Intelligent scheduling management method and system for robot cluster

Abstract

The invention relates to the field of robot cluster control, and particularly discloses an intelligent scheduling management method and system for a robot cluster, and the method comprises the steps: precisely predicting a future energy consumption curve of a robot through fusing distributed tasks of the robot with map data; and meanwhile, the available charging window in the future is mastered by performing time sequence prediction on the resources of the charging station. Based on the two predictions, the decision engine can further combine the busy degree of the current task queue to quantitatively evaluate the energy risk and opportunity cost of the robot for charging, and comprehensively calculate a dynamic charging urgency degree. According to the method, the charging behavior is converted from an isolated emergency event into a profound planning link which is integrated into the whole task flow of the system, so that the problems of charging station congestion and robot downtime caused by lack of perspectiveness are effectively avoided, and the energy safety of the robot is ensured while the energy safety of the robot is ensured. And the operation efficiency of the whole cluster is maximized.

Description

Technical Field

[0001] This application relates to the field of robot swarm control, and more specifically, to an intelligent scheduling and management method and system for robot swarms. Background Technology

[0002] With the rapid development of automation and intelligent technologies, robot swarms have been widely applied in numerous fields such as warehousing and logistics, intelligent manufacturing, and security inspection. To maximize the operational efficiency and return on investment of robot swarms, building an efficient and intelligent scheduling and management solution is crucial. The core objective of this solution is to ensure the swarm can continuously and stably execute tasks. Effective energy management of the robots, i.e., charging scheduling, is a key factor determining the overall operational efficiency of the swarm. A poorly managed charging strategy can lead to robots unexpectedly shutting down due to depleted power or causing congestion at charging stations, severely impacting the continuity of task execution and overall operational throughput.

[0003] Currently, most mainstream robot swarm charging management solutions in the industry adopt a reactive strategy based on fixed power thresholds. The logic of this strategy is relatively simple: when a robot's power level falls below a preset threshold (e.g., 20%), the system generates a high-priority charging task for it. However, this reactive management approach has significant drawbacks. First, it lacks foresight, only responding when the robot's power level is already low, failing to plan ahead. Second, this strategy decouples charging decisions from task scheduling, failing to fully consider the overall operational status of the system. For example, during peak business periods, multiple robots may consume large amounts of power within a similar timeframe, causing them to trigger the charging threshold almost simultaneously, leading to a rush to limited charging stations, resulting in severe queuing and congestion, and ultimately reducing the number of effective robots the system can effectively utilize when most demanding. Conversely, during off-peak periods, charging station resources may be idle, and some robots with sufficient power may miss the optimal charging window because they haven't triggered the threshold. This strategy inherently cannot quantify and assess the opportunity cost of charging at different times—the loss from missing high-value tasks due to charging—leading to suboptimal charging timing.

[0004] Therefore, an optimized intelligent scheduling and management scheme for robot swarms is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide an intelligent scheduling and management method and system for robot swarms.

[0006] According to one aspect of this application, an intelligent scheduling and management method for a robot swarm is provided, comprising: Acquire the first robot status data, the charging station status data set, the task queue data, and the map data; Based on map data and the state data of the first robot, the future energy curve of the first robot is predicted to obtain the future energy curve of the first robot; Predict future charging station availability based on charging station status data set to obtain charging station availability timeline set; The task queue data, the charging station availability timeline set, and the future energy curve of the first robot are input into the charging decision engine to obtain the charging decision results. Based on the charging decision results and task queue data, determine whether to generate a charging task instruction for the first robot.

[0007] According to another aspect of this application, an intelligent scheduling and management system for a robot swarm is provided, comprising: The data acquisition module is used to acquire the first robot's status data, the charging station's status data set, the task queue data, and the map data; The energy curve prediction module is used to predict the future energy curve of the first robot based on map data and the state data of the first robot to obtain the future energy curve of the first robot. The charging station availability prediction module is used to predict the future availability of charging stations based on the charging station status data set to obtain a charging station availability timeline set. The charging decision module is used to input task queue data, charging station availability timeline set and the future energy curve of the first robot into the charging decision engine to obtain charging decision results; The charging task generation module is used to determine whether to generate a charging task instruction for the first robot based on the charging decision result and task queue data.

[0008] Compared with existing technologies, this application provides an intelligent scheduling and management method and system for robot swarms. By integrating assigned tasks and map data, it accurately predicts the future energy consumption curve of the robots. Simultaneously, it predicts charging station resources over time to determine future available charging windows. Based on these two predictions, the decision engine further combines the current task queue's busyness to quantitatively assess the energy risk and opportunity cost of charging the robots, and comprehensively calculates a dynamic charging urgency. This method transforms charging from an isolated emergency event into a well-thought-out planning step integrated into the overall system task flow, effectively avoiding charging station congestion and robot downtime caused by a lack of foresight. It maximizes the operational efficiency of the entire swarm while ensuring robot energy safety. Attached Figure Description

