A robot dynamic scheduling control method and device

Abstract

The present disclosure provides a robot dynamic scheduling control method and device, and relates to the technical field of robots. The method comprises the following steps: obtaining a task template pool comprising a plurality of task templates and a capability module pool comprising a plurality of robot capability models; performing grouping processing on all single robots in the capability module pool, and determining a combined capability vector corresponding to each robot group; comparing an environment constraint vector and a capability demand vector with the combined capability vector of each robot group respectively, determining a candidate robot set based on the comparison result; calculating a loss expectation vector and a comprehensive matching score of each candidate object in the candidate robot set, and determining a target execution object; performing deduction processing on the job capability vector of the target execution object according to the rated energy consumption and the rated load of the target execution object, and updating the capability module pool and the job capability vector. The present disclosure can improve the collaborative success rate of heterogeneous robot jobs.

Description

Technical Field

[0001] This disclosure relates to the field of robotics, and in particular to a method and apparatus for dynamic scheduling and control of robots. Background Technology

[0002] With the rapid development of automation and intelligent technologies, multi-robot collaborative systems have been widely used in complex industrial operation scenarios. Existing robot scheduling technologies are mostly geared towards single tasks or simple scenarios, mainly relying on static configuration methods that directly bind tasks to specific individual robots, or matching tasks with robots according to preset fixed rules when a task request arrives.

[0003] However, the above scheduling method is based on the premise that the tasks at the work site are usually independent, functionally simple, and have clear boundaries. When the system is deployed to complex scenarios, a single task often involves multiple robots with strong time-series coordination and different hardware functional requirements to participate in the execution. Existing technology cannot effectively support the need for dynamic matching between robots and tasks under such complex and heterogeneous collaboration. In actual dispatch and execution, problems such as asynchronous actions of multiple machines and misaligned timing of transportation resources are very likely to occur, ultimately causing the entire multi-machine collaborative operation chain to be forcibly interrupted and the on-site task to be unable to be completed. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this disclosure provides a robot dynamic scheduling and control method and device, which can avoid the problem of interruption in multi-robot collaborative operations in complex scenarios, thereby improving the success rate of collaborative operations of heterogeneous robots.

[0005] According to a first aspect of the present disclosure, a robot dynamic scheduling and control method is provided, the method comprising: Obtain a task template pool containing multiple task templates and a capability module pool containing multiple robot capability models. Match the corresponding target task template from the task template pool according to the target task, and determine the environmental constraint vector, capability requirement vector, cooperative constraint vector and loss expectation vector corresponding to the target task. Based on the environmental constraint vector and the capability requirement vector, all individual robots in the capability module pool are screened to obtain a candidate robot set; Based on the cooperative constraint vector, the individual robots in the candidate robot set are dynamically grouped to obtain multi-body heterogeneous robot groups and independent individual robot groups, and the combination capability vector corresponding to each robot group is determined. The multi-body heterogeneous robot group includes multiple individual robots, and the independent individual robot group includes one individual robot. Calculate the comprehensive matching score between the expected loss vector and the combined capability vector of each robot group, and determine the target execution object from each robot group based on the comprehensive matching score, wherein the expected loss vector includes the expected preference value, robot health index and energy reserve index; The target task is dispatched to the target execution object, and the operation capability vector of each individual robot corresponding to the target execution object is deducted according to the rated energy consumption and rated load of the target execution object. After the target task is completed, the capability module pool is updated.

[0006] In one exemplary embodiment of this disclosure, the step of matching a corresponding target task template from the task template pool according to the target task, and determining the environmental constraint vector, capability requirement vector, collaborative constraint vector, and loss expectation vector corresponding to the target task, includes: Based on the requirement parameters of each task template in the task template pool, the target task is matched to obtain the target task template; Obtain the collaborative constraint parameters, environmental constraint parameters, and capability requirement parameters corresponding to the target task; The multi-body cooperative mode in the cooperative constraint parameters is subjected to feature mapping processing with the execution path to obtain the cooperative constraint vector; The environmental constraint parameters are numericalized and normalized to obtain the environmental constraint vector. The capability requirement parameters are divided into a first capability set and a second capability set. The capability requirement vector is generated based on the first capability set, and the loss expectation vector is constructed based on the robot health index, energy reserve index and expected preference value in the second capability set. The first capability set is used to characterize the rigid capability constraints corresponding to the robot completing the target task, and the second capability set is used to characterize the optimization constraint index corresponding to the robot completing the target task.

[0007] In one exemplary embodiment of this disclosure, the step of dynamically grouping individual robots in the candidate robot set according to the cooperative constraint vector to obtain multi-body heterogeneous robot groups and independent individual robot groups, and determining the combined capability vector corresponding to each robot group, includes: Based on the collaborative constraint vector corresponding to the target task, a collaborative matching relationship is determined, and multiple individual robots that satisfy the collaborative matching relationship are selected from the candidate robot set, and a multi-body heterogeneous robot group is constructed. Individual robots that do not participate in the combination are identified as independent individual robot groups; The static capability vector and operational capability vector of each individual robot in each robot group are aggregated and calculated to obtain the combined capability vector of each robot group.

[0008] In one exemplary embodiment of this disclosure, the step of aggregating and calculating the static capability vector and operational capability vector of each individual robot in each robot group to obtain the combined capability vector corresponding to each robot group includes: For each robot group, the capability dimension components of the robot group in the preset capability dimension are determined based on the static capability vector and operational capability vector of each individual robot. For each capability dimension component, based on the type of the capability dimension, the corresponding target mapping function is matched and aggregated to obtain the corresponding capability dimension value; The values ​​of each capability dimension are aggregated to obtain the combined capability vector of the robot group. Specifically, for the first type of capability dimension, the capability dimension components are summed according to the corresponding linear accumulation function; for the second type of capability dimension, the maximum or minimum value of the capability dimension components is obtained according to the corresponding nonlinear extremum function; for the third type of capability dimension, the determination result of whether there is at least one individual robot in the robot group whose capability dimension component satisfies the corresponding task threshold is determined according to the corresponding logic function.

[0009] In one exemplary embodiment of this disclosure, the step of filtering all individual robots in the capability module pool based on the environmental constraint vector and the capability requirement vector to obtain a candidate robot set includes: Obtain the environmental adaptation threshold in the environmental constraint vector and the rigid capability threshold in the capability requirement vector; Traverse the capability module pool and determine the candidate robot set as all individual robots whose static capability components are greater than or equal to the environmental adaptation threshold and whose remaining operational capability components are greater than or equal to the rigid capability threshold.

