelligent control method and system for robot cluster response according to the present invention are specifically implemented as follows:

[0054] S1. Collect historical response data from the robot cluster, extract cluster state features, and construct an initial candidate set of robot clusters: The data acquisition module collects historical response data from the robot cluster and extracts cluster state features from this historical response data. These cluster state features include spatial distribution features and task load features. Spatial distribution features include coordinate deviation, coverage, and communication quality, while task load features include task processing volume, energy consumption, and device temperature. Based on these spatial distribution features, the cluster management module selects clusters that meet the spatial distribution criteria and outputs them as an initial candidate set.

[0055] S2. Analyze the functional matching between historical tasks and clusters in the initial candidate set based on the historical response data of the robot cluster, and select the second candidate set of the robot cluster: obtain the historical response data of the robot cluster, and analyze the characteristics of the historical tasks based on the data. The characteristics of the historical tasks include task identification characteristics and functional characteristics. Filter out historical tasks that meet the task load conditions based on the task identification characteristics and task load characteristics, and obtain the functional characteristics of the clusters in the initial candidate set at the same time. Analyze the functional characteristics of the filtered historical tasks and the functional characteristics of the clusters in the initial candidate set item by item, and calculate the ratio of the number of matching functional items to the total number of functional items to analyze the functional matching degree. Sort the functional matching degree in descending order, filter out the clusters corresponding to the set sequence number, and output them in the form of the second candidate set.

[0056] S3. Analyze the response behavior pattern of the cluster based on the historical response data of the robot cluster: extract the response duration and execution accuracy from the filtered historical response data, obtain the response time interval of two adjacent historical response data, average the obtained response time intervals to obtain the average response interval, and average the execution accuracy to obtain the average execution accuracy. Compare the average response interval with the preset first interval threshold and second interval threshold to determine the response behavior pattern of the cluster: if the average response interval is less than or equal to the first interval threshold, the cluster response task execution speed is determined to be fast; if the average response interval is greater than the first interval threshold and less than the second interval threshold, the execution speed is determined to be medium; if the average response interval is greater than or equal to the second interval threshold, the execution speed is determined to be slow. Obtain the cluster response preference score value based on the cluster's response behavior pattern.

[0057] S4. Analyze the response compatibility of the cluster with the clusters in the second candidate set based on the cluster's response behavior pattern, and select a third candidate set of robot clusters: Obtain the response characteristics and task difficulty of the clusters in the second candidate set, and combine the cluster response preference score and average execution accuracy to analyze the response compatibility of the cluster with the clusters in the second candidate set. This response compatibility analysis includes the degree of match between the cluster response preference score and the cluster response characteristics, as well as the degree of fit between the average execution accuracy and the task difficulty. Sort the response compatibility in descending order, select the clusters corresponding to the set sequence number, and output them as the third candidate set.

[0058] S5. Obtain the corresponding cluster information in the third candidate set based on the cluster's real-time response feedback records, and optimize control instructions during the response process: Obtain the cluster's real-time response feedback records, calculate the ratio of the number of times multiple tasks are executed simultaneously in the records to the total number of responses, and if this ratio exceeds a preset collaborative habit threshold, determine that the cluster has a collaborative habit; otherwise, determine that it does not. If the cluster does not have a collaborative habit, obtain the cluster information with the highest response match in the third candidate set, and optimize control instructions during the response process. If a collaborative habit does exist, obtain the cluster information corresponding to the set sequence number in the third candidate set, and optimize control instructions during the response process.

[0059] Example 1:

[0060] The implementation of step S1 involves the collection of historical response data of the robot cluster, the extraction of cluster state features, and the construction of an initial candidate set.

[0061] The data acquisition module collects historical response data generated by the robot cluster in different mission scenarios. This data comes from real-time feedback from the robots as they perform various tasks, including but not limited to the robots' dynamic position changes in space, communication data with other robots or the control center, and parameters related to the robots' own operating status.