[0009] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 This is a flowchart of an intelligent scheduling and management method for a robot swarm according to an embodiment of this application; Figure 2 This is a schematic diagram of the data flow in the intelligent scheduling and management method for a robot swarm according to an embodiment of this application; Figure 3 This is a flowchart illustrating how a robot cluster intelligent scheduling and management method based on map data and first robot state data predicts the future energy curve of a first robot to obtain the future energy curve of the first robot, according to an embodiment of this application. Figure 4 This is a flowchart illustrating the process of inputting task queue data, charging station availability timeline set, and the future energy curve of the first robot into a charging decision engine to obtain a charging decision result in the intelligent scheduling and management method for robot clusters according to an embodiment of this application. Figure 5 This is a flowchart illustrating the calculation of a charging urgency score based on quantified decision factors and a basic weight vector in an intelligent scheduling and management method for a robot swarm according to an embodiment of this application. Figure 6 This is a block diagram of an intelligent scheduling and management system for a robot swarm according to an embodiment of this application. Detailed Implementation

[0011] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0012] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0013] While this application makes various references to certain modules of the systems according to embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0014] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0015] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0016] Existing robot swarm charging scheduling generally employs reactive strategies based on static energy thresholds. This lack of foresight often leads to charging station congestion and unexpected robot downtime during peak periods, severely impacting the overall operational efficiency of the swarm. To address these technical issues, this application proposes an intelligent scheduling and management method for robot swarms, the core of which is to transform charging decisions from passive response to proactive planning. Specifically, this method first accurately predicts the future energy state of each robot. By analyzing the robot's assigned task sequence and combining it with map data, it precisely calculates the energy consumption of future paths and task execution, generating a detailed future energy consumption curve. Simultaneously, it dynamically predicts the future availability time windows of each charging station, forming a time-series map of charging resources. Based on these two forward-looking data points, the decision engine further integrates an analysis of the current task queue's busyness, comprehensively quantifying and evaluating it from three dimensions: energy risk, opportunity cost, and charging accessibility, ultimately calculating a dynamic charging urgency score. This score will serve as the basis for decision-making, replacing the rigid power threshold. This allows the system to make the optimal trade-off between ensuring energy security and maximizing task benefits, and proactively generate charging instructions at the most appropriate time, thereby fundamentally solving the congestion and inefficiency problems caused by reactive strategies.

[0017] The technical solution of this application proposes an intelligent scheduling and management method for robot swarms. Figure 1 This is a flowchart of an intelligent scheduling and management method for a robot cluster according to an embodiment of this application. Figure 2 This is a schematic diagram of the data flow in an intelligent scheduling and management method for a robot swarm according to an embodiment of this application. Figure 1 andFigure 2 As shown, the intelligent scheduling and management method for a robot cluster according to an embodiment of this application includes the following steps: S100, acquiring first robot state data, charging station state data set, task queue data, and map data; S200, based on the map data and the first robot state data, predicting the future energy curve of the first robot to obtain the future energy curve of the first robot; S300, based on the charging station state data set, predicting the future availability of charging stations to obtain a charging station availability timeline set; S400, inputting the task queue data, the charging station availability timeline set, and the future energy curve of the first robot into a charging decision engine to obtain a charging decision result; S500, based on the charging decision result and the task queue data, determining whether to generate a charging task instruction for the first robot.

[0018] Specifically, in step S100, the first robot state data, charging station state data set, task queue data, and map data are acquired. It should be understood that an isolated, static data point, such as the robot's current battery level, is insufficient to support a forward-looking, globally optimal scheduling decision. An intelligent decision must be based on a dynamic, real-time, and comprehensive understanding of the entire operating environment. Therefore, in the technical solution of this application, by acquiring the first robot state data, charging station state data set, task queue data, and map data, a complete, accurate, and time-synchronized data foundation is provided for the subsequent energy curve prediction and charging decision-making process. This transforms the scheduling problem from processing fragmented and partial information to data analysis and optimization based on a unified, structured system state snapshot, thereby laying the foundation for proactive and predictive intelligent scheduling.

[0019] More specifically, this acquisition process is implemented through communication and data aggregation between the scheduling management server and various entities within the cluster and external systems. In a specific example of this application, the scheduling management server first establishes a connection with all devices in the field via a wireless communication network. Firstly, the server periodically sends query commands to the first robot or receives heartbeat data packets proactively reported by the robot, thereby obtaining the first robot's status data, which includes its unique identifier, current coordinates provided by the onboard positioning system, precise battery percentage feedback from the battery management system, and current operating status. Secondly, the server polls each charging station or subscribes to its status update events through standard communication protocols to obtain a set of charging station status data containing information such as the charging station's unique identifier, current availability, occupancy, and fault status. Thirdly, the server interfaces with an upper-level task management system, such as a warehouse management system, to pull or receive a list of tasks to be executed in real time. This list constitutes task queue data containing information such as task priority, target work point coordinates, and estimated working hours. Fourth, during the initialization phase, the server loads a pre-drawn electronic map representing the physical layout and feasible paths of the operating environment from local storage or configuration center. This map data exists in the form of a graph structure containing location nodes and path distances.

