Network resource scheduling method and device for cloud robot tasks and medium

Abstract

The invention discloses a network resource scheduling method and device for cloud robot tasks and a medium. According to the method, the cloud robot task deployment scheme in the cloud side-end cooperative system is regularly and dynamically adjusted, and the task completion quality and the task completion real-time performance of the cloud robot task in a complex network environment are improved. Compared with the existing cloud robot task deployment method, the scheme can ensure the task completion quality and task completion real-time performance of the cloud robot in a high-performance manner in a complex network environment. According to the scheme, the task completion quality and the task completion real-time performance of the cloud robot task are considered in the dynamic adjustment of the cloud robot task deployment scheme, the network transmission resource consumption and the network computing resource consumption of the cloud robot task are comprehensively considered, and the optimal network resource scheduling performance of the cloud robot task is ensured.

Description

Technical Field

[0001] This invention relates to the field of cloud robotics, and in particular to a method, device and medium for scheduling network resources for cloud robotic tasks. Background Technology

[0002] Most current cloud robotics systems are entirely based on cloud computing resources. For example, tasks involving deformable object manipulation robots, which have high computational demands and large amounts of sensor data, face challenges when deployed as cloud robotics systems, such as high data transmission resource consumption and difficulty in meeting real-time requirements. There are relatively few solutions for utilizing the computing power of network edge devices along the upload path to deploy deformable object manipulation robots in stages and meet the computational resource requirements of cloud robotics tasks at different nodes during the upload process. Therefore, it is imperative to deploy a network resource scheduling method for cloud robotics tasks based on dynamic programming.

[0003] Deploying high-computational-power-demanding deformable object manipulation robot tasks directly on classic cloud robotics systems presents challenges related to data and environmental impact. Firstly, deformable object manipulation robot tasks involve diverse data modalities and varying data sizes; uploading all data to the cloud would incur enormous transmission resource consumption. Secondly, deformable object manipulation robots are highly susceptible to environmental changes, and the deformability of the manipulated objects makes them sensitive to real-time performance during operations. Transmitting all data to the cloud would degrade the real-time performance of robot control information, thereby reducing the quality of task completion.

[0004] Efficiently allocating and scheduling tasks across different computing nodes to maximize overall system performance and resource utilization is a pressing issue. Edge devices have relatively limited computing power and storage resources; therefore, minimizing resource consumption on edge devices while maintaining computational efficiency is another significant research challenge. Cloud-edge-device collaborative computing systems need to process large amounts of real-time data, placing high demands on system real-time performance and reliability. Reducing data transmission latency and bandwidth consumption while ensuring real-time performance and reliability in data transmission and processing remains a challenging research topic.

[0005] In summary, designing a network resource scheduling method for cloud robot tasks based on dynamic programming is quite challenging. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by providing a network resource scheduling method, device, and medium for cloud robot tasks, thereby improving the performance of cloud robot tasks.

[0007] According to a first aspect of this specification, a network resource scheduling method for cloud robot tasks is provided, the method comprising the following steps:

[0008] S1: When deploying cloud robot tasks in a cloud-edge-device collaborative system, for a cloud robot task M with an ordered number of modules m and a cloud-edge-device collaborative system E with an ordered number of computing nodes n, in network state B, set a cloud robot task deployment scheme P, and solve the cloud robot task completion quality QoS and task completion real-time performance T corresponding to the cloud robot task deployment scheme P.

[0009] S2: Set an impact coefficient α for the completion quality of a cloud robot task. Each cloud robot task deployment scheme P corresponds to an optimization target Target = T - αQoS.

[0010] S3: Using dynamic programming, find the cloud robot task deployment scheme P that minimizes the optimization objective Target value, and deploy and execute the cloud robot tasks according to the cloud robot task deployment scheme P;

[0011] S4: During the deployment and execution of cloud robot tasks based on the cloud robot task deployment scheme P solved in S3, the update status of network state B is collected periodically, and the cloud robot task deployment scheme P is updated periodically based on the updated network state B.