[0062] When collecting historical response data, it is necessary to ensure the integrity and accuracy of the data. For example, for data related to spatial distribution characteristics, it is necessary to accurately record the coordinate information of each robot at different time points, including the specific values ​​of the X, Y, and Z axes, as well as the deviation between the coordinates and the preset coordinates. At the same time, it is also necessary to collect the coverage data of the entire cluster in space. This can be determined by calculating the boundaries of the spatial geometric figures formed by all robot coordinates, such as the area of ​​a polygon formed with all robot coordinates as vertices on a two-dimensional plane, or the volume in three-dimensional space. In addition, data related to communication quality is also crucial. Indicators such as the success rate of signal transmission, transmission delay time, and signal strength attenuation need to be collected to fully reflect the communication status between robots.

[0063] When collecting data related to task load characteristics, attention should be paid to aspects such as the robot's task throughput, energy consumption, and device temperature during task execution. Task throughput can include parameters such as the number of tasks each robot handles per unit time and the complexity of the tasks. Energy consumption data includes the robot's real-time battery charge, the rate of power consumption per unit time, and the total accumulated energy consumption. Device temperature collection requires recording temperature changes in key robot components, such as motor and processor temperatures, including maximum, minimum, and average values, as well as temperature trends.

[0064] When extracting cluster status features, the spatial distribution characteristics require analysis of the collected coordinate data and calculation of each robot's coordinate deviation—the difference between the actual coordinates and the preset coordinates. For coverage, the spatial coverage of the cluster is determined using geometric calculations based on the coordinate data of all robots. For example, on a two-dimensional plane, a convex hull algorithm can be used to calculate the minimum convex polygon containing all the robot coordinate points. The area of ​​this polygon is the cluster's coverage. Extracting communication quality features requires statistical analysis of the collected signal transmission data, such as calculating the average signal transmission success rate and the median transmission delay time, to characterize communication quality.

[0065] To extract task load characteristics, the task processing capacity characteristics can be normalized by processing the collected task processing data to obtain the task processing capability index of each robot. Energy consumption characteristics extraction requires analyzing the relationship between energy consumption rate, task type, and task intensity to form an energy consumption feature vector. Device temperature characteristics extraction requires classifying collected temperature data according to different components and analyzing the correlation between temperature, task execution time, environmental factors, and other factors to form the device temperature load characteristics.

[0066] After extracting cluster status features, the cluster management module needs to filter out clusters that meet the spatial distribution criteria based on their spatial distribution characteristics. Specifically, various filtering thresholds for spatial distribution features are set, such as maximum coordinate deviation, minimum coverage, and minimum communication quality standards. Based on these thresholds, the cluster management module will screen all collected cluster data one by one, selecting clusters whose coordinate deviations are within the set range, whose coverage meets the mission area requirements, and whose communication quality meets the minimum standards. During the screening process, a comprehensive evaluation of each spatial distribution feature is required. This can be achieved using a weighted scoring method, assigning weights to different features based on their importance. A comprehensive spatial distribution score is then calculated for each cluster, and screening is then performed based on the preset score thresholds.

[0067] After selecting clusters that meet the spatial distribution characteristics, these clusters are organized into an initial candidate set and output in a specific data structure. This data structure must contain basic information about each cluster, spatial distribution characteristics, and task load characteristics, enabling subsequent processing and analysis. When outputting the initial candidate set, data accuracy and completeness must be ensured, while ensuring the stability and reliability of data transmission to avoid data loss or errors during transmission.

[0068] Example 2:

[0069] The implementation of step S2 involves the acquisition of historical response data, the analysis of historical task characteristics and cluster function characteristics, the calculation of function matching, and the screening of the second candidate set.

[0070] The robot cluster's historical response data must be obtained from the data storage module or related database. This data contains a complete record of the robot cluster's past execution of various tasks, including the time the task was issued, the execution process, the completion status, and the status changes of each robot in the cluster.

[0071] After obtaining the historical response data, the characteristics of the historical tasks need to be parsed. The characteristics of historical tasks include task identification characteristics and functional characteristics. Task identification characteristics are information used to uniquely identify each historical task, such as the task number, task creation time, etc. These identifiers can be used to accurately distinguish different historical tasks. Functional characteristics describe the robot functions required to complete the task. For example, a handling task requires the robot to have grasping and moving functions, and a detection task requires the robot to have sensor perception and data transmission functions. When parsing functional characteristics, it is necessary to sort out the specific requirements of the task for the robot function in detail, including the type of function, performance indicators, etc., such as the load capacity of the grasping function in a handling task and the speed range of the moving function.