[0020] Specifically, in step S200, based on map data and the state data of the first robot, the future energy curve of the first robot is predicted to obtain the future energy curve of the first robot. It should be understood that reactive strategies that make decisions solely based on the robot's current battery level lack foresight and cannot identify and mitigate the risk of energy depletion due to future task execution. Therefore, in the technical solution of this application, the future energy curve of the first robot is further predicted based on map data and the state data of the first robot to obtain the future energy curve of the first robot. This quantifies and temporally tracks the future energy state changes of the robot, providing a crucial and forward-looking judgment basis for the subsequent charging decision engine. This enables the system to shift from passively responding to battery alarms to proactively planning energy replenishment, thereby significantly improving the stability and continuity of cluster operations.

[0021] Figure 3 This document describes a flowchart illustrating how a robot swarm intelligent scheduling and management method, based on map data and first robot state data, predicts the future energy curve of a first robot to obtain the future energy curve of that first robot, according to an embodiment of this application. Figure 3As shown, step S200 includes: S210, extracting assigned task data from the task queue data; S220, deconstructing and quantizing the task energy consumption parameters of the first robot state data based on the assigned task data and map data to obtain quantized energy consumption parameters; S230, calculating segmented energy consumption and time consumption based on the quantized energy consumption parameters to obtain a segmented consumption profile; S240, synthesizing the current battery level, current timestamp, and segmented consumption profile in the first robot state data into a future energy curve to obtain the future energy curve of the first robot.

[0022] Accordingly, in steps S210 and S220, assigned task data is extracted from the task queue data, and based on the assigned task data and map data, the task energy consumption parameters of the first robot state data are deconstructed and quantized to obtain quantified energy consumption parameters. It should be understood that since the original task instructions, such as target position coordinates, do not directly contain the physical quantity information required for energy consumption calculation, it is impossible to directly and accurately model future energy consumption. Therefore, in the technical solution of this application, assigned task data is further extracted from the task queue data, and based on the assigned task data and map data, the task energy consumption parameters of the first robot state data are deconstructed and quantified to obtain quantified energy consumption parameters. This transforms the abstract task objective into a set of specific, quantified physical parameters that can be used for subsequent energy consumption modeling. Specifically, the quantified energy consumption parameters here include the planned path, travel distance, and task energy consumption, which can provide accurate input for subsequent segmented energy consumption calculations, ensuring the accuracy and reliability of future energy curve predictions.

[0023] Specifically, in a specific example of this application, the process first retrieves and extracts the complete record of the corresponding assigned task data from the task queue data based on the assigned task identifier contained in the first robot's state data. Then, it obtains the current position coordinates from the first robot's state data and the target position coordinates of the task from the extracted assigned task data. Based on these two coordinates, the A* path planning algorithm is invoked on preloaded map data to calculate an optimal planned path connecting the starting point and the ending point, which consists of a series of continuous map nodes. By summing the lengths of all path segments on the planned path, the total travel distance required to complete the path movement is calculated. Finally, the calculated planned path, travel distance, and the estimated task energy consumption directly obtained from the assigned task data are encapsulated together into a structured quantified energy consumption parameter object for use in subsequent steps.

[0024] Accordingly, in step S230, segmented energy consumption and time consumption are calculated based on the quantified energy consumption parameters to obtain a segmented consumption profile. It should be understood that since the quantified energy consumption parameters are only a static set of physical quantities, they do not directly reflect the dynamic process of time and energy consumption at different stages in completing the task, and therefore cannot be directly used to construct a time-series energy curve. Therefore, in the technical solution of this application, segmented energy consumption and time consumption are further calculated based on the quantified energy consumption parameters to obtain a segmented consumption profile. This decomposes the entire future task execution process into independent, calculable consumption stages, providing a structured data foundation for the final synthesis of an accurate, time-series future energy curve, ensuring the precision and accuracy of the prediction.

[0025] Specifically, in this embodiment, the calculation of segmented energy consumption and time consumption based on quantified energy consumption parameters to obtain a segmented consumption profile includes: inputting the travel distance from the quantified energy consumption parameters into the robot energy consumption model and the robot kinematics model to obtain travel time and travel energy consumption; extracting the estimated execution time associated with the quantified energy consumption parameters from the assigned task data; and encapsulating the travel time, travel energy consumption, estimated execution time, and task energy consumption to obtain the segmented consumption profile. More specifically, the calculation process first extracts the travel distance from the input quantified energy consumption parameters. This travel distance is input into the pre-calibrated robot kinematics model and robot energy consumption model. The robot kinematics model calculates the travel distance as a specific travel time based on a preset average travel speed; the robot energy consumption model calculates the travel distance as the corresponding travel energy consumption based on the energy consumption coefficient per unit distance. Simultaneously, the estimated execution time is extracted from the assigned task data through the task identifier associated with the quantified energy consumption parameters. Finally, the calculated travel time, travel energy consumption, and task energy consumption included in the extracted estimated execution time and quantified energy consumption parameters are encapsulated into structured data to generate a segmented consumption profile containing at least two segments: the travel phase and the task execution phase. Each segment records its corresponding duration and energy consumption value, providing clear and orderly data input for subsequent energy curve synthesis.