[0012] Furthermore, in S1, the cloud robot task M with an ordered number of modules m has the characteristic of module serialization, where each module receives the data generated by the previous module and processes and generates the data to transmit to the next module.

[0013] Furthermore, in S1, the cloud-edge-device collaborative system E with n ordered computing nodes has the characteristic of node serialization. The n computing nodes are connected in series. Each computing node communicates only with the computing nodes before and after it. Each computing node collects the data generated by the previous computing node and processes and generates data to transmit to the next computing node.

[0014] Furthermore, in S1, the cloud robot task deployment scheme P is set as follows: based on the ordered number of computing nodes n of the cloud-edge-device collaborative system E and the ordered number of modules m of the cloud robot task M in S1, the cloud robot task deployment scheme P is set as (q1,q2,…,q…). n An array of q, where 0 ≤ q1 ≤ q2 ≤ … ≤ q n For the first n computation nodes, the cumulative number of modules that each node and its predecessor nodes are responsible for is q1,q2,…,q.n indivual.

[0015] Furthermore, in S1, the solution for the Quality of Service (QoS) of cloud robot task completion includes the following steps:

[0016] A: Based on the cloud robot task M with m ordered modules in S1, collect the update frequency ν of each module. i The influencing factor η on the QoS (Quality of Service) of task completion i The influence factor mapping relationship f corresponding to the i-th module is obtained. i The expression η i =f i (ν i );

[0017] B: For each cloud robot task deployment scheme P, solve for its task completion quality (QoS). P The MethodQoS approach is based on the deployment node of the i-th module in scheme P, where the update frequency of the i-th module is ν. i According to the influence factor mapping relationship f corresponding to the i-th module i The task completion quality corresponding to scheme P is obtained by multiplying all influencing factors.

[0018] Furthermore, in S1, the solution for the task completion timeliness T includes the following steps:

[0019] a: Based on the cloud robot task M with m ordered modules in S1, let TC be the computation time required for the i-th module on the j-th computing node. ij The computation time required for each module on each computing node was collected through experiments and denoted as TC = {TC}. 11 ,TC 12 ,…,TC 1n ,…,TC mn};

[0020] b: For each cloud robot task deployment scheme P, calculate its computation time TC. P MethodTC is to obtain TC from the set TC based on the j-th computing node where the i-th module in scheme P is deployed. ij The computation time of each module is summed to obtain TC. P ;

[0021] c: Based on the cloud robot task M with m ordered modules in S1, let TT be the transmission time required for the data calculated by the i-th module at the j-th computing node to be transmitted from that node to the next node. ijThe set of transmission times required for the data computed by each module at each computing node to be transmitted from that node to the next node is collected through experiments and is denoted as TT = {TT}. 11 ,TT 12 ,…,TT 1n ,…,TT mn};

[0022] d: For each cloud robot task deployment scheme P, solve for its transmission time TT. P Method TT is to obtain TT from the set TT based on the last module i deployed on the j-th computing node in scheme P. ij The transmission time required to transfer the data calculated by the last module i deployed on each computing node to the next node is accumulated to obtain TT. P ;

[0023] e: Based on the calculation time TC in steps b and d P and transmission time TT P Solve the real-time performance T of the task. P =TC P +TT P .

[0024] Furthermore, S3 specifically includes the following sub-steps:

[0025] aa: For the cloud robot task deployment scheme P, a two-dimensional state transition table DP is constructed based on the number of modules m and the number of nodes n, with a total size of dp[n+1][m+1]. Among them, dp[j][i] stores the minimum total cost of allocating the first i modules to the first j computing nodes. dp[0][0] = 0 is initialized, and all other dp[j][i] are initialized to infinity. A path backtracking table path[n+1][m+1] is constructed to record the optimal allocation path.