[0072] Filter out historical tasks that meet the task load conditions based on task identification features and task load features. Task load features include task processing volume, energy consumption, and device temperature. These features reflect the difficulty of executing the task and the resource consumption. When filtering, set the filtering conditions for task load features. For example, when the task processing volume exceeds a certain value, the energy consumption reaches a specific threshold, or the device temperature exceeds the safe range, these tasks will be filtered out. For example, when analyzing a certain type of industrial handling task, historical tasks with a task processing volume greater than 100 pieces / hour or an energy consumption exceeding 500Wh can be filtered out as the object of subsequent analysis.

[0073] While screening historical tasks, it's necessary to obtain the functional characteristics of the clusters in the initial candidate set. Each cluster in the initial candidate set consists of multiple robots, each with different capabilities. Therefore, the functional characteristics of a cluster are a comprehensive reflection of the capabilities of all robots within the cluster. To obtain the functional characteristics of a cluster, it's necessary to analyze the function type, quantity, and functional combination of each robot in the cluster. For example, if a cluster contains five robots with transport capabilities and three robots with detection capabilities, its functional characteristics are the combination of transport and detection capabilities, as well as parameters such as the corresponding number of robots.

[0074] After filtering historical tasks and acquiring cluster functional characteristics, the functional compatibility between these filtered historical tasks and the clusters in the initial candidate set needs to be analyzed. Specifically, the functional characteristics of the filtered historical tasks are compared against the functional characteristics of each cluster in the initial candidate set. For example, if a historical task requires handling (load capacity ≥ 10kg) and communication (transmission distance ≥ 50 meters), each cluster in the initial candidate set is checked to see if it possesses these two functions and whether their functional performance meets the task requirements.

[0075] During the item-by-item correspondence analysis, for each functional feature, we determine whether the cluster possesses that feature and whether the performance indicators of that feature meet the task requirements. If the cluster possesses the feature and the performance indicators meet the task requirements, it is considered a functional match; if it does not possess the feature or the performance indicators do not meet the task requirements, it is considered a mismatch. The ratio of the number of matching functional features to the total number of functional features is calculated, and this ratio is the functional match degree. For example, if a historical task has five functional features and a cluster matches four of them, the functional match degree between the cluster and the historical task is 80%.

[0076] When calculating functional compatibility, it's important to note that different functional features may have varying degrees of importance. Different weights can be assigned to each feature based on actual needs. For example, core functional features can be given a higher weight, while auxiliary functional features can be given a lower weight. This weighted calculation yields a more accurate functional compatibility. However, in specific implementation, the rules and basis for setting weights must be clearly defined to ensure the rationality and scientific nature of the functional compatibility calculation.

[0077] After calculating the functional match between all clusters and the filtered historical tasks, the functional match is sorted in descending order. After sorting, the corresponding clusters are selected based on the set sequence number. For example, to select the top 20% of clusters, after sorting from high to low by functional match, the top 20% of clusters are selected to form the second candidate set. The sequence number can be set as a specific number, such as the top 10 clusters, or as a percentage, such as the top 20%, depending on the actual application scenario and requirements.

[0078] During the screening process, it is important to ensure accurate sorting and screening to avoid biased screening results due to data processing errors. Furthermore, consideration should be given to cases where functional compatibility is similar. Additional screening criteria, such as the cluster's workload capacity and energy efficiency, can be set to distinguish clusters with similar functional compatibility.

[0079] After completing the second candidate set screening, the set needs to be output in a specific format for use in subsequent steps. The output should include detailed information about each cluster, such as cluster number, functional characteristics, and functional matching, to ensure that subsequent steps can accurately obtain and use this information.

[0080] Example 3:

[0081] The implementation of step S3 involves extracting the response duration and execution accuracy from the filtered historical response data, calculating the average response interval and average execution accuracy, judging the response behavior pattern of the cluster by comparing with the preset threshold, and obtaining the cluster response preference score value accordingly.

[0082] The historical response data needs to be screened to select valid data relevant to the current analysis. This data should come from the response records corresponding to the historical tasks screened in step S2 to ensure the data is targeted and valid.