[0026] Accordingly, in step S240, the current battery level, current timestamp, and segmented consumption profile in the first robot state data are synthesized into a future energy curve to obtain the future energy curve of the first robot. It should be understood that since the segmented consumption profile only provides discrete, non-time-series consumption data fragments, it cannot intuitively reflect the trend of continuous change in the robot's battery level over time. Therefore, in the technical solution of this application, the current battery level, current timestamp, and segmented consumption profile in the first robot state data are further synthesized into a future energy curve to obtain the future energy curve of the first robot. This combines the discrete energy consumption calculation results with the robot's initial state to generate a continuous, time-indexed energy state prediction, ultimately producing clear future energy state time-series data that can be directly queried and analyzed by the decision engine, providing an accurate basis for predicting energy risk points.

[0027] Specifically, the synthesis of the future energy curve is an iterative calculation process based on the initial state. In a specific example of this application, the synthesis process first extracts the current battery level and current timestamp from the first robot state data, and uses this pair of values ​​as the first data point of the energy curve, i.e., the starting point of the prediction. Subsequently, each data segment in the segmented consumption profile is traversed sequentially. For the first segment, the driving stage, its duration (driving time) is added to the initial current timestamp to obtain the time point at the end of the stage; at the same time, its energy consumption (driving energy consumption) is subtracted from the initial current battery level to obtain the predicted battery level at the end of the stage. This new time point and the predicted battery level constitute the second data point of the energy curve. Next, the next segment, the task execution stage, is processed. Based on the second data point, the duration of this stage (estimated execution time) is added again and its energy consumption (task energy consumption) is subtracted to calculate the third data point of the energy curve. This process is repeated sequentially until all segments of the segmented consumption profile have been traversed, ultimately forming an ordered list of multiple data points arranged in chronological order (timestamp, predicted power). This list is the future energy curve of the first robot.

[0028] Specifically, in step S300, future charging station availability is predicted based on the charging station status data set to obtain a charging station availability timeline set. It should be understood that knowing only the current occupancy status of a charging station is insufficient for effective planning. A robot dispatched to a currently idle charging station may conflict with another robot due to information lag, or miss an optimal charging location that is about to become available. Therefore, in the technical solution of this application, future charging station availability is further predicted based on the charging station status data set to obtain a charging station availability timeline set. This transforms the static snapshot information of charging resources into a dynamic, time-indexed availability view, providing the charging decision engine with accurate spatiotemporal berth information. This allows for advance reservation and path optimization in the allocation of charging tasks, thereby maximizing the utilization rate of charging facilities and avoiding ineffective scheduling.

[0029] More specifically, the prediction process for future availability is a computational flow that extrapolates and synthesizes data based on the current state. In a specific example of this application, the scheduling management server first traverses each charging station in the charging station state data set. For charging stations in the "currently available" state, an open availability time window starting from the current timestamp and without a set end time is directly generated for them. For charging stations in the "currently occupied" state, the unique identifier of the robot currently charging is extracted from its state data. Subsequently, the robot's state data is queried to obtain its current battery level and the preset target charging level. Based on the charging station's rated charging power and the robot's battery capacity, the remaining time required to complete the charging process is calculated. Adding this remaining time to the current timestamp yields the exact starting time of the charging station's future availability, and a new availability time window is generated accordingly. Finally, the availability time windows calculated for each charging station are aggregated to form a set of charging station availability timelines indexed by the charging station identifier and containing a list of availability time windows.

[0030] Specifically, in step S400, task queue data, charging station availability timeline set, and the future energy curve of the first robot are input into the charging decision engine to obtain a charging decision result. It should be understood that since the robot's future energy state, the availability of charging resources, and the system's current task load are multiple dimensions that interact and jointly determine the optimal charging time, information from any single dimension is insufficient to support a globally optimal decision. Therefore, in the technical solution of this application, task queue data, charging station availability timeline set, and the future energy curve of the first robot are further input into the charging decision engine to obtain a charging decision result. This allows for the fusion, weighing, and comprehensive evaluation of all relevant and forward-looking data within a unified computational framework. This enables the final charging decision to no longer be a simple judgment based on isolated rules, but rather an intelligent optimization result that seeks to maximize the overall system benefit under multiple constraints.

[0031] Figure 4 This is a flowchart illustrating how the intelligent scheduling and management method for a robot swarm, according to an embodiment of this application, inputs task queue data, a set of charging station availability timelines, and the future energy curve of the first robot into a charging decision engine to obtain a charging decision result. Figure 4 As shown, step S400 includes: S410, calculating energy risk score, opportunity cost score and charging accessibility score to obtain quantitative decision factors based on task queue data, charging station availability timeline set and future energy curve of the first robot; S420, calculating charging urgency score based on quantitative decision factors and basic weight vector; S430, generating the charging decision result based on comparison between charging urgency score and charging urgency threshold.