[0026] bb: Traverse j from 1 to n in the order of computed nodes; for each value of j, traverse the number of modules i from 1 to m; for each value of i, traverse the predecessor state i ′ From max(0,i) max Iterate through i to i-1;

[0027] cc: Based on the traversal of j, i, i in bb ′ The value of corresponds to a cloud robot task deployment scheme P, and the optimization target Target is determined according to the deployment scheme P. P =T P -αQoS P Solve for Target P ;

[0028] dd: If dp[j - 1][i′]+Target P < dp[j][i], update dp[j][i]=dp[j - 1][i′]+

[0029] Target P , record path[j][i]=(j - 1, i ′ );

[0030] ee: Starting from the final state dp[n + 1][m + 1], backtrack the optimal solution. By initializing i = m, j = n, and performing the following steps until i = 0, obtain the cloud robot task deployment plan P that minimizes the optimization target Target P value:

[0031] ee1: Obtain the predecessor state, read (j″, i″)=path[j][i], record the allocation plan P′, update j = j ′′ , i = i″, repeat ee1 until i = 0;

[0032] ff: Output the cloud robot task deployment plan P = P′ that minimizes the optimization target Target P value

[0033] Furthermore, in S4, each computing node periodically collects the data generated by the previous computing node and records the time interval t for collecting the data. When the time interval t is greater than the network interruption time interval threshold T, reset the cloud-edge-end collaborative system E in S1 to the cloud-edge-end collaborative system E′ composed of the computing nodes before the current node, and re-execute the network resource scheduling method

[0034] According to the second aspect of this specification, there is provided an electronic device, including a memory and a processor, the memory is coupled to the processor; wherein, the memory is used to store program data, and the processor is used to execute the program data to implement the network resource scheduling method for cloud robot tasks as described in the first aspect

[0035] According to the third aspect of this specification, there is provided a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the network resource scheduling method for cloud robot tasks as described in the first aspect

[0036] Compared with the prior art, the present invention has the following advantages

[0037] First, the deployment of cloud robot tasks in the prior art is static and difficult to adapt to complex and changeable network environments. The dynamic update of the cloud robot task deployment plan in the present invention takes into account the real-time nature of the cloud robot platform and ensures the stability of cloud robot task deployment

[0038] Second, the network resource scheduling method for this cloud robot task has very little additional operating overhead.

[0039] Third, most current cloud robot systems are entirely based on cloud computing resources, which consume a lot of transmission resources. This invention is deployed in a cloud-edge-device collaborative system, which has stronger overall performance. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram illustrating a cloud robot task deployment scheme as an exemplary embodiment.

[0042] Figure 2 This is a schematic diagram illustrating a specific ordered module and ordered computing node as an exemplary embodiment.

[0043] Figure 3 This is a schematic diagram of the structure of an electronic device as an exemplary embodiment. Detailed Implementation

[0044] To better understand the technical solution of this application, the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0045] It should be understood that the described embodiments are merely some, not all, of the embodiments in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0046] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. The singular forms “a,” “the,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0047] like Figure 1 As shown in the figure, the network resource scheduling method for cloud robot tasks proposed in this application includes the following steps:

[0048] S1. When deploying cloud robot tasks in a cloud-edge-device collaborative system using this method, for a cloud robot task M with an ordered number of modules m and a cloud-edge-device collaborative system E with an ordered number of computing nodes n, in network state B, set a cloud robot task deployment scheme P, and solve the cloud robot task completion quality QoS and task completion real-time performance T corresponding to the cloud robot task deployment scheme P.

[0049] A cloud robot task M with m ordered modules has a module-chaining characteristic, where each module receives data generated by the previous module, processes and generates data, and then transmits it to the next module. In the prototype system provided in this embodiment, m is 7; in common application scenarios, m is 7 or more.

[0050] A cloud-edge-device collaborative system E with n ordered computing nodes exhibits a node serial connection characteristic. The n computing nodes form a series link, and each computing node communicates only with its immediate adjacent computing nodes. Each computing node collects data processed and generated by the previous computing node, processes it, and transmits the data to the next computing node. In practical applications, n is often 3 or more in common application scenarios.

[0051] like Figure 2 In the example shown, the cloud robot is a deformable object manipulation robot. The ordered modules are the following functional modules in sequence: image preprocessing module, CNN module, MLP module... robotic arm control module. The ordered computing nodes are the end computing node, edge computing node, and cloud computing node in sequence.