[0010] In one exemplary embodiment of this disclosure, the step of calculating a comprehensive matching score between the loss expectation vector and the combined capability vector of each of the robot groups, and determining the target execution object from each of the robot groups based on the comprehensive matching score, includes: For each candidate object in each of the robot groups, obtain the values ​​of each indicator in the second capability set of the candidate object; Based on the preset scoring weight coefficients, the values ​​of each indicator are weighted and fused with the corresponding expected preference values ​​in the loss expectation vector to obtain the comprehensive matching score of the candidate object. The candidate with the highest comprehensive matching score is determined as the target execution object.

[0011] In one exemplary embodiment of this disclosure, the method further includes: Real-time monitoring of the operating parameters of each target execution object, and calculation of the task execution deviation between the operating parameters and the target task requirements; If the task execution deviation exceeds a preset abnormal threshold, the candidate robot set is dynamically rescheduled based on the task requirement vector of the remaining tasks in the target task.

[0012] In one exemplary embodiment of this disclosure, the method further includes: Obtain the preset template inheritance rules in the task template pool, and determine the inheritance relationship between the sub-task template and the parent task template according to the template inheritance rules, wherein the sub-task template inherits the environmental constraint vector, capability requirement vector, collaborative constraint vector and loss expectation vector corresponding to the parent task template; When performing a task for a new scenario, the inherited vectors are incrementally modified based on the environmental feedback information of the new scenario, and the modified task template is registered as a new subtask template to the task template pool. The task templates in the task template pool are managed in a versioned manner. When multiple task templates are associated with the same job type, the task templates are iterated and their parameters are optimized based on the execution effect of historical scheduling data.

[0013] According to a second aspect of the present disclosure, a robot dynamic scheduling and control device is provided, comprising: The acquisition module is used to acquire a task template pool containing multiple task templates and a capability module pool containing multiple robot capability models. Based on the target task, it matches the corresponding target task template from the task template pool and determines the environmental constraint vector, capability requirement vector, cooperative constraint vector and loss expectation vector corresponding to the target task. The filtering module is used to filter all individual robots in the capability module pool according to the environmental constraint vector and the capability requirement vector to obtain a candidate robot set. The grouping module is used to dynamically group the individual robots in the candidate robot set according to the cooperative constraint vector to obtain multi-body heterogeneous robot groups and independent individual robot groups, and determine the combination capability vector corresponding to each robot group. The multi-body heterogeneous robot group includes multiple individual robots, and the independent individual robot group includes one individual robot. The determination module is used to calculate the comprehensive matching score between the loss expectation vector and the combined capability vector of each robot group, and to determine the target execution object from each robot group according to the comprehensive matching score, wherein the loss expectation vector includes the expectation preference value, the robot health index and the energy reserve index; The update module is used to dispatch the target task to the target execution object, and deduct the operation capability vector of each individual robot corresponding to the target execution object according to the rated energy consumption and rated load of the target execution object, and update the capability module pool after the target task is completed.

[0014] According to a third aspect of the present disclosure, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described robot dynamic scheduling and control method.

[0015] This disclosure provides a method for dynamic scheduling and control of robots. By acquiring a task template pool containing multiple task templates and a capability module pool containing multiple robot capability models, and determining the corresponding environmental constraint vector, capability requirement vector, cooperative constraint vector, and loss expectation vector based on matching the target task with the target task template, the method decouples task requirements from robot hardware capabilities. Therefore, compared to the traditional method of statically binding robots to specific scenarios or tasks, this method supports flexible derivation and dynamic integration of business solutions.

[0016] Furthermore, by using environmental constraint vectors and capability requirement vectors to filter all individual robots in the capability module pool to obtain a candidate robot set, rigid hard constraints such as environmental adaptability level are applied at the forefront of scheduling control for filtering. Thus, before entering complex heterogeneous combination computation, unqualified or non-idle individual devices are preferentially eliminated, achieving dimensionality reduction in algorithm processing scale, reducing the computational overhead of the control system, and ensuring the safety and reliability of the final execution scheme in complex scenarios.

[0017] Furthermore, by dynamically grouping individual robots within the candidate robot set based on the collaborative constraint vector, and determining the combined capability vectors for multi-unit heterogeneous robot groups and independent individual robot groups respectively, a dynamic parameterized modeling mechanism for multi-unit collaboration is provided for robots with different shapes and functional characteristics. This overcomes the limitations of traditional scheduling, which is confined to isomorphic or individual state comparisons. It enables heterogeneous robots to dynamically assemble and quantitatively evaluate their capabilities through preset combination rules, eliminating the risk of control system disorder such as multi-machine action desynchronization and capacity timing mismatch, and improving the success rate of collaboration.

[0018] Furthermore, by using a loss expectation vector that includes expected preference values, robot health indicators, and energy reserve indicators, and combining the combined capability vectors of each robot group for comprehensive matching and scoring, the target execution object is determined. Under the premise of ensuring the rigidity of the execution subject, excessive equipment wear caused by relying solely on spatial distance or simple availability ranking is avoided, thus achieving a globally optimal scheduling decision in terms of overall cluster health, service life balance, and energy consumption.

[0019] Finally, target tasks are dispatched to the target execution objects. Based on their corresponding rated energy consumption and rated load, these tasks are mapped to the operational capability vectors of each individual robot, and corresponding deduction processing is performed. After the task is completed, the capability module pool is updated in a closed loop, constructing a closed-loop feedback control flow. This ensures that the robot capability data in the scheduling backend is aligned with actual physical consumption in real time and with high accuracy, effectively preventing hardware resource conflicts and overload during multi-task concurrency. Even when facing operational deviations such as equipment failures or sudden anomalies, dynamic rescheduling can still be triggered based on the latest available resources, guaranteeing the continuity and collaborative success rate of highly dynamic and complex work environments.

[0020] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0021] The accompanying drawings, which are incorporated in and form part of this disclosure, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0022] Figure 1 This is a flowchart illustrating a robot dynamic scheduling and control method according to an exemplary embodiment of the present disclosure.

[0023] Figure 2 This is a schematic diagram of the structure of a robot dynamic scheduling and control device according to an exemplary embodiment of the present disclosure.

