A method for cooperative game-based resource collaborative allocation in a cleaning robot cluster

Abstract

The present application relates to resource cooperative processing technical field, specifically a kind of cleaning robot cluster based on cooperation game resource cooperative allocation method, all robots are regarded as node to participate in coalition game, and task offloading can be carried out in coalition.When constructing coalition, all nodes are initialized to be located in same coalition, and directed graph is constructed based on the preference value of offloading object to each neighbor node;Randomly select a node as target point on the directed graph, search neighbor node according to the edge with maximum preference value, until a closed loop path is formed, and the nodes corresponding to the closed loop path form a coalition, remove the nodes and connected edges in the coalition from the directed graph;Repeat the closed loop path search until all nodes are divided into corresponding coalition.The present application overcomes the problems in the prior art that bulk cargo automatic cleaning is difficult to adapt to complex environment and has low efficiency, and can fully utilize the computing and storage capacity of nodes to improve the overall performance of the system.

Description

Technical Field

[0001] This invention relates to the field of resource collaborative processing technology, specifically a resource collaborative allocation method based on cooperative game theory in a cleaning robot cluster. Background Technology

[0002] In the complex scenario of automated cleaning of bulk carriers, significant heterogeneity exists between the cleaning tasks and the nodes of the cleaning robot cluster, directly leading to a resource supply-demand imbalance. From a task perspective, the cleaning tasks exhibit multi-dimensional heterogeneity: the map building phase requires the use of LiDAR, visual sensors, and other equipment to collect complex spatial information inside the bulk carrier, transforming irregular cabin structures and pipe layouts into digital maps suitable for path planning. This process demands extremely high data processing accuracy and real-time performance. High-pressure water gun control not only requires precise adjustment of water pressure and spray angle based on the stubbornness of stains on the hull surface but also must ensure cleaning effectiveness while avoiding damage to the hull. Cleaning path planning requires comprehensive consideration of factors such as the distribution of obstacles within the cabin and cleaning coverage to plan an efficient and comprehensive cleaning route. Furthermore, the resource heterogeneity among the nodes of the cleaning robot cluster is equally significant. The computing power of different nodes varies significantly. Some nodes may be equipped with high-performance processors, capable of quickly processing large amounts of sensor data and complex algorithms, while others have weaker computing performance due to cost or size limitations. Storage capabilities also differ, causing different nodes to face varying capacity pressures when storing map data and cleaning task records. This dual heterogeneity of tasks and node resources makes it difficult to accurately match resources, resulting in a supply-demand imbalance where some nodes experience task backlog and resource shortages, while others have idle resources.

[0003] Currently, existing research on this resource allocation problem mainly falls into two categories: centralized resource management and distributed cooperative game theory methods. The mainstream research direction of centralized resource management methods relies on a central controller to uniformly schedule the resources of multiple cleaning robot nodes. The central controller acts as the "brain" of the entire cleaning system, collecting information such as the resource status and task progress of each node, and then rationally allocating cleaning tasks to each node according to a preset algorithm, thereby improving local resource efficiency. In some simple scenarios or when resources are relatively stable, it can achieve certain optimization effects. However, this approach has many drawbacks: in terms of architectural adaptability, once the cleaning environment of the bulk carrier changes, such as temporary modifications to the ship's structure or the appearance of new types of obstacles, the central controller struggles to quickly adjust its resource allocation strategy; in terms of dynamic response, since all decisions must be made through the central controller, there are delays in information transmission and processing, making it unable to react promptly to sudden cleaning tasks or node failures; in terms of scalability, as the scale of the cleaning robot cluster increases, the computational and management burden of the central controller grows exponentially, easily leading to performance bottlenecks and making it difficult to meet the needs of large-scale cleaning operations. Another, less commonly used, distributed cooperative game theory approach attempts to break the limitations of centralized systems by improving resource efficiency through collaboration among cleaning robot nodes. Each node makes autonomous decisions under certain rules, allocating tasks and sharing resources based on its own resource status and the strategies of other nodes. However, this method faces key bottlenecks in practical applications: insufficient resource coordination manifests as a lack of effective information exchange and coordination mechanisms between nodes, easily leading to multiple nodes competing for the same task resources or inefficient collaborative processing of complex tasks; lack of service caching coupling prevents nodes from sharing cached information such as processed map data and cleaning strategies, resulting in extensive redundant calculations and reduced overall efficiency; and an imperfect stable alliance mechanism leads to unstable cooperative relationships between nodes, making cooperation prone to breakdown when faced with conflicts of interest or resource fluctuations, thus failing to guarantee the successful completion of cleaning tasks. These problems severely restrict the efficient operation of automated bulk carrier cleaning systems; therefore, exploring better resource allocation solutions has become an urgent challenge. Summary of the Invention

[0004] To avoid and overcome the problems of low efficiency and difficulty in adapting to complex environments in existing automated bulk cargo cleaning technologies, this invention provides a cooperative game-theoretic resource allocation method for cleaning robot clusters. This method, through the collaborative allocation of computing and storage resources among nodes in the cleaning robot cluster, can fully utilize the computing and storage capabilities of the nodes, thereby improving the overall performance of the system.

[0005] This invention proposes a resource allocation method based on cooperative game theory in a cleaning robot cluster. All robots participate in alliance game as nodes, and nodes in the same alliance are neighbor nodes. Neighbor nodes can unload tasks from each other.

[0006] The alliance is constructed as follows:

[0007] First, initialize all nodes to be in the same federation, take the tasks that the nodes cannot complete as the offload objects, and calculate the preference values ​​of the offload objects for each neighbor node;

[0008] Construct a directed graph, where each edge corresponds to the unloading direction of each unloading task, and label the corresponding preference value.