[0052] The cloud robot task deployment scheme P is set according to the ordered number of computing nodes n of the cloud-edge-device collaborative system E in S1 and the ordered number of modules m of the cloud robot task M. Scheme P is set as (q1, q2, ..., q n An array of q, where 0 ≤ q1 ≤ q2 ≤ … ≤ q n For the first n computation nodes, the cumulative number of modules that each node and its predecessor nodes are responsible for is q1,q2,…,q. n One, such as Figure 1 As shown, the first computing node is responsible for the first 3 modules, so q1 = 3. The second computing node is responsible for the 4th and 5th modules, plus the 3 modules responsible for the preceding node, so q2 = 5. The third computing node is responsible for the 6th and 7th modules, plus the 5 modules responsible for the preceding node, so q3 = 7, and so on.

[0053] The solution for the Quality of Service (QoS) of cloud robot task completion includes the following steps:

[0054] For a cloud robotics task M consisting of m ordered modules, deploying them on different computing nodes will result in different update frequencies for each module. Since the importance of each module varies, the impact of each module's update frequency on the overall task completion quality differs. Furthermore, since there is no absolute standard for "whether a task is completed," we need to define and quantify the task completion quality of cloud robotics tasks to compare the completion quality under different conditions.

[0055] In an example of a deformable object manipulation robot task, taking the task of folding clothes as an example, the prototype provided in this application defines and quantifies the task completion quality. The specific method is to collect the coordinates of several fabric sampling points (e.g., six), and after folding, sum the L2 norm between the coordinates of these fabric sampling points and the coordinates of the corresponding folding target points to obtain the task completion quality QoS. That is, the distance difference is used as a metric, and the smaller the QoS value, the higher the task completion quality.

[0056] Let the update frequency of the i-th module be ν. i The impact factor on QoS (Quality of Service) of task completion is η. i The method for solving the influence factor is as follows: the update frequency ν of the i-th module. i The change will cause the task completion quality (QoS) to change to QoS′. i For each module, the minimum task completion quality value among all experiments is taken as the baseline value QoS_base, and the influence factor η after frequency change is updated. i =QoS′ i / QoS_base is used for assignment and collection.

[0057] The specific method for collecting data to solve the Quality of Service (QoS) of cloud robot tasks is as follows:

[0058] A: Based on the cloud robot task M with m ordered modules in S1, collect the update frequency ν of each module. i The influencing factor η on the QoS (Quality of Service) of task completion i The influence factor mapping relationship f corresponding to the i-th module is obtained. i The expression η i =f i (ν i In practical applications, η i The variation range is between 0.1 and 0.9.

[0059] Collect the update frequency ν of each module i The influencing factor η on the QoS (Quality of Service) of task completion iDuring the process, the update frequency of other modules remains constant, while the update frequency ν is adjusted within the range of 100Hz to 0.1Hz. i And record each update frequency ν i The corresponding task completion quality (QoS) is then used to obtain the influence factor mapping relationship f for the i-th module through spline interpolation. i The expression η i =f i (ν i );

[0060] B: For each cloud robot task deployment scheme P, solve for its task completion quality (QoS). P The MethodQoS approach is based on the deployment node of the i-th module in scheme P, where the update frequency of the i-th module is ν. i According to the influence factor mapping relationship f corresponding to the i-th module i The task completion quality corresponding to scheme P is obtained by multiplying all influencing factors.

[0061] The method for determining the real-time performance T of the task completion includes the following steps:

[0062] a: Based on the cloud robot task M with m ordered modules in S1, let TC be the computation time required for the i-th module on the j-th computing node. ij The computation time required for each module on each computing node was collected through experiments and denoted as TC = {TC}. 11 ,TC 12 ,…,TC 1n ,…,TC mn};

[0063] b: For each cloud robot task deployment scheme P, calculate its computation time TC. P MethodTC is to obtain TC from the set TC based on the j-th computing node where the i-th module in scheme P is deployed. ij The computation time of each module is summed to obtain TC. P .