[0083] When extracting response time, it's important to clearly define response time: the time interval between task issuance and completion. For each historical response data entry, accurately record the task issuance and completion times, and calculate the response time using the time difference. For example, if a task is issued at 10:00:00 on June 1, 2025, and completed at 10:15:30 on June 1, 2025, the response time is 15 minutes and 30 seconds. During the extraction process, ensure the accuracy of time recording to avoid miscalculation of response time due to timestamp errors.

[0084] The extraction of execution accuracy requires different indicators depending on the task type. For example, for a handling task, execution accuracy can be expressed as the deviation distance between the object placement position and the target position; for a detection task, execution accuracy can be expressed as the error range of the detection data. When extracting execution accuracy, it is necessary to clarify the specific accuracy indicators and measurement methods to ensure the consistency and comparability of the data. For example, in a handling task, the straight-line distance between the center point of the object and the center point of the target position is used as the position deviation, with the unit being centimeters; in a detection task, the relative error between the detection data and the actual value is used as the execution accuracy indicator, with the unit being percentage.

[0085] To obtain the response time interval between two adjacent historical response data sets, arrange the historical response data in chronological order and calculate the interval between the completion time of the previous task and the issuance time of the next task. For example, if the previous task is completed at 10:15:30 and the next task is issued at 10:20:00, the response time interval is 4 minutes and 30 seconds. For multiple historical response data sets, calculate the response time interval for each pair of adjacent data sets sequentially to form a response time interval sequence.

[0086] When averaging several acquired response intervals, first check for outliers in the response interval sequence, such as significantly longer or shorter intervals. These outliers may be caused by data recording errors or unique task scenarios. If outliers exist, they should be addressed based on the actual situation, such as removal or correction. After processing, add up all valid response intervals and divide by the number of intervals to obtain the average response interval. For example, if there are five response intervals of 5, 6, 4, 7, and 5 minutes, the total is 27 minutes, and the average is 5.4 minutes.

[0087] When averaging several acquired execution accuracies, the accuracy data must also be preprocessed to remove outliers or erroneous data. Calculations are then performed based on the accuracy metric type. If the accuracy is a numerical metric such as deviation distance, the arithmetic mean can be directly calculated. If it is a percentage metric such as relative error, the arithmetic mean method can also be used. For example, the position deviations for five handling tasks were 2 cm, 3 cm, 1 cm, 4 cm, and 2 cm, respectively, with an average of 2.4 cm.

[0088] When presetting the first and second interval thresholds, they should be appropriately set based on the actual performance of the robot cluster and the task requirements. For example, for certain rapid-response tasks, the first interval threshold could be set to 10 minutes, and the second interval threshold to 30 minutes. For routine tasks, the thresholds can be adjusted accordingly. These thresholds should be subject to multiple tests and field validation to ensure they accurately reflect the cluster's response speed level.

[0089] When comparing the average response interval with the preset first and second interval thresholds to determine the cluster's response behavior pattern, the judgment must be made strictly according to the preset logic. If the average response interval is less than or equal to the first interval threshold, for example, if the average response interval is 8 minutes and the first interval threshold is 10 minutes, the cluster is judged to have a fast response task execution speed. If the average response interval is greater than the first interval threshold and less than the second interval threshold, for example, if the average response interval is 15 minutes, the first interval threshold is 10 minutes, and the second interval threshold is 30 minutes, the cluster is judged to have a medium execution speed. If the average response interval is greater than or equal to the second interval threshold, for example, if the average response interval is 35 minutes and the second interval threshold is 30 minutes, the cluster is judged to have a slow execution speed.

[0090] When deriving cluster response preference scores based on their response behavior patterns, a mapping between response behavior patterns and scores must be established. For example, clusters with fast execution speeds can be assigned higher scores, such as 80-100 points; clusters with medium execution speeds can be assigned intermediate scores, such as 50-79 points; and clusters with slow execution speeds can be assigned lower scores, such as 0-49 points. The specific range of scores and their mapping can be adjusted based on actual needs to ensure that the scores accurately reflect the cluster's response preferences.

[0091] Throughout the implementation process, careful attention must be paid to rigorous data preprocessing to ensure the accuracy of data such as response time, execution accuracy, and response interval. Clear rules and guidelines must be in place for handling abnormal data to avoid data bias caused by human factors. The setting of preset thresholds must be based on actual application scenarios, fully considering the robot's hardware performance, task type, and environmental factors to ensure that the thresholds are reasonable and scientific. The calculation method for the response preference score must be clear and unambiguous for easy reference and use in subsequent steps.