[0032] Accordingly, in step S410, based on task queue data, the charging station availability timeline set, and the future energy curve of the first robot, an energy risk score, opportunity cost score, and charging accessibility score are calculated to obtain quantitative decision factors. It should be understood that, due to the different dimensions and physical meanings of the original, multi-source, heterogeneous input data, such as time-series energy curves and structured task lists, they cannot be directly subjected to unified mathematical operations or weighted comparisons. Therefore, in the technical solution of this application, the energy risk score, opportunity cost score, and charging accessibility score are further calculated based on task queue data, the charging station availability timeline set, and the future energy curve of the first robot to obtain quantitative decision factors. This abstracts and maps the three core decision dimensions—the urgency of the robot's energy state, the busyness of the system tasks, and the accessibility of charging resources—from the complex original data into standardized, dimensionless numerical scores. This provides a unified and standardized input vector for the subsequent synthesis of the charging urgency score, enabling effective weighted and coupled analysis of decision factors with different physical meanings within the same mathematical framework.

[0033] Specifically, this quantification process involves parallel and independent calculation and normalization of three core dimensions. More specifically, the process first quantifies energy risk by analyzing the future energy curve of the first robot to locate its predicted minimum energy level. This minimum energy level is then mapped to a range of 0 to 1 using a piecewise linear function: a score of 0 when the energy level is above a safety threshold, a score of 1 when the energy level is below a critical threshold, and a score that increases linearly with decreasing energy level when the energy level is between the two, thus obtaining an energy risk score. Simultaneously, opportunity cost is quantified by iterating through the task queue data, weighting and summing the priorities of all pending tasks to obtain a total value representing the current system task load. This total value is then processed by a preset normalization function to convert it into an opportunity cost score between 0 and 1. Furthermore, by quantifying charging accessibility, it combines the first robot's current location with the charging station availability timeline to calculate the shortest travel time to each future available charging bay, selects the minimum value, and normalizes this minimum travel time to a maximum acceptable time value, thus obtaining a charging accessibility score. Finally, these three independently calculated scores are collectively encapsulated into a quantified decision factor object for use in subsequent steps.

[0034] Accordingly, in step S420, a charging urgency score is calculated based on the quantified decision factors and the basic weight vector. It should be understood that the original charging decision mechanism relies on a fixed linear weighted sum of three independent quantified decision factors—energy risk, opportunity cost, and charging accessibility—when synthesizing the charging urgency score. The inherent flaw of this method is that it presupposes that the contributions of each decision factor to the final urgency are independent, thus ignoring the strong nonlinear coupling effects between these factors in the real physical world. Especially in extreme scenarios where the robot's battery is nearly depleted and it is far from available charging facilities, the risk is not a simple summation of individual risk values, but rather an exponential amplification of the crisis. Linear models cannot capture and quantify this surge in urgency under combined risks, tending to underestimate the severity of such exacerbated scenarios, potentially leading to delayed or erroneous decisions by the scheduling system at critical moments, resulting in serious consequences such as robot downtime and task failure. Therefore, in the technical solution of this application, a dynamic synthesis method based on decision manifolds and state metric tensors is introduced, elevating the charging decision process from a linear space to a nonlinear geometric space that better reflects physical reality. Specifically, in the technical solution of this application, a charging urgency score is calculated based on quantitative decision factors and basic weight vectors to construct a nonlinear decision model that can measure and amplify the risk of key factor combination. This makes the final urgency assessment result closer to the complex risk scenarios in actual operation, avoids decision-making errors caused by underestimating the combination risk, and thus significantly improves the robustness and accuracy of decision-making.

[0035] Figure 5 This is a flowchart illustrating the calculation of a charging urgency score based on quantified decision factors and a basic weight vector in an intelligent scheduling and management method for a robot swarm according to an embodiment of this application. Figure 5 As shown, step S420 includes: S421, constructing a state metric tensor based on the quantized decision factor and the basic weight vector, wherein the state metric tensor includes a coupling term between the energy risk score and the charging accessibility score; S422, calculating the charging urgency score based on the state metric tensor and the quantized decision factor.

[0036] Specifically, in step S421, a state metric tensor is constructed based on the quantified decision factors and the basic weight vector. This state metric tensor includes a coupling term between the energy risk score and the charging accessibility score. It should be understood that a simple weight vector can only define the independent importance of each decision factor, but cannot characterize the synergistic amplification risk generated when a specific combination of factors occurs, i.e., the nonlinear coupling effect. Therefore, in the technical solution of this application, a state metric tensor is further constructed based on the quantified decision factors and the basic weight vector. This state metric tensor includes a coupling term between the energy risk score and the charging accessibility score, thereby upgrading the decision model from linear weighting to a quadratic metric capable of capturing and quantifying the interaction between key risk factors, thus describing the dynamic coupling relationship between decision factors. This ensures that when the robot faces both low battery risk and difficulty in reaching a charging station, this combined risk is disproportionately amplified in the final urgency assessment, thereby making the decision system more sensitive to the most dangerous critical states.