[0064] In practical applications, TC ij Typically less than 0.1 seconds.

[0065] c: Based on the cloud robot task M with m ordered modules in S1, let TT be the transmission time required for the data calculated by the i-th module at the j-th computing node to be transmitted from that node to the next node. ij The set of transmission times required for the data computed by each module at each computing node to be transmitted from that node to the next node is collected through experiments and is denoted as TT = {TT}.11 ,TT 12 ,…,TT 1n ,…,TT mn};

[0066] d: For each cloud robot task deployment scheme P, solve for its transmission time TT. P Method TT is to obtain TT from the set TT based on the last module i deployed on the j-th computing node in scheme P. ij The transmission time required to transfer the data calculated by the last module i deployed on each computing node to the next node is accumulated to obtain TT. P .

[0067] In practical applications, TT ij Typically less than 0.1 seconds.

[0068] e: Based on the calculation time TC in steps b and d P and transmission time TT P Solve the real-time performance T of the task. P =TC P +TT P .

[0069] S2, set an impact coefficient α for the quality of cloud robot task completion. Based on MethodQoS, MethodTC, and MethodTT from steps B, b, and d, each deployment scheme P corresponds to an optimization target Target. P =T P -αQoS P In practical applications, the influence coefficient α of the cloud robot task completion quality is usually around 0.01.

[0070] S3, using dynamic programming, solves for the optimal target Target in S2. P The cloud robot task deployment scheme P with the smallest value is determined, and the cloud robot task is deployed and executed according to the cloud robot task deployment scheme P; the specific method includes the following steps:

[0071] aa: For the cloud robot task deployment scheme P, a two-dimensional state transition table DP is constructed based on the number of modules m and the number of nodes n, with a total size of dp[n+1][m+1]. Among them, dp[j][i] stores the minimum total cost of allocating the first i modules to the first j computing nodes. dp[0][0] is initialized to 0, and all other dp[j][i] are initialized to infinity (∞). A path backtracking table path[n+1][m+1] is constructed to record the optimal allocation path.

[0072] bb: Traverse j from 1 to n in the order of computing nodes; for each value of j, traverse module number i from 1 to m; for each value of i, traverse the predecessor state i ′ from max(0, i max ) to i - 1;

[0073] cc: According to the values of j, i, and i ′ in the traversal in bb, it corresponds to a cloud robot task deployment plan P. According to each deployment plan P in step S2, there is an optimization objective Target P = T P - αQoS P , solve for Target P ;

[0074] dd: If dp[j - 1][i′] + Target P < dp[j][i], update dp[j][i] = dp[j - 1][i′] +

[0075] Target P , record path[j][i] = (j - 1, i ′ );

[0076] ee: Start backtracking the optimal solution from the final state dp[n + 1][m + 1]. By initializing i = m, j = n, and performing the following steps until i = 0, obtain the cloud robot task deployment plan P that minimizes the optimization objective Target P value:

[0077] ee1: Obtain the predecessor state, read (j″, i″) = path[j][i], record the allocation plan P′, update j = j″, i = i″, and repeat ee1 until i = 0;

[0078] ff: Output the cloud robot task deployment plan P = P′ that minimizes the optimization objective Target P value.

[0079] During the process of solving using the dynamic programming method, complexity optimization measures can be taken, including realizing a time complexity of O(1) by pre - calculating the prefix sum of the two - dimensional state transition table;

[0080] By pruning strategies, reduce unnecessary calculations during the process of solving using the dynamic programming method;

[0081] By parallel computing, adopt multi - thread parallel processing for independent state transition processes to improve the performance of the embodiment.

[0082] S4. During the deployment and execution of cloud robot tasks based on the cloud robot task deployment scheme P solved in S3, the update status of network state B is collected periodically, and the cloud robot task deployment scheme P is updated periodically based on the updated network state B.

[0083] Each computing node periodically collects the data generated by the previous computing node and records the data collection time interval t. When the time interval t is greater than the network interruption time interval threshold T, the cloud-edge-device collaborative system E in S1 is reset to the cloud-edge-device collaborative system E′ composed of the computing nodes before the current node, and the above network resource scheduling method is re-executed.