[0092] The sample size of historical response data must be considered, ensuring it is large enough to ensure statistical significance for the calculation of average response interval and average execution accuracy. A sample size that is too small may lead to inaccurate calculations, affecting the determination of response behavior patterns and the acquisition of response preference scores. Therefore, during implementation, it is necessary to accumulate sufficient historical response data or employ appropriate statistical methods to process small sample data to improve the reliability of the analysis results.

[0093] A comprehensive data management mechanism must also be established to record and store historical response data, calculation processes, and results for subsequent query and verification. When determining response behavior patterns and deriving response preference scores, logical consistency and repeatability must be maintained to ensure that different batches of data processing yield the same conclusions, thereby improving the stability and reliability of the entire control method.

[0094] Example 4:

[0095] The implementation of step S4 involves obtaining the response characteristics and task difficulty of the clusters in the second candidate set, analyzing the response adaptability in combination with the cluster response preference score and average execution accuracy, and screening out the third candidate set by sorting.

[0096] The response characteristics of each cluster must be extracted from the second candidate set. These characteristics include dynamic behavioral data such as the fluctuation range of the cluster's response time in historical tasks, the stability of its accuracy across different task types, and task switching efficiency. For example, a cluster's response time in historical transport tasks fluctuated between 5 and 10 minutes, and its execution accuracy fluctuated between 1 and 3 centimeters. This data must be compiled from historical response records.

[0097] Determining task difficulty requires a comprehensive analysis of the current task's attributes, including task complexity, environmental interference, and resource constraints. Task complexity can be reflected in the number of steps in the task process and the diversity of functional combinations. For example, a task requiring both transport and detection is more complex than a single transport task. Environmental interference includes factors such as communication interference intensity and obstacle density. For example, executing a task in an environment with strong electromagnetic interference is more difficult than in a normal environment. Resource constraints include energy constraints and time window requirements. For example, a task that must be completed within 30 minutes is more difficult than a task without a strict time limit.

[0098] The cluster response preference score is determined in step S3 based on the cluster's response behavior pattern. For example, a cluster with fast execution speed is scored 90 points, a cluster with medium execution speed is scored 70 points, and a cluster with slow execution speed is scored 50 points. The average execution accuracy is the average execution accuracy of historical tasks calculated in step S3. For example, the average position deviation of a cluster in historical tasks is 2.5 cm.

[0099] When analyzing the response compatibility of a cluster with clusters in the second candidate set, two core dimensions must be considered. First, the degree of match between the cluster's response preference score and its response characteristics. For example, a cluster with a high score (fast execution speed) should have response characteristics characterized by a small fluctuation in response time and stable task execution speed. If a cluster scores 90 points but has a historical response time fluctuation of 20 minutes, the compatibility is low. Conversely, if the response time fluctuates between 5 and 8 minutes, the compatibility is high. Second, the degree of fit between average execution accuracy and task difficulty is considered. For high task difficulty (e.g., requiring a position deviation of ≤1 cm), the cluster's average execution accuracy must meet or exceed this standard. For example, if a task requires an accuracy of ≤1.5 cm and the cluster's average execution accuracy is 1.2 cm, the compatibility is high. If the cluster's average execution accuracy is 2 cm, the compatibility is low.

[0100] In specific implementations, evaluation rules need to be established for each dimension. Regarding the degree of match between the cluster response preference score and the response characteristics, clusters with scores above 80 can be set to have a response time fluctuation of ≤10 minutes. If this exceeds, the matching degree is reduced. Response characteristics also include the cluster's adaptability to different task types. For example, if a cluster historically excels at transport tasks, but the current task is detection, the matching degree of its response characteristics with the current task needs to be adjusted based on the difference in task type.

[0101] To assess the degree of fit between average execution accuracy and task difficulty, the task difficulty must be broken down into specific accuracy requirements. For example, if a detection task requires data error ≤ 5%, then the average error of the cluster in historical detection tasks is compared. If it is 3%, the fit is high; if it is 8%, the fit is low. For tasks with multi-dimensional accuracy metrics (such as requiring both position accuracy and time accuracy), the fit of each dimension must be evaluated separately, and then the overall fit is calculated comprehensively.