[0024] Figure 3 This is a hardware structure diagram of a computer device shown in an embodiment of this disclosure. Detailed Implementation

[0025] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0026] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0027] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are used only to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0028] The embodiments of this disclosure will now be described in detail.

[0029] like Figure 1 As shown, Figure 1 This disclosure is a flowchart illustrating a robot dynamic scheduling and control method according to an exemplary embodiment, comprising the following steps: Step 101: Obtain a task template pool containing multiple task templates and a capability module pool containing multiple robot capability models. Match the corresponding target task template from the task template pool according to the target task, and determine the environmental constraint vector, capability requirement vector, collaborative constraint vector and loss expectation vector corresponding to the target task.

[0030] Step 102: Based on the environmental constraint vector and the capability requirement vector, all individual robots in the capability module pool are screened to obtain a candidate robot set.

[0031] Step 103: Based on the collaborative constraint vector, the individual robots in the candidate robot set are dynamically grouped to obtain multi-body heterogeneous robot groups and independent individual robot groups, and the combined capability vector corresponding to each robot group is determined.

[0032] Step 104: Calculate the comprehensive matching score between the expected loss vector and the combined capability vector of each robot group, and determine the target execution object from each robot group based on the comprehensive matching score. The expected loss vector includes the expected preference value, robot health index and energy reserve index.

[0033] Step 105: Dispatch the target task to the target execution object, and deduct the operation capability vector of each individual robot corresponding to the target execution object according to the rated energy consumption and rated load of the target execution object. After the target task is completed, update the capability module pool.

[0034] The robot dynamic scheduling and control method provided in this exemplary embodiment obtains a task template pool containing multiple task templates and a capability module pool containing multiple robot capability models. It then matches the corresponding target task template to the target task to determine the environmental constraint vector, capability requirement vector, cooperative constraint vector, and loss expectation vector, thereby achieving pooled decoupling between task requirement description and robot hardware capability model. This breaks the static binding limitation between tasks and specific hardware in traditional scheduling.

[0035] Furthermore, by screening all individual robots according to environmental constraint vectors and capability requirement vectors to obtain a candidate robot set, strong constraints such as environmental adaptability and rigidity capability are applied to the forefront of scheduling and control for filtering. This allows unqualified entities to be eliminated before entering complex heterogeneous combination calculations, reducing the computing power overhead of the control system, avoiding the explosion of full exhaustive combination during large-scale cluster scheduling, and ensuring the safety and reliability of the execution scheme.

[0036] Furthermore, by dynamically grouping individual robots in the candidate robot set according to the collaborative constraint vector, and determining the combined capability vectors corresponding to the multi-body heterogeneous robot group and the independent single robot group, it is possible to adaptively assemble multiple heterogeneous units with complementary functions and spatial correlation characteristics into a unified robot group according to the strong temporal collaboration requirements of the target task in the multi-machine collaborative mode. This overcomes the limitations of traditional scheduling, which is limited to isomorphic or single-unit state comparison. It enables heterogeneous robots to achieve dynamic combination and quantitative evaluation of capabilities through preset combination rules, eliminates the risk of control system disorder such as multi-machine action desynchronization or capacity timing mismatch, and improves the collaborative success rate of complex operation chains.

[0037] Furthermore, by calculating the comprehensive matching score between the loss expectation vector (which includes expected preference value, robot health index, and energy reserve index) and each combined capability vector, and based on this, the target execution object is determined from each robot group. Under the rigid premise of ensuring the availability of the execution subject, a multi-index refined comparison and selection that conforms to the actual industrial site preferences is introduced, avoiding excessive equipment wear and achieving overall global optimal scheduling decision.

[0038] Finally, by dispatching target tasks to the target execution objects and deducting the operational capability vectors of each individual robot based on rated energy consumption and rated load, and updating the capability module pool after task completion, a closed-loop feedback control flow is constructed. This ensures that the robot dynamic capability data in the scheduling backend is aligned in real time and with high precision with the actual physical operation consumption, avoiding the core transport resources of the same robot being disorderly occupied by multiple tasks in a short period of time, which would cause capacity contention and timing misalignment. It also reduces execution interruption failures caused by overload or mid-process power depletion of heterogeneous robots during execution, and maximizes the guarantee of the final successful completion of the on-site task.

[0039] The robot dynamic scheduling and control method in this example embodiment will now be described in detail.

[0040] In step 101, a task template pool containing multiple task templates and a capability module pool containing multiple robot capability models are obtained. The corresponding target task template is matched from the task template pool according to the target task, and the environmental constraint vector, capability requirement vector, collaborative constraint vector and loss expectation vector corresponding to the target task are determined.

[0041] In the exemplary implementation of this disclosure, the task template pool contains multiple parameterized task templates pre-arranged and registered for different typical work scenarios. These task templates define basic requirement parameters such as task type and task area, and also encapsulate requirement parameter features such as multi-body collaboration mode, execution path, environmental constraint parameters (such as explosion-proof level and protection level), capability requirement parameters (such as rated load and movement speed), and expected loss parameters (such as health and power). Through this pooled management, when facing new and sporadic task scenarios, the template matching mechanism can be used to directly retrieve or reuse the corresponding task template, thereby avoiding the redundancy of building task descriptions from scratch and improving the deployment and response timeliness of task scheduling.

[0042] Meanwhile, the capability module pool contains capability models that independently correspond to each individual robot. Each individual robot's capability model is mapped to a unified global capability dimension space. Furthermore, this capability model is structurally divided into static capability vectors and operational capability vectors. The static capability vectors represent the robot's fixed factory hardware parameters or rated attribute configurations, such as maximum climbing ability, factory explosion-proof rating, rated load capacity, and whether it possesses specific mounting tools. The operational capability vectors represent dynamic resource attributes that change in real-time with operational time, environment, or task assignment, such as the robot's current real-time location, current remaining battery percentage, health status, and currently unoccupied available remaining load capacity.

[0043] In the exemplary implementation of this disclosure, a process can be described as determining the collaborative constraint vector, environmental constraint vector, capability requirement vector, and loss expectation vector corresponding to the target task based on the requirement parameters of each task template in the task template pool. The specific implementation process may include the following steps.

[0044] First, based on the requirement parameters of each task template in the task template pool, a specific target task is matched to obtain the corresponding target task template.