[0084] In practical applications, the network interruption time interval threshold T is set at around 0.5 seconds. When the data transmission time interval is greater than this threshold, it usually indicates that the network transmission has been interrupted.

[0085] Accordingly, this application also provides an electronic device, including: one or more processors; a memory for storing one or more programs; and when the one or more programs are executed by the one or more processors, causing the one or more processors to implement the network resource scheduling method for cloud robot tasks as described above. Figure 3 The diagram shown illustrates a hardware structure of any data processing-capable device for implementing the network resource scheduling method for cloud robot tasks provided in this embodiment of the invention, except... Figure 3 In addition to the processor, memory, and network interface shown, any data processing device in the embodiment may also include other hardware depending on the actual function of the data processing device, which will not be described in detail here.

[0086] Accordingly, this application also provides a computer-readable storage medium storing computer instructions thereon, which, when executed by a processor, implement the network resource scheduling method for cloud robot tasks as described above. The computer-readable storage medium can be an internal storage unit of any data-processing device as described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data-processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data-processing device, and can also be used to temporarily store data that has been output or will be output.

[0087] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations or adaptations to this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only.

[0088] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

[0089] The above description is merely a preferred embodiment of the present invention. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make many possible variations and modifications to the technical solutions of the present invention using the methods and techniques disclosed above, or modify them into equivalent embodiments with equivalent changes, without departing from the scope of the technical solutions of the present invention. Therefore, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall still fall within the protection scope of the technical solutions of the present invention.

Claims

1. A network resource scheduling method for cloud robot tasks, characterized in that, Includes the following steps: S1: When deploying cloud robot tasks in a cloud-edge-device collaborative system, for a cloud robot task M with an ordered number of modules m and a cloud-edge-device collaborative system E with an ordered number of computing nodes n, in network state B, set a cloud robot task deployment scheme P, and solve the cloud robot task completion quality QoS and task completion real-time performance T corresponding to the cloud robot task deployment scheme P. S2: Set an impact coefficient α for the completion quality of a cloud robot task. Each cloud robot task deployment scheme P corresponds to an optimization target Target = T - αQoS. S3: Using dynamic programming, find the cloud robot task deployment scheme P that minimizes the optimization objective Target value, and deploy and execute the cloud robot tasks according to the cloud robot task deployment scheme P; S4: During the deployment and execution of cloud robot tasks based on the cloud robot task deployment scheme P solved in S3, the update status of network state B is collected periodically, and the cloud robot task deployment scheme P is updated periodically based on the updated network state B.

2. The method according to claim 1, characterized in that, In S1, the cloud robot task M with m ordered modules has the characteristic of module serialization. Each module receives the data generated by the previous module and processes and generates the data to transmit to the next module.

3. The method according to claim 1, characterized in that, In S1, the cloud-edge-device collaborative system E with n ordered computing nodes has the characteristic of node serialization. The n computing nodes are connected in series. Each computing node communicates only with the computing nodes before and after it. Each computing node collects the data generated by the previous computing node and processes and generates data to transmit to the next computing node.

4. The method according to claim 1, characterized in that, In S1, the cloud robot task deployment scheme P is set as follows: based on the ordered number of computing nodes n in the cloud-edge-device collaborative system E and the ordered number of modules m in the cloud robot task M, the cloud robot task deployment scheme P is set as (q1, q2, ..., q n An array of q, where 0 ≤ q1 ≤ q2 ≤ ... ≤ q n For the first n computation nodes, the cumulative number of modules that each node and its predecessor nodes are responsible for is q1,q2,…,q. n indivual.