[0102] A comprehensive assessment of response adaptability can be performed using a weighted summation approach, assigning weights to the two core dimensions. For example, the weight for the match between the cluster's response preference score and the response characteristics is set at 0.6, while the weight for the match between average execution accuracy and task difficulty is set at 0.4. The specific weights should be determined based on task requirements and historical experience. To calculate the adaptability score for each cluster, first calculate the scores for each dimension separately, then multiply them by the corresponding weights and add them together.

[0103] For example, if a cluster's response preference score matches its response characteristics with a score of 85, and its average execution accuracy matches its task difficulty with a score of 75, its total adaptability score is 85 × 0.6 + 75 × 0.4 = 81. If another cluster's match score is 70 and its fit score is 90, its total score is 70 × 0.6 + 90 × 0.4 = 78, indicating that the former is more adaptable than the latter.

[0104] After calculating the response suitability of all clusters, sort them in descending order based on their suitability scores. Sorting should strictly follow the highest-scoring order, ensuring that high-scoring clusters are prioritized. When selecting clusters, select clusters based on a set range of numbers, such as the top 10 or top 30%. The specific number can be determined based on the number of clusters required for the task. For example, if the current task requires the coordinated execution of five clusters, select the clusters with the top five suitability scores.

[0105] The selected clusters form the third candidate set, which must include detailed information about each cluster, such as its cluster ID, response characteristics, suitability score, and historical task execution records, to facilitate subsequent steps. When outputting the third candidate set, ensure that the data format is uniform and the content is complete to facilitate reading and processing by the real-time optimization control module.

[0106] Example 5:

[0107] The implementation of step S5 involves analyzing whether the cluster has a collaborative habit through the cluster's real-time response feedback record, and obtaining corresponding cluster information from the third candidate set based on the presence or absence of the collaborative habit to optimize the control instructions.

[0108] It is necessary to obtain real-time cluster response feedback records, which contain various real-time data during the cluster's current task execution, such as each robot's task execution status, task switching time, and multi-task parallel execution. This data is collected and transmitted in real time through sensors and communication modules deployed on the robots and the central control system to ensure data timeliness and accuracy.

[0109] To determine whether a cluster has a collaborative habit, we use a statistical analysis method: we count the percentage of simultaneous executions of multiple tasks in the cluster's real-time response feedback records relative to the total number of responses. "Simultaneous execution of multiple tasks" is defined as the number of times within a time period (e.g., a 5-minute window) at least two robots in the cluster execute different subtasks, or the same robot performs different tasks alternately in a time-sliced ​​manner. The total number of responses refers to the total number of task instructions received by the cluster within the same time period, including both single-task instructions and multi-task combinations.

[0110] For example, if a cluster receives 100 task instructions during a certain time period, and 35 of them require simultaneous execution of multiple tasks (e.g., instructions requiring Robot A to perform a transport task while Robot B performs a detection task), then the proportion of simultaneous execution of multiple tasks in the total response time is 35%. A preset collaborative habit threshold is set based on factors such as the type of robot cluster and the characteristics of the task scenario, for example, 30%. If the statistically calculated ratio exceeds the preset collaborative habit threshold (e.g., 35% > 30%), the cluster is deemed to have a collaborative habit; if the ratio does not exceed the threshold (e.g., 25% < 30%), the cluster is deemed to have no collaborative habit.

[0111] If it is determined that the clusters do not have a collaborative habit, it is necessary to obtain the cluster information with the highest response match from the third candidate set. The response match is calculated based on the response adaptability evaluation results in step S4. The clusters in the third candidate set are sorted in descending order of response adaptability, so the cluster with the highest response match is the first cluster in the sorting. When obtaining the cluster information, it is necessary to extract its complete feature data, including the cluster's spatial distribution characteristics (such as coordinate deviation, coverage range, and communication quality), functional characteristics (functional types and combinations), response characteristics (response time fluctuation range, execution accuracy stability), and historical task execution records.