[0045] For example, multimodal task information (such as text, action or semantic information) corresponding to the target task is obtained, and word segmentation and feature extraction are performed on the multimodal task input information to obtain multimodal task features; based on the multimodal task features, traversal and classification retrieval are performed in the task template pool to match and obtain the target task template corresponding to the target task.

[0046] Secondly, obtain the collaborative constraint parameters, environmental constraint parameters, and capability requirement parameters corresponding to the target task.

[0047] For example, after obtaining the target task template, in order to transform the operational requirements of the target task into indicators that the control algorithm can recognize, the target task template can be invoked, and the specific operational parameters input by the user for the target task (such as task subclass, station sequence, cargo information, priority, and execution count) can be obtained. By filling these specific operational parameters into the target task template, a target task instance to be dispatched is generated, and from this target task instance, the collaborative constraint parameters, environmental constraint parameters, and capability requirement parameters that meet the actual execution requirements of the task can be obtained.

[0048] For example, after obtaining the station sequence entered by the user, the path is calculated by retrieving the pre-stored electronic map topology network. If it is detected that the distance from the starting point to the end point is long and there may be communication blind spots due to terrain obstruction, it is determined that the execution of the target task requires multi-machine time-series coordination. Thus, the multi-body collaboration mode in the target task instance is determined and assigned as a relay transportation mode, and the preset execution path is determined based on each station in the station sequence, thereby determining the collaboration constraint parameters of the target task instance.

[0049] After obtaining the task subclass filled in by the user, the physical explosion-proof level requirement of the target task instance is determined by searching the security protection level comparison table; at the same time, according to the operating environment through which the site sequence passes, the corresponding protection level requirement is retrieved from the gateway configuration file, and the environmental constraint parameters of the target task instance are determined accordingly.

[0050] After obtaining the cargo information entered by the user, the rated load capacity threshold of the target task instance is determined based on the total loaded weight of the cargo. Simultaneously, based on the cargo information's remarks, a material tool association table is retrieved to determine the robot's corresponding upper equipment requirements (such as a specific explosion-proof gripping robotic arm). After obtaining the user-entered priority, a timeliness interval is determined based on a priority-to-preset timeliness level comparison table. Then, based on the distance span from the starting point to the ending point determined by the station sequence, the maximum moving speed and corresponding horsepower output requirement required for the target task instance are determined, thus constructing the capability requirement parameters characterizing rigid constraints. Furthermore, the task optimization direction label from the target task template can be obtained and assigned as the expected preference value of the target task instance. The robot health index and energy reserve index, non-mandatory optimization dimensions, are added to the capability requirement parameters of the target task instance, forming the loss expectation parameter. This expected preference value serves as the weighted calculation coefficient for each optimization index item during subsequent multi-objective scoring, while the loss expectation parameter is used to perform multi-objective comprehensive scoring optimization on candidate objects while meeting rigid requirements. This determines the capability requirement parameters characterizing soft constraints.

[0051] By using the task subclasses, station sequences, cargo information, priorities, and execution counts based on user input, the target operation requirements are transformed into collaborative constraint parameters, environmental constraint parameters, and capacity requirement parameters contained in the current target task instance, which can be directly read and solved by the scheduling algorithm.

[0052] Then, the obtained collaborative constraint parameters, environmental constraint parameters, and capability requirement parameters are converted into corresponding parameter vectors.

[0053] For example, collaborative constraint parameters can be extracted from the target task instance, and the multi-body collaborative modes (such as relay or formation modes) and preset execution paths can be processed by feature mapping to obtain a collaborative constraint vector. This collaborative constraint vector is used as a quantization parameter on the spatial and temporal axes to characterize the multi-robot collaborative form required to complete the target task.

[0054] It can also extract environmental constraint parameters from the target task instance (such as the explosion-proof rating, waterproof / temperature and humidity protection rating of the work surface, etc.), and perform numerical and normalization processing to obtain an environmental constraint vector. Through this environmental constraint vector, discrete labels (such as explosion-proof rating) are transformed into values ​​in a unified numerical domain.

[0055] It is also possible to extract capability requirement parameters from target task instances and divide these parameters into a first capability set and a second capability set. The first capability set characterizes the rigid capability constraints required for the robot to complete the target task (i.e., the hardware foundation the robot must possess, such as the presence or absence of a robotic arm, the rated maximum carrying capacity boundary, etc.). Based on the rigid indicators and thresholds in the first capability set, normalization is performed to generate a capability requirement vector. The second capability set characterizes the optimization constraint indicators required for the robot to complete the target task (i.e., the soft optimization features affecting the long-term effectiveness and economy of the operation). Based on the robot health indicators, energy reserve indicators (such as real-time remaining battery percentage), and expected preference values ​​in the second capability set, a fusion and quantification process can be performed to obtain a loss expectation vector, which is used to perform multi-objective comprehensive scoring optimization on candidate objects while meeting rigid requirements.

[0056] By outputting four standardized vectors representing the environment, capabilities, coordination, and losses of the target task through step-by-step output, the separation between task description and specific hardware implementation is achieved. This allows the robot's main controller to directly drive subsequent initial screening and dynamic grouping logic using the partitioned vector data during scheduling and control, improving its scalability and scheduling capabilities in complex industrial environments.

[0057] In step 102, all individual robots in the capability module pool are screened based on the environmental constraint vector and the capability requirement vector to obtain a candidate robot set.

[0058] In the exemplary implementation of this disclosure, environmental adaptation thresholds (such as minimum satisfaction thresholds for explosion-proof rating, protection rating, etc.) can be extracted from the environmental constraint vector, and rigid capability thresholds (such as minimum load requirement components, minimum accuracy requirement components, etc.) required to complete the target task can be extracted from the capability requirement vector. Subsequently, multi-dimensional logical judgments are performed on the static capability vectors and dynamic capability vectors of all currently registered individual robots in the capability module pool.

[0059] For example, the judgment criteria include the following three items: First, determine whether the current state of the individual robot is idle, directly eliminating non-idle devices that are currently operating, experiencing abnormal malfunctions, or undergoing charging and maintenance. Second, determine whether the environmental adaptation component and static capability component in the static capability vector of the individual robot are respectively greater than or equal to the corresponding environmental adaptation constraints and static capability requirement thresholds. For example, verify whether the physical explosion-proof rating of the individual robot meets the mandatory capability threshold required by the task, and whether its factory-installed fixed configuration is equipped with the specific upper-mount equipment or specific tools specified by the target task. Third, determine whether the current remaining working capacity component in the dynamic capability vector of the individual robot is greater than or equal to the rigid dynamic resource threshold required for task execution. For example, verify whether the current remaining load capacity component or current battery percentage component of the individual robot can completely cover the dynamic consumption baseline required for the execution of the target task.