[0037] More specifically, in this embodiment of the application, the specific implementation process for constructing the state metric tensor based on quantized decision factors and basic weight vectors is as follows: The quantified energy risk, opportunity cost, and accessibility factor are regarded as state vectors in the decision space, and a state-dependent metric tensor matrix is ​​constructed for them. Specifically, based on the quantized decision factors and basic weight vectors, the state metric tensor is constructed using the following formula:

[0038] in, and This is the coupling term between the energy risk score and the charging accessibility score. For the state metric tensor, , and These are the preset weight values ​​for the energy risk score, opportunity cost score, and charging accessibility score in the basic weight vector. Specifically, the process first initializes a square matrix matching the dimensions of the quantified decision factors; this matrix serves as the mathematical carrier of the state metric tensor. Then, each element of the basic weight vector, corresponding to the weight values ​​for energy risk, opportunity cost, and charging accessibility, is sequentially filled into the main diagonal of this square matrix. Next, the coupling term is introduced, filling the coupling coefficient values ​​into the intersection of the row and column corresponding to both the energy risk score and the charging accessibility score in the square matrix. To maintain matrix symmetry, the same value is filled into the corresponding transpose position. For example, the coupling term for energy risk and accessibility can be expressed by the following formula: When energy risk increases, this term grows exponentially, dynamically distorting the geometry of the decision space:

[0039] in, and The scores are energy risk score and charging accessibility score. and These are the preset trainable coupling parameters. All other off-diagonal elements in the matrix are set to zero to indicate that there is no direct coupling relationship between other factors. Ultimately, this filled symmetric matrix has diagonal elements defining the independent influence of each factor, while the non-zero coupling terms on the off-diagonal precisely mathematically represent the synergistic risk relationship between energy risk and charging accessibility. Together, they constitute a complete state metric tensor, generating a state metric tensor capable of encoding nonlinear effects such as crisis amplification into the spatial geometry itself. Thus, the originally flat and monotonous Euclidean decision space is transformed into a curved Riemannian decision manifold that can dynamically change according to the robot's real-time state.

[0040] Specifically, in step S422, the charging urgency score is calculated based on the state metric tensor and the quantized decision factor. It should be understood that since the state metric tensor and the quantized decision factor are two independent mathematical entities—the former defining the measurement and coupling rules for decision-making, and the latter representing the current state of the system—they must be combined through a defined computational process to obtain a single, final index usable for decision-making. Furthermore, in non-Euclidean geometry, the length or size of a vector cannot be measured by simple component summation; a measurement method adapted to the spatial geometry must be used. Therefore, in the technical solution of this application, the charging urgency score is further calculated based on the state metric tensor and the quantized decision factor to perform a comprehensive evaluation operation. This projects the multi-dimensional state vector onto a one-dimensional urgency scale according to a preset nonlinear measurement rule, ultimately generating a precise quantitative decision-making basis that includes the weights of each independent risk while also amplifying key coupled risks.

[0041] More specifically, in this embodiment, the charging urgency score is calculated based on the state metric tensor and the quantization decision factor, and the charging urgency score is defined as a state vector. In the metric tensor The square root of the norm (i.e., length) under the defined local geometry is calculated using the following quadratic form, where the cross term reflecting the coupling effect is explicitly included, as shown in the formula:

[0042] in, It is a vector composed of quantified decision factors. For the state metric tensor, Score for charging urgency.

[0043] Expanding, we get:

[0044] Specifically, the calculation process will first be based on the energy risk score. Opportunity cost score and charging accessibility score The quantified decision factors are organized into a column vector. Then, a standard quadratic calculation is performed, involving three matrix multiplication operations: First, the transpose of the quantified decision factor column vector is calculated, resulting in a row vector; next, this row vector is multiplied by a pre-constructed state metric tensor, yielding a new row vector; finally, this new row vector is multiplied by the original quantified decision factor column vector. The final result of this series of operations is a single numerical value, a scalar. This scalar value precisely reflects the norm of the quantified decision factors in the space defined by the state metric tensor. Its magnitude comprehensively reflects the superposition of the independent contributions of each factor and the key coupling effects; this is the charging urgency score, calculating a geometrically corrected single scalar value that truly reflects the portfolio risk. As a score reflecting the urgency of charging, the resulting urgency score inherently and non-linearly amplifies the combined risk. When energy risk and charging accessibility risk coexist, the presence of the cross term will lead to a final score that is much greater than the linear sum of the two, thereby giving the system a higher sensitivity to potential crisis states and a stronger early warning capability.