[0112] When optimizing control commands during the response process, for clusters that lack a collaborative approach, the focus is on the task execution efficiency of individual clusters. For example, the interval between task issuances is adjusted based on the cluster's response characteristics to avoid cluster overload caused by excessively frequent task issuance. Based on the cluster's functional characteristics, tasks of a single type are rationally allocated to ensure efficient cluster completion. Furthermore, by integrating the cluster's real-time response feedback, such as the current task's execution progress and device status, control command parameters, such as task execution priority and speed settings, are dynamically adjusted to suit the cluster's actual execution capabilities.

[0113] When a cluster is determined to have a collaborative habit, the information for the cluster within the set number in the third candidate set is obtained. The set number should take into account factors such as the complexity of the task and the number of robots required to collaborate, such as the top 5 or top 20% of clusters. When obtaining information on these clusters, not only is the characteristic data of each cluster extracted, but also the historical collaboration records between clusters, such as the number of tasks they have performed together, the quality of the collaborative tasks, and the roles each cluster played in the collaboration.

[0114] For clusters with a history of collaboration, optimizing control instructions during the response process is crucial to promoting inter-cluster collaboration. For example, based on each cluster's historical collaboration records, the collaborative roles and task allocation methods for the current task are determined, combining clusters with complementary functional characteristics to form efficient collaborative units. Based on the communication quality characteristics between clusters, communication protocols and data transmission paths are optimized to reduce communication delays and data loss during the collaborative process. Furthermore, based on the cluster's historical performance in executing multiple tasks simultaneously, a reasonable task parallel execution strategy is established to avoid resource conflicts caused by multi-tasking concurrency.

[0115] During implementation, the frequency of real-time response feedback recording must match the pace of task execution to ensure timely capture of changes in the cluster's collaborative behavior. For example, for rapid response tasks, the collection frequency can be set to once per second; for routine tasks, the collection frequency can be set to once per minute. Furthermore, an efficient data processing mechanism must be established to clean and preprocess the collected real-time data in real time, removing noise and invalid data to ensure the accuracy of statistical analysis.

[0116] Adjustments to the preset collaboration habit threshold should be based on extensive historical data and practical application experience. For example, in industrial production scenarios, where robot swarms often need to collaborate frequently to complete complex processes, the collaboration habit threshold could be set at 40%. In simple handling scenarios, where the need for swarm collaboration is lower, the threshold could be set at 20%. Furthermore, the threshold can be dynamically adjusted based on different task types to improve the adaptability of collaboration habit determination.

[0117] When acquiring cluster information, ensure data security and integrity, employing encrypted transmission and backup mechanisms to prevent data loss or tampering during transmission and storage. When optimizing control instructions, establish an instruction verification mechanism and conduct simulation tests on the adjusted control instructions to ensure their feasibility and effectiveness, and avoid mission execution failures or robot malfunctions due to improper instruction optimization.