[0060] All individual robots that simultaneously meet the aforementioned idle state, static capability requirements, and dynamic resource thresholds are retained and uniformly included in the candidate robot set. By constructing this candidate robot set, subsequent heterogeneous multi-robot combination calculations are only performed within the local intersection of these initially qualified robots. This avoids the combinatorial explosion and computational disaster caused by indiscriminate exhaustive combination of the entire heterogeneous cluster, compresses the matching optimization space, and ensures the response speed and execution security of complex industrial site scheduling and control.

[0061] In step 103, based on the collaborative constraint vector, the individual robots in the candidate robot set are dynamically grouped to obtain multi-body heterogeneous robot groups and independent individual robot groups, and the combined capability vector corresponding to each robot group is determined.

[0062] In the example implementation of this disclosure, the cooperative matching relationship is first determined according to the cooperative constraint vector corresponding to the target task, and multiple individual robots that meet the cooperative matching relationship are selected from the candidate robot set, and a multi-body heterogeneous robot group is constructed.

[0063] For example, the collaborative constraint vector declares the collaborative pattern required to complete the target task, such as a relay transportation mode or a multi-robot formation mode. Based on this pattern requirement, multiple spatially related or functionally complementary individual robots are logically bound together within the candidate robot set, thereby adaptively assembling multiple heterogeneous individuals into a unified multi-robot heterogeneous robot group. Simultaneously, the remaining individual robots that do not participate in the combination are independently identified as independent individual robot groups, where each independent individual robot group contains only one individual robot.

[0064] Secondly, after constructing different robot groups (including multi-body heterogeneous robot groups and independent single-unit robot groups), the static capability vector and operational capability vector of each single robot in each robot group are aggregated and calculated to obtain the true combined capability vector of each robot group.

[0065] For example, for each robot group, the aggregation calculation process may include the following steps: Based on the static capability vectors and operational capability vectors of each individual robot in the group, determine each capability dimension component of the robot group within the preset capability dimension space. For example, extract the load component, movement speed component, and interface state component indicating whether a specific tool is available for individual robots A and B in the group.

[0066] For each capability dimension component, based on the type of the capability dimension, the corresponding target mapping function is matched for aggregation operation to obtain the corresponding capability dimension value. For the first type of capability dimension, this dimension belongs to the superimposed physical attribute (such as maximum transport load, energy storage capacity, etc.). For the first type of capability dimension, a corresponding linear summation function can be matched to sum the capability dimension components of all individual robots in the group, and the summation result is used as the capability dimension value of the robot group in this dimension. For the second type of capability dimension, which is a nonlinear extreme value physical attribute constrained by the bottleneck individual in the group (such as the maximum moving speed in multi-robot formation operations), a corresponding nonlinear extreme value function can be matched to obtain the maximum or minimum value of the capability dimension component of each individual robot in the group (for example, the overall movement speed of the formation depends on the minimum speed of the individual robots in the group), and this extreme value is used as the capability dimension value of the robot group in this dimension. For the third type of capability dimension, which is a Boolean, existence-based physical attribute such as tool or environmental adaptability (such as whether at least one robot in the group carries a specific robotic arm or working tool), a corresponding logic function can be matched to determine whether there is at least one individual robot in the robot group whose capability dimension component satisfies the corresponding task threshold, and output the corresponding judgment result (output 1 if it exists, output 0 if it does not exist), and use this judgment result as the capability dimension value of the robot group in this dimension.

[0067] The capability dimension values ​​obtained through the above differential aggregation operation are subjected to matrix aggregation processing to finally reconstruct and generate a unified and unique combined capability vector for the robot group.

[0068] This step addresses the distortion of combined capabilities in multi-machine heterogeneous clusters caused by individual differences in space and time through a classification-based cascaded mapping function aggregation method. This approach enables the acquisition of parameterized capability boundaries aligned with the actual physical representation before task dispatch, effectively preventing asynchronous actions and spatiotemporal divergence in path planning caused by speed differences, uneven load distribution, or incomplete tool coverage during subsequent actual dispatch and execution in multi-machine collaboration. This ensures the continuity of the overall operation control chain.

[0069] In step 104, a comprehensive matching score is calculated between the expected loss vector and the combined capability vector of each robot group, and the target execution object is determined from each robot group based on the comprehensive matching score. The expected loss vector includes the expected preference value, robot health index and energy reserve index.

[0070] In the exemplary implementation of this disclosure, after obtaining the combined capability vector of each robot group, each candidate object can be comprehensively scored by non-rigid soft optimization indicators, thereby dynamically selecting the optimal execution entity that takes into account both task suitability and cluster health.

[0071] Specifically, the process of calculating the comprehensive matching score between the expected loss vector and the combined capability vector of each robot group, and determining the target execution object from each robot group based on the comprehensive matching score, can be achieved in the following way.

[0072] First, for each candidate robot in each robot group, obtain the numerical values ​​of each indicator in the candidate robot's second capability set. The candidate robots here include the multi-body heterogeneous robot groups and independent single-unit robot groups identified in the previous steps.

[0073] For example, the real-time operational capability vector of each individual robot in the candidate group is obtained, and status parameters such as health, battery level, and current load, which dynamically fluctuate with the operation time, are extracted. Based on these parameters, the values ​​of various indicators of the candidate group under the second capability set dimension are calculated, such as the robot health index and energy reserve index of the robot group as a whole. Among them, the energy reserve index can be specifically represented as the actual percentage of remaining battery power or remaining endurance of each individual robot in the current group.

[0074] Secondly, after obtaining the values ​​of each indicator, based on the preset scoring weight coefficients, the values ​​of each indicator are weighted and fused with the corresponding expected preference values ​​in the loss expectation vector to obtain the comprehensive matching score of the candidate object.