[0045] In summary, this method of calculating charging urgency scores, by introducing mathematical tools such as decision manifolds and metric tensors, can dynamically perceive and amplify combined risks, especially in high-risk edge scenarios, making judgments far more accurate and timely than linear models. Its technical effects are reflected in the fact that the charging behavior of robot swarms is no longer a passive response based on static rules, but rather an active planning based on the geometric risk assessment of future states. This can significantly reduce the rate of unexpected downtime due to battery depletion, optimize the utilization rate of charging resources, and ultimately improve the operational continuity, stability, and overall operational efficiency of the entire robot swarm.

[0046] Accordingly, in step S430, the charging decision result is generated based on the comparison between the charging urgency score and the charging urgency threshold. It should be understood that since the calculated charging urgency score is a continuous quantitative indicator, it does not provide a clear, directly executable action instruction. The decision-making system needs a clear triggering standard to transform the quantitative assessment into a concrete action. Therefore, in the technical solution of this application, the charging decision result is further generated based on the comparison between the charging urgency score and the charging urgency threshold. This transforms the continuous, quantitative urgency assessment value into a discrete, binary decision instruction, providing a final, deterministic execution trigger for the entire intelligent scheduling system. This allows complex, multi-dimensional state assessments to ultimately converge into a clear, executable scheduling action, thereby ensuring the integrity of the decision-making closed loop and the reliability of automated execution.

[0047] Specifically, in a specific example of this application, the process first obtains the final charging urgency score from the aforementioned calculation steps. Simultaneously, it reads a pre-set charging urgency threshold, adjustable by the system administrator, from its internal configuration parameters. Then, a conditional judgment operation is performed, directly comparing the obtained charging urgency score with the charging urgency threshold. Based on the comparison result, a final charging decision is generated. If the charging urgency score is greater than or equal to the threshold, a positive charging decision is generated, which is structurally identified as recommending immediate execution of the charging task. Conversely, if the charging urgency score is less than the threshold, a negative charging decision is generated, indicating that the robot should continue executing the current or pending task. The generated charging decision is then output to the upper-level scheduling module of the scheduling management server for execution.

[0048] Specifically, in step S500, based on the charging decision result and task queue data, it is determined whether to generate a charging task instruction for the first robot. It should be understood that since the charging decision result itself is only a logical judgment signal, it does not include consideration of the current task execution status of the first robot. If it is directly converted into an instruction without verification, it may interrupt high-priority or uninterruptible critical business processes that are currently being executed. Therefore, in the technical solution of this application, it is further determined whether to generate a charging task instruction for the first robot based on the charging decision result and task queue data. This allows for a final arbitration and coordination of the charging demand and the priority of the current business execution before the final instruction is issued. This ensures that the charging behavior is intelligently and non-destructively integrated into the robot's workflow, avoiding critical task interruptions caused by forced charging, thereby ensuring the continuity and stability of the entire cluster operation.

[0049] More specifically, in a concrete example of this application, the process first receives and parses the charging decision result. If the result is negative, no operation is performed within the current decision cycle, and the process terminates. If the result is positive, the task queue data is further accessed to retrieve and analyze the status attributes of the currently executing task assigned to the first robot. This is done by checking whether the task contains a flag indicating non-interruptible execution. If the first robot currently has no task executing, or its current task's flag allows interruption, it is determined that a charging task instruction can be generated. Otherwise, if the current task is a non-interruptible task, the generation of this charging instruction will be suspended until the task is completed. After determining that an instruction can be generated, a new charging task instruction with the highest execution priority is created and inserted at the head of the first robot's task queue to ensure that it is executed immediately.

[0050] In summary, the intelligent scheduling and management method for robot swarms according to the embodiments of this application is explained. It accurately predicts the future energy consumption curve of the robots by integrating the robots' assigned tasks and map data; simultaneously, it determines the available charging windows by performing time-series predictions of charging station resources. Based on these two predictions, the decision engine further combines the current task queue's busyness level to quantitatively assess the energy risk and opportunity cost of the robots charging, and comprehensively calculates a dynamic charging urgency. This method transforms charging behavior from an isolated emergency event into a well-thought-out planning step integrated into the overall system task flow, effectively avoiding charging station congestion and robot downtime caused by a lack of foresight. While ensuring robot energy safety, it maximizes the operational efficiency of the entire swarm.

[0051] Furthermore, an intelligent scheduling and management system for robot swarms is also provided.

[0052] Figure 6 This is a block diagram of an intelligent scheduling and management system for a robot swarm according to an embodiment of this application. Figure 6As shown, the intelligent scheduling and management system 100 for a robot cluster according to an embodiment of this application includes: a data acquisition module 110, used to acquire first robot status data, charging station status data set, task queue data, and map data; an energy curve prediction module 120, used to predict the future energy curve of the first robot based on the map data and the first robot status data to obtain the future energy curve of the first robot; a charging station availability prediction module 130, used to predict the future availability of charging stations based on the charging station status data set to obtain a charging station availability timeline set; a charging decision module 140, used to input the task queue data, the charging station availability timeline set, and the future energy curve of the first robot into a charging decision engine to obtain a charging decision result; and a charging task generation module 150, used to determine whether to generate a charging task instruction for the first robot based on the charging decision result and the task queue data.