[0118] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0119] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent control method for robot cluster response, characterized in that: The specific steps include: S1. Collect historical response data of the robot cluster, extract cluster state features, and construct an initial candidate set of the robot cluster based on the cluster state features; S2. Analyze the historical tasks based on the historical response data of the robot cluster, and then determine the degree of matching between the historical tasks and the functions of each cluster in the initial candidate set based on the analysis results, and select a second candidate set of robot clusters; S3. Analyze the cluster’s response behavior pattern based on the robot cluster’s historical response data; S4. Analyze the clusters according to the response behavior patterns of the clusters, determine the response adaptability of the clusters in the second candidate set based on the results of the cluster analysis, and select a third candidate set of robot clusters; S5. Obtaining corresponding cluster information in the third candidate set based on the cluster real-time response feedback record, and optimizing control instructions during the response process; The S1 includes the following specific steps: S101. Inputting historical response data of a robot cluster into a data acquisition module, and extracting cluster state features from the response data, wherein the cluster state features include spatial distribution features and task load features; S102, screening out clusters containing spatial distribution features in the cluster management module according to the spatial distribution features and outputting them in the form of an initial candidate set; The S2 includes the following specific steps: S201. Obtain historical response data of a robot cluster, analyze features of historical tasks based on the historical response data, wherein the features of the historical tasks include task identification features and functional features, select historical tasks containing task load features based on the task identification features and the task load features, and simultaneously obtain functional features of clusters in the initial candidate set; S202: Analyze the functional matching degree between the filtered historical tasks and the clusters in the initial candidate set based on the functional characteristics of the filtered historical tasks and the functional characteristics of the clusters in the initial candidate set; S203, sorting the functional matching degrees in descending order, screening out the clusters corresponding to the set sequence numbers and outputting them as a second candidate set; The S3 includes the following specific steps: S301. Extract the response duration and execution accuracy from the filtered historical response data, obtain the response time interval between two adjacent historical response data, average the obtained response time intervals to obtain an average response interval, and average the obtained execution accuracies to obtain an average execution accuracy. S302, comparing the average response interval with a preset first interval threshold and a second interval threshold and determining the response behavior pattern of the cluster; if the average response interval is less than or equal to the first interval threshold, determining that the cluster response task execution speed is fast; if the average response interval is greater than the first interval threshold and the response interval is less than the second interval threshold, determining that the cluster response task execution speed is medium; if the average response interval is greater than or equal to the second interval threshold, determining that the cluster response task execution speed is slow; S303, obtaining a cluster response preference score value according to the cluster response behavior pattern; The S4 includes the following specific steps: S401, obtaining the response characteristics and task difficulty of the clusters in the second candidate set, and analyzing the response compatibility of the cluster with the clusters in the second candidate set based on the cluster response preference score, average execution accuracy, the response characteristics and task difficulty of the clusters in the second candidate set; S402, sorting the response adaptability in descending order, screening out the clusters corresponding to the set sequence number and outputting them as a third candidate set; The S5 includes the following specific steps: Obtaining real-time cluster response feedback records to analyze whether the cluster has a collaborative habit. If the cluster does not have a collaborative habit, obtaining the cluster information with the highest response match in the third candidate set and optimizing control instructions during the response process. If the cluster does have a collaborative habit, obtaining the cluster information corresponding to the set sequence number in the third candidate set and optimizing control instructions during the response process. When optimizing control instructions during the response process, for clusters that do not have a collaborative habit, the focus of optimizing control instructions is on the task execution efficiency of a single cluster. For clusters that do have a collaborative habit, the core of optimizing control instructions during the response process is to promote collaborative operations between clusters. The method for analyzing whether the cluster has a collaborative habit is to count the proportion of the number of times multiple tasks are executed simultaneously in the cluster's real-time response feedback records to the total number of responses. If the proportion exceeds a preset collaborative habit threshold, it is determined that the cluster has a collaborative habit.

2. The intelligent control method for robot cluster response according to claim 1, characterized in that: The spatial distribution characteristics include specific distribution characteristics of coordinate deviation, coverage range, and communication quality, and the task load characteristics include load characteristics of task processing volume, energy consumption, and device temperature.

3. The intelligent control method for robot cluster response according to claim 1, characterized in that: The response adaptability analysis of the cluster and the clusters in the second candidate set includes the matching degree between the cluster response preference score value and the cluster response characteristics, and the matching degree between the average execution accuracy and the task difficulty.

4. The intelligent control method for robot cluster response according to claim 1, characterized in that: The functional matching degree is analyzed by performing a correspondence analysis on the functional features of the filtered historical tasks and the functional features of the clusters in the initial candidate set, and calculating the ratio of the number of matching functional items to the total number of functional items.

5. An intelligent control system for robot cluster response, used to implement the intelligent control method for robot cluster response according to any one of claims 1 to 4, characterized in that: include: The data collection and feature extraction module is used to collect historical response data of the robot cluster, extract cluster state features and construct the initial candidate set of the robot cluster; a function matching screening module, connected to the data acquisition and feature extraction module, configured to analyze historical tasks based on the historical response data of the robot cluster, and then determine the degree of functional matching between the historical tasks and each cluster in the initial candidate set based on the analysis results, thereby screening out a second candidate set of robot clusters; A behavior pattern analysis module, connected to the data acquisition and feature extraction module, for analyzing the response behavior pattern of the cluster based on the historical response data of the robot cluster; a response adaptation screening module, connected to the behavior pattern parsing module and the function matching screening module, configured to parse the clusters according to their response behavior patterns, determine the response adaptability of the clusters in the second candidate set based on the cluster parsing results, and screen out a third candidate set of robot clusters; The real-time optimization control module is connected to the response adaptation screening module and is used to obtain corresponding cluster information in the third candidate set according to the cluster real-time response feedback record, and optimize the control instructions during the response process.

Images