[0075] For example, the actual values ​​of robot health indicators and energy reserve indicators of each candidate can be compared or the deviation can be calculated with the corresponding expected preference values ​​in the loss expectation vector, so that parameters of different physical dimensions are transformed into scoring components of a unified level. Based on the preset scoring weight coefficient, the scoring weight coefficient can be strategically set in advance according to the actual task type. For example, the weight coefficient of energy reserve is increased in emergency tasks, and the weight coefficient of health indicators is increased in routine operations. Each scoring component is multiplied and accumulated with the corresponding scoring weight coefficient and expected preference value to calculate a comprehensive matching score value for each candidate that can objectively represent its comprehensive loss cost and performance advantage.

[0076] In the specific calculations, different dimensions of the loss expectation vector correspond to preset preference control methods, which guide the performance evaluation of different types of capability indicators. To accurately reflect the dynamic preferences of complex industrial environments for different soft constraint indicators, the individual score of each capability component is calculated by weighting it according to the mapping functions of three expected preference values ​​for the corresponding dimension of the loss expectation vector. First, when the expected preference value corresponding to the corresponding dimension in the loss expectation vector is as high as possible, for example, for capability indicators such as robot health, movement speed and energy reserve, the first proportional mapping scoring function is matched and the boundary is truncated based on the preset maximum expected value. This ensures that the maximum expected value is output when the actual capability of the robot group reaches or exceeds the task expectation benchmark, while effectively limiting the unlimited amplification of the excess part to the total score, thus ensuring the numerical stability of the joint scoring of multiple indicators.

[0077] Second, when the expected preference value of the corresponding dimension in the loss expectation vector is as small as possible, such as for capability indicators such as equipment service life, long-term failure rate or depreciation, the second inverse mapping scoring function is matched and the boundary is truncated based on the preset minimum expected value. This second inverse mapping scoring function achieves negative incentive control of long-term operation and maintenance loss indicators by making the candidate with the lower actual value obtain a higher single score.

[0078] Third, when the expected preference value corresponding to the corresponding dimension in the loss expectation vector has no clear target value and only a preference for better, the current combined capability value of the robot group is determined as the corresponding individual score to ensure that the operator with better software and hardware performance can gain a competitive advantage.

[0079] Finally, after calculating the overall matching score of all candidate objects, the candidate object with the highest overall matching score is determined as the target execution object.

[0080] By introducing a weighted fusion scoring mechanism that balances multiple objectives such as expected preferences, equipment health, and energy reserves, the limitations of traditional scheduling, which relies solely on spatial distance or static parameters for single-objective matching, are overcome. This approach enables the scheduling control layer to identify which individual robots have better endurance redundancy and long-term hardware stability when faced with multiple robot groups that also meet rigid requirements, thereby achieving optimal allocation of global capacity at the control decision level.

[0081] In step 105, a target task is dispatched to the target execution object, and the operation capability vector of each individual robot corresponding to the target execution object is deducted according to the rated energy consumption and rated load of the target execution object. After the target task is completed, the capability module pool is updated.

[0082] In the example implementation of this disclosure, after the candidate with the highest comprehensive matching score is determined and identified as the target execution object, a target task instruction is formally issued to the target execution object to drive it to the designated industrial work site to perform production, transportation or rescue tasks.

[0083] Simultaneously with task assignment, if the target is a multi-unit heterogeneous robot group, the individual operational capability vectors of each robot in the capability module pool are locked and deducted in real time based on their respective rated energy consumption, rated load, and assigned sub-task type. If the target is an independent single robot, its own operational capability vector is locked and deducted directly. For the stackable capability dimension, the remaining load component of the corresponding individual robot is deducted based on the actual load allocated to the current task. For the resource dimension that is consumed linearly over time, the corresponding power or fuel percentage component is pre-deducted and locked based on the estimated execution time or execution path topology distance in the target task template, combined with the rated energy consumption parameters of the individual robot. For the equipment status and scheduling exclusivity dimension, the available status components of all individual robots participating in the current task are modified from idle status and locked to working status. This directly blocks the disorderly occupation of these hardware entities by other scheduling requests at the control front end, ensuring the synchronization and sequential continuity of the current multi-machine collaborative operation chain.

[0084] After the target task is fully completed, the exclusive lock on the relevant hardware resources is released, and the operation capability vector of each individual robot in the capability module pool is globally updated based on environmental feedback data and the actual power consumption, mechanical wear or operation deviation status of the equipment during actual operation.

[0085] This closed-loop control mechanism, which evaluates at the group level and deducts at the individual level, ensures the accuracy and real-time performance of dynamic data in the capability module pool. This allows for calculations based on the latest and most accurate physical hardware consumption status, even in complex, highly dynamic, and volatile industrial environments, guaranteeing the continuity of overall operations and control precision.

[0086] In some example implementations, this robot dynamic scheduling and control method also provides a real-time control closed loop and a rescheduling mechanism for occasional abnormal situations. Specifically, during the process of dispatching and executing target tasks to target execution objects, the main controller monitors the operating parameters of each target execution object in real time and calculates the task execution deviation between the operating parameters and the target task requirements.

[0087] Specifically, operational parameters refer to the real-time status data fed back by the target execution object during operation in physical space, including but not limited to the real-time three-dimensional spatial coordinates of a single robot, its current moving speed, the posture of the robotic arm's end effector, real-time remaining battery power, the nonlinear attenuation coefficient of the motor torque, and the current actual load status. These operational parameters are compared with the preset expected timing nodes, path trajectories, and functional output baselines in the target task template using difference calculations or topology matching to deduce task execution deviations. For example, deviations can manifest as position tracking errors at the spatial trajectory level, arrival time delays at the timing control level (such as mismatches due to highly dynamic terrain obstacles), or abnormal excessive power consumption at the physical resource level.

[0088] Subsequently, the calculated task execution deviation is quantitatively compared with a preset anomaly threshold. If the task execution deviation exceeds the preset anomaly threshold (e.g., single-machine position deviation continuously exceeds the preset sliding mode boundary, equipment gets stuck due to terrain causing desynchronization time to exceed the tolerance limit, or sudden hardware failure causes complete shutdown), it is determined that the current executing entity can no longer maintain the continuity and control accuracy of the overall operation. At this time, a dynamic rescheduling process is triggered: first, the critical state of the current site that has been damaged is locked, the remaining tasks that have not yet been completed in the target task are extracted, and the task requirement vector of the remaining tasks is re-derived. Then, based on this, the latest dynamic data of the capability module pool is retrieved again, the candidate robot set is re-determined, and dynamic rescheduling is performed. In the rescheduling calculation, equipment that has failed or has severely degraded performance is removed, and a new decision is made on whether to replace the executing entity or reorganize the formation. This embodiment, through a fully closed-loop deviation rescheduling mechanism, ensures that the entire collaborative operation chain will not be forcibly interrupted when faced with sudden anomalies or occasional failures in complex industrial special operation sites, thus ensuring the continuity of on-site operations.