[0053] As described above, the intelligent scheduling and management system 100 for robot swarms according to embodiments of this application can be implemented in various wireless terminals, such as servers with intelligent scheduling and management algorithms for robot swarms. In one possible implementation, the intelligent scheduling and management system 100 for robot swarms according to embodiments of this application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the intelligent scheduling and management system 100 for robot swarms can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the intelligent scheduling and management system 100 for robot swarms can also be one of many hardware modules of the wireless terminal.

[0054] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for intelligent scheduling management of a robot cluster, characterized in that, The method comprises: obtaining first robot state data, a charging station state data set, task queue data, and map data; performing future energy curve prediction on the first robot based on the map data and the first robot state data to obtain a first robot future energy curve; performing future charging station availability prediction based on the charging station state data set to obtain a charging station availability timeline set; inputting the task queue data, the charging station availability timeline set, and the first robot future energy curve into a charging decision engine to obtain a charging decision result; determining whether to generate a charging task instruction for the first robot based on the charging decision result and the task queue data. 2.The intelligent scheduling management method of the robot cluster according to claim 1, characterized in that, The method of performing future energy curve prediction on the first robot based on the map data and the first robot state data to obtain a first robot future energy curve comprises: extracting assigned task data from the task queue data; performing task energy consumption parameter deconstruction and quantization on the first robot state data based on the assigned task data and the map data to obtain quantized energy consumption parameters; performing segmented energy consumption and time consumption calculation based on the quantized energy consumption parameters to obtain a segmented consumption profile; performing future energy curve synthesis on the current power, the current timestamp, and the segmented consumption profile in the first robot state data to obtain the first robot future energy curve. 3.The intelligent scheduling management method of the robot cluster according to claim 2, characterized in that, The quantized energy consumption parameters comprise a planned path, a travel distance, and a task energy consumption. 4.The intelligent scheduling management method of the robot cluster according to claim 3, characterized in that, The method of performing segmented energy consumption and time consumption calculation based on the quantized energy consumption parameters to obtain a segmented consumption profile comprises: inputting the travel distance in the quantized energy consumption parameters into a robot energy consumption model and a robot kinematics model to obtain travel time and travel energy consumption; extracting an estimated execution time associated with the quantized energy consumption parameters from the assigned task data; performing data encapsulation on the travel time, the travel energy consumption, the estimated execution time, and the task energy consumption to obtain the segmented consumption profile.

5. The intelligent scheduling management method of robot swarm according to claim 1, wherein, The method of inputting the task queue data, the charging station availability timeline set, and the first robot future energy curve into a charging decision engine to obtain a charging decision result comprises: based on the task queue data, the charging station availability timeline set, and the first robot future energy curve, calculating an energy risk score, an opportunity cost score, and a charging accessibility score to obtain quantized decision factors; based on the quantized decision factors and a base weight vector, calculating a charging urgency score; based on a comparison between the charging urgency score and a charging urgency threshold, generating the charging decision result.

6. The intelligent scheduling management method of the robot cluster according to claim 5, wherein, The method of calculating a charging urgency score based on quantized decision factors and a base weight vector comprises: based on the quantized decision factors and the base weight vector, constructing a state metric tensor, the state metric tensor comprising a coupling term between the energy risk score and the charging accessibility score; based on the state metric tensor and the quantized decision factors, calculating the charging urgency score.

7. The intelligent scheduling management method of robot swarm according to claim 6, wherein, The method of constructing a state metric tensor based on quantized decision factors and a base weight vector comprises: based on the quantized decision factors and the base weight vector, constructing a state metric tensor in accordance with the following formula: ； ； wherein, and are coupling terms between the energy risk score and the charging accessibility score, is a state metric tensor, , and are preset weight values for the energy risk score, the opportunity cost score, and the charging accessibility score in the base weight vector, respectively, and are the energy risk score and the charging accessibility score, and are preset trainable coupling parameters. 8.The intelligent scheduling management method of the robot cluster according to claim 7, characterized in that, The charging urgency score is calculated based on the state metric tensor and the quantized decision factor, including: the charging urgency score is calculated based on the state metric tensor and the quantized decision factor according to the following formula: ； wherein, is a vector of quantification decision factors, is a state metric tensor, is a charge urgency score.

9. An intelligent scheduling and management system for a robot swarm, characterized in that, including: The data acquisition module is configured to acquire first robot state data, a charging station state data set, task queue data, and map data. The energy curve prediction module is configured to perform future energy curve prediction on the first robot based on the map data and the first robot state data to obtain a first robot future energy curve. The charging station availability prediction module is configured to perform future charging station availability prediction based on the charging station state data set to obtain a charging station availability timeline set. The charging decision module is configured to input the task queue data, the charging station availability timeline set, and the first robot future energy curve into a charging decision engine to obtain a charging decision result. The charging task generation module is configured to determine whether to generate a charging task instruction for the first robot based on the charging decision result and the task queue data.

Images