5. The method according to claim 1, characterized in that, In S1, the solution for the Quality of Service (QoS) of cloud robot task completion includes the following steps: A: Based on the cloud robot task M with m ordered modules in S1, collect the update frequency ν of each module. i The influencing factor η on the QoS (Quality of Service) of task completion i The influence factor mapping relationship f corresponding to the i-th module is obtained. i The expression η i =f i (ν i ); B: For each cloud robot task deployment scheme P, solve for its task completion quality (QoS). P The MethodQoS approach is based on the deployment node of the i-th module in scheme P, where the update frequency of the i-th module is ν. i According to the influence factor mapping relationship f corresponding to the i-th module i The task completion quality corresponding to scheme P is obtained by multiplying all influencing factors.

6. The method according to claim 1, characterized in that, In S1, the solution for the real-time performance T of task completion is given. Includes the following steps: a: Based on the cloud robot task M with m ordered modules in S1, let TC be the computation time required for the i-th module on the j-th computing node. ij The computation time required for each module on each computing node was collected through experiments and denoted as TC = {TC}. 11 ,TC 12 ,…,TC 1n ,…,TC mn }; b: For each cloud robot task deployment scheme P, calculate its computation time TC. P MethodTC is to obtain TC from the set TC based on the j-th computing node where the i-th module in scheme P is deployed. ij The computation time of each module is summed to obtain TC. P ; c: Based on the cloud robot task M with m ordered modules in S1, let TT be the transmission time required for the data calculated by the i-th module at the j-th computing node to be transmitted from that node to the next node. ij The set of transmission times required for the data computed by each module at each computing node to be transmitted from that node to the next node is collected through experiments and is denoted as TT = {TT}. 11 ,TT 12 ,…,TT 1n ,…,TT mn }; d: For each cloud robot task deployment scheme P, solve for its transmission time TT. P Method TT is to obtain TT from the set TT based on the last module i deployed on the j-th computing node in scheme P. ij The transmission time required to transfer the data calculated by the last module i deployed on each computing node to the next node is accumulated to obtain TT. P ; e: Based on the calculation time TC in steps b and d P and transmission time TT P Solve the real-time performance T of the task. P =TC P +TT P .

7. The method according to claim 1, characterized in that, S3 specifically includes the following sub-steps: aa: For the cloud robot task deployment scheme P, a two-dimensional state transition table DP is constructed based on the number of modules m and the number of nodes n, with a total size of dp[n+1][m+1]. Among them, dp[j][i] stores the minimum total cost of allocating the first i modules to the first j computing nodes. dp[0][0] = 0 is initialized, and all other dp[j][i] are initialized to infinity. A path backtracking table path[n+1][m+1] is constructed to record the optimal allocation path. bb: Traverse j from 1 to n in the order of computed nodes; for each value of j, traverse the number of modules i from 1 to m; for each value of i, traverse the predecessor state i ′ From max(0,i) max Iterate through i to i-1; cc: Based on the traversal of j, i, i in bb ′ The value of corresponds to a cloud robot task deployment scheme P, and the optimization target Target is determined according to the deployment scheme P. P =T P -αQoS P Solve for Target P ; dd: If dp[j - 1][i'] + Target P < dp[j][i], update dp[j][i] = dp[j - 1][i'] + Target P Record path[j][i] = (j-1, i ′ ); ee: Starting from the final state dp[n+1][m+1], backtrack to the optimal solution by initializing i=m, j=n, and performing the following steps until i=0, to obtain the optimization objective Target. P The cloud robot task deployment scheme with the smallest value P: ee1: Get the predecessor state, read (j″,i″) = path[j][i], record the allocation scheme P′, update j = j″, i = i″, repeat ee1 until i = 0; ff: Output that optimizes the target. P The cloud robot task deployment scheme with the smallest value is P = P′.

8. The method according to claim 1, characterized in that, In S4, each computing node periodically collects the data generated by the previous computing node and records the time interval t of data collection. When the time interval t is greater than the network interruption time interval threshold T, the cloud-edge-device collaborative system E in S1 is reset to the cloud-edge-device collaborative system E′ composed of the computing nodes before the current node, and the network resource scheduling method is re-executed.

9. An electronic device comprising a memory and a processor, characterized in that, The memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the network resource scheduling method for cloud robot tasks as described in any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the network resource scheduling method for cloud robot tasks as described in any one of claims 1-8.

Images