[0089] In some example implementations, preset template inheritance rules can be obtained from the task template pool, and the inheritance relationship between sub-task templates and parent task templates can be determined according to these rules. Under this inheritance relationship, an inheritance topology relationship is established between the sub-task templates and parent task templates. Under this inheritance topology, when a sub-task template is statically registered, it automatically and completely inherits the basic data structure and constraint benchmarks of the environmental constraint vector, capability requirement vector, collaborative constraint vector, and loss expectation vector corresponding to the parent task template, thereby eliminating the computational redundancy and control configuration overhead of repeatedly modeling similar operating scenarios or variant operating conditions.

[0090] When executing tasks for entirely new scenarios, there's no need to modify the underlying scheduling engine. Instead, the main controller extracts incremental features from the environmental feedback information (such as a sudden change in the explosion-proof safety level at a specific work site) and performs localized, differentiated, online incremental modifications to each vector inherited from the parent template. For example, if the new scenario triggers higher protection indicators due to environmental corrosion or upgraded safety requirements, the scheduling engine only performs upward incremental corrections to the waterproof / explosion-proof level components in the inherited environmental constraint vector. If the work distance changes in the new scenario, adaptive incremental fine-tuning is performed on the execution path parameters in the collaborative constraint vector, the load capacity in the capacity requirement vector, and the energy consumption threshold in the capacity requirement vector. After modification, the modified task template is registered as a new sub-task template in the task template pool.

[0091] Furthermore, to ensure the long-term stability of the task template pool, version management can be implemented for the task templates within the pool. When multiple task templates are associated with the same job type, the task template is iterated and its parameters optimized based on the execution results of historical scheduling data. Specifically, the actual time consumption, multi-machine timing mismatch frequency, actual power consumption, and operational stability data of each individual robot in multiple historical scheduling processes are extracted, and the rationality of the task template is evaluated using a preset performance evaluation function. If it is found that a certain hard constraint filter is too lenient or the soft constraint weight allocation is unreasonable in the historical execution results, the feature vector parameters in the corresponding task template will be corrected, and an updated version number will be assigned to it (e.g., iterating from V1.0 to V1.1). Through incremental inheritance and dynamic version iteration of templates, the scheduling constraints caused by the fixed task descriptions of traditional scheduling technologies are avoided, enabling the scheduling process to have adaptive evolution capabilities and improving scheduling efficiency under long-cycle control flow.

[0092] Corresponding to the embodiments of the foregoing methods, this disclosure also provides embodiments of the apparatus and the terminal to which it is applied.

[0093] like Figure 2 As shown, Figure 2This is a schematic diagram of a robot dynamic scheduling and control device according to an exemplary embodiment of the present disclosure. The device includes: an acquisition module 210, a filtering module 220, a grouping module 230, a determination module 240, and an update module 250. The acquisition module 210 is used to acquire a task template pool containing multiple task templates and a capability module pool containing multiple robot capability models. It matches the corresponding target task template from the task template pool according to the target task and determines the environmental constraint vector, capability requirement vector, cooperative constraint vector and loss expectation vector corresponding to the target task. The screening module 220 is used to screen all individual robots in the capability module pool according to the environmental constraint vector and the capability requirement vector to obtain a set of candidate robots. Grouping module 230 is used to dynamically group individual robots in the candidate robot set according to the cooperative constraint vector to obtain multi-body heterogeneous robot group and independent individual robot group, and determine the combination capability vector corresponding to each robot group. The multi-body heterogeneous robot group includes multiple individual robots, and the independent individual robot group includes one individual robot. The determination module 240 is used to calculate the comprehensive matching score between the loss expectation vector and the combined capability vector of each robot group, and to determine the target execution object from each robot group based on the comprehensive matching score. The loss expectation vector includes the expectation preference value, robot health index and energy reserve index.

[0094] The update module 250 is used to dispatch target tasks to the target execution object, and deduct the operation capability vector of each individual robot corresponding to the target execution object according to the rated energy consumption and rated load of the target execution object. After the target task is completed, the capability module pool is updated.

[0095] Embodiments of the robot dynamic scheduling and control device disclosed herein can be applied to computer devices, such as servers or terminal devices. The device embodiments can be implemented through software, hardware, or a combination of both. Taking software implementation as an example, as a logical device, it is formed by the processor reading the corresponding computer program instructions from non-volatile memory into memory and executing them. From a hardware perspective, such as... Figure 3 The diagram shown is a hardware structure diagram of a computer device containing a robot dynamic scheduling and control device according to an embodiment of this disclosure. Except for... Figure 3 In addition to the processor 310, memory 330, network interface 320, and non-volatile memory 340 shown, the server or electronic device where the robot dynamic scheduling and control device is located in the embodiment may also include other hardware depending on the actual function of the computer device, which will not be described in detail here.

[0096] The specific implementation process of the functions and roles of each module in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0097] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0098] In one example embodiment of this disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the methods described above is stored. In some possible embodiments, various aspects of this disclosure may also be implemented as a program product including program code that, when the program product is run on a terminal device, causes the terminal device to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure.

[0099] The foregoing has described specific embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0100] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention applied herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not claimed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0101] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

[0102] The above description is merely a preferred embodiment of this disclosure and is not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for dynamic scheduling and control of a robot, characterized in that, include: Obtain a task template pool containing multiple task templates and a capability module pool containing multiple robot capability models. Match the corresponding target task template from the task template pool according to the target task, and determine the environmental constraint vector, capability requirement vector, cooperative constraint vector and loss expectation vector corresponding to the target task. Based on the environmental constraint vector and the capability requirement vector, all individual robots in the capability module pool are screened to obtain a candidate robot set; Based on the cooperative constraint vector, the individual robots in the candidate robot set are dynamically grouped to obtain multi-body heterogeneous robot groups and independent individual robot groups, and the combination capability vector corresponding to each robot group is determined. The multi-body heterogeneous robot group includes multiple individual robots, and the independent individual robot group includes one individual robot. Calculate the comprehensive matching score between the expected loss vector and the combined capability vector of each robot group, and determine the target execution object from each robot group based on the comprehensive matching score, wherein the expected loss vector includes the expected preference value, robot health index and energy reserve index; The target task is dispatched to the target execution object, and the operation capability vector of each individual robot corresponding to the target execution object is deducted according to the rated energy consumption and rated load of the target execution object. After the target task is completed, the capability module pool is updated.

2. The method according to claim 1, characterized in that, The step of matching a corresponding target task template from the task template pool based on the target task, and determining the environmental constraint vector, capability requirement vector, collaborative constraint vector, and loss expectation vector corresponding to the target task, includes: Based on the requirement parameters of each task template in the task template pool, the target task is matched to obtain the target task template; Obtain the collaborative constraint parameters, environmental constraint parameters, and capability requirement parameters corresponding to the target task; The multi-body cooperative mode in the cooperative constraint parameters is subjected to feature mapping processing with the execution path to obtain the cooperative constraint vector; The environmental constraint parameters are numericalized and normalized to obtain the environmental constraint vector. The capability requirement parameters are divided into a first capability set and a second capability set. The capability requirement vector is generated based on the first capability set, and the loss expectation vector is constructed based on the robot health index, energy reserve index and expected preference value in the second capability set. The first capability set is used to characterize the rigid capability constraints corresponding to the robot completing the target task, and the second capability set is used to characterize the optimization constraint index corresponding to the robot completing the target task.

3. The method according to claim 1, characterized in that, The step of dynamically grouping individual robots in the candidate robot set according to the cooperative constraint vector to obtain multi-body heterogeneous robot groups and independent individual robot groups, and determining the combined capability vector corresponding to each robot group, includes: Based on the collaborative constraint vector corresponding to the target task, a collaborative matching relationship is determined, and multiple individual robots that satisfy the collaborative matching relationship are selected from the candidate robot set, and a multi-body heterogeneous robot group is constructed. Individual robots that do not participate in the combination are identified as independent individual robot groups; The static capability vector and operational capability vector of each individual robot in each robot group are aggregated and calculated to obtain the combined capability vector of each robot group.

4. The method according to claim 3, characterized in that, The aggregation calculation of the static capability vector and operational capability vector of each individual robot in each robot group to obtain the combined capability vector corresponding to each robot group includes: For each robot group, the capability dimension components of the robot group in the preset capability dimension are determined based on the static capability vector and operational capability vector of each individual robot. For each capability dimension component, based on the type of the capability dimension, the corresponding target mapping function is matched and aggregated to obtain the corresponding capability dimension value; The values ​​of each capability dimension are aggregated to obtain the combined capability vector of the robot group. Specifically, for the first type of capability dimension, the capability dimension components are summed according to the corresponding linear accumulation function; for the second type of capability dimension, the maximum or minimum value of the capability dimension components is obtained according to the corresponding nonlinear extremum function; for the third type of capability dimension, the determination result of whether there is a capability dimension component of at least one individual robot in the robot group that satisfies the corresponding task threshold is determined according to the corresponding logic function.

5. The method according to claim 1, characterized in that, The process of filtering all individual robots in the capability module pool based on the environmental constraint vector and the capability requirement vector yields a candidate robot set, including: Obtain the environmental adaptation threshold in the environmental constraint vector and the rigid capability threshold in the capability requirement vector; Traverse the capability module pool and determine the candidate robot set as all individual robots whose static capability components are greater than or equal to the environmental adaptation threshold and whose remaining operational capability components are greater than or equal to the rigid capability threshold.

6. The method according to claim 1, characterized in that, The step of calculating a comprehensive matching score between the expected loss vector and the combined capability vector of each robot group, and determining the target execution object from each robot group based on the comprehensive matching score, includes: For each candidate object in each of the robot groups, obtain the values ​​of each indicator in the second capability set of the candidate object; Based on the preset scoring weight coefficients, the values ​​of each indicator are weighted and fused with the corresponding expected preference values ​​in the loss expectation vector to obtain the comprehensive matching score of the candidate object. The candidate with the highest comprehensive matching score is determined as the target execution object.

7. The method according to claim 1, characterized in that, The method further includes: Real-time monitoring of the operating parameters of each target execution object, and calculation of the task execution deviation between the operating parameters and the target task requirements; If the task execution deviation exceeds a preset abnormal threshold, the candidate robot set is dynamically rescheduled based on the task requirement vector of the remaining tasks in the target task.

8. The method according to claim 1, characterized in that, The method further includes: Obtain the preset template inheritance rules in the task template pool, and determine the inheritance relationship between the sub-task template and the parent task template according to the template inheritance rules, wherein the sub-task template inherits the environmental constraint vector, capability requirement vector, collaborative constraint vector and loss expectation vector corresponding to the parent task template; When performing a task for a new scenario, the inherited vectors are incrementally modified based on the environmental feedback information of the new scenario, and the modified task template is registered as a new subtask template to the task template pool. The task templates in the task template pool are managed in a versioned manner. When multiple task templates are associated with the same job type, the task templates are iterated and their parameters are optimized based on the execution effect of historical scheduling data.

9. A robot dynamic scheduling and control device, characterized in that, include: The acquisition module is used to acquire a task template pool containing multiple task templates and a capability module pool containing multiple robot capability models. Based on the target task, it matches the corresponding target task template from the task template pool and determines the environmental constraint vector, capability requirement vector, cooperative constraint vector and loss expectation vector corresponding to the target task. The filtering module is used to filter all individual robots in the capability module pool according to the environmental constraint vector and the capability requirement vector to obtain a candidate robot set. The grouping module is used to dynamically group the individual robots in the candidate robot set according to the cooperative constraint vector to obtain multi-body heterogeneous robot groups and independent individual robot groups, and determine the combination capability vector corresponding to each robot group. The multi-body heterogeneous robot group includes multiple individual robots, and the independent individual robot group includes one individual robot. The determination module is used to calculate the comprehensive matching score between the loss expectation vector and the combined capability vector of each robot group, and to determine the target execution object from each robot group according to the comprehensive matching score, wherein the loss expectation vector includes the expectation preference value, the robot health index and the energy reserve index; The update module is used to dispatch the target task to the target execution object, and deduct the operation capability vector of each individual robot corresponding to the target execution object according to the rated energy consumption and rated load of the target execution object, and update the capability module pool after the target task is completed.

10. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the robot dynamic scheduling and control method as described in any one of claims 1-8.

Images