[0009] Randomly select a node as the target point in the directed graph, search for neighboring nodes based on the edge with the largest preference value until a closed loop path is formed, make the nodes on the closed loop path form an alliance, and remove the nodes in the alliance and the connected edges from the directed graph.

[0010] Repeat the closed-loop path search until all nodes are assigned to their corresponding alliances.

[0011] Preferably, after the alliance is divided, verification is performed through the following steps:

[0012] First, within the alliance, each unloaded object is unloaded to the neighbor node corresponding to the maximum preference value. After unloading is completed, the task reward of each node is calculated.

[0013] Select any node in any alliance as the migration node to perform cross-alliance migration. After the migration, each unloaded object within the alliance is unloaded to the neighbor node corresponding to the maximum preference value. After the unloading is completed, calculate the task revenue of each node. If the task revenue of the migration node and the total task revenue of the alliance both increase after the migration, it means that the alliance is unstable and the alliance needs to be re-divided.

[0014] If, after traversing every cross-alliance migration direction for every node in each alliance, it is impossible to satisfy the condition that the task revenue of the migrated node and the total task revenue of the alliance both increase after migration, then the alliance is considered stable.

[0015] Preferably, the task reward calculation method for a node is as follows:

[0016] ;

[0017] In the formula, Indicates the first The task rewards of each node; Indicates the first Let the total number of cleaning task categories assigned to node i be the number of nodes in the nth node. The cleaning service program corresponding to a cleaning task is referred to as the cleaning service program. ; To indicate the first Individual node cache cleaning service program A binary number; Indicates the execution of cleaning service procedures Available fees; Indicates the first Each node stores the execution cleaning service program. The amount of data; To indicate the first Individual node cache cleaning service program The binary number of the data; Indicates the execution of cleaning service procedures Delay weighting; Indicates the first Each node is assigned to the cleaning service program. The optimal ratio of computing resources; This indicates a 1-bit cleaning service program. Required CPU clock cycles; Indicates the first Each node receives user-uploaded cleaning service programs. The amount of data; Represents a node Computational power; This indicates the network transmission rate.

[0018] Preferred, the first Cleaning service tasks for the first The preference values ​​for each node are:

[0019] ;

[0020] Among them, the The cleaning service procedure is the first... The execution program corresponding to each cleaning service task; the resource ratio is the ratio of computing resources to storage resources.

[0021] Preferably, within the alliance, the unloading target is selected from the neighboring nodes corresponding to the highest preference value for unloading.

[0022] Preferably, the service caching strategy for the robot is verified and optimized before building the alliance;

[0023] First, read the current cleaning service program cache status and current service caching strategy of each node; then generate a new service caching strategy with the goal of maximizing the total node revenue.

[0024] Calculate the total node revenue under the current service caching strategy and the new service caching strategy, respectively, and the probability of fluctuation for each node after implementing the new service caching strategy; fluctuation probability. The calculation formula is:

[0025] ;

[0026] in, It is a natural number; For the set temperature parameters, The difference in node revenue between the new service caching strategy and the current service caching strategy;

[0027] If the total revenue of nodes increases after implementing the new service caching strategy, and the fluctuation probability of each node... p If all values ​​are less than the set value, the current service caching strategy will be maintained; otherwise, a new service caching strategy will be implemented.

[0028] Preferably, the formula for calculating node revenue is:

[0029] ;

[0030] In the formula, Indicates the first The revenue of each node; This indicates the total number of categories of cleaning service procedures; To determine the service caching strategy, the first Does the first node cache the first...? A binary number representing a cleaning service procedure; Indicates execution of the first Fees available for this type of cleaning service procedure; Indicates the first The node stores the execution of the first node. Data volume of cleaning service programs; To indicate the first Does the node cache the first one? The binary number of data corresponding to the task in the cleaning service program; Indicates execution of the first Delay weighting for cleaning service programs; Indicates the first The node is assigned to the first The optimal ratio of computing resources for a cleaning service program; Indicates processing 1 bit of the first bit The CPU clock cycles required by the cleaning service program; Indicates the first The node receives the user-uploaded first... Data volume of cleaning service programs; Represents a node Computational power; This indicates the network transmission rate.

[0031] The preferred method for calculating the optimal computing resource ratio is as follows:

[0032] ;

[0033] in, Indicates the service caching strategy for the first The set of cleaning service program categories corresponding to the tasks assigned to each node.

[0034] The present invention proposes a resource allocation system based on cooperative game theory in a cleaning robot cluster, characterized in that it includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the resource allocation method based on cooperative game theory in a cleaning robot cluster.

[0035] The present invention proposes a storage medium storing a computer program, which, when executed, is used to implement a cooperative game-theoretic resource allocation method in a cleaning robot cluster.

[0036] Compared with the prior art, the beneficial effects of the present invention are:

[0037] 1. This method proposes a cooperative game-theoretic resource allocation approach for cleaning robot clusters. By dividing the cluster into alliances, a dynamic resource complementarity mechanism is constructed. Within each alliance, task unloading paths are optimized based on the heterogeneous characteristics of nodes, improving the utilization efficiency of computing and storage resources in a distributed environment. Through task category and alliance division, this invention prioritizes computationally intensive nodes for high-computational-demand tasks such as path planning and stain identification, while storage-intensive nodes prioritize cleaning task caching. Combined with an improved bilateral matching algorithm between unloading and receiving nodes, precise resource adaptation is achieved, effectively improving resource utilization.

[0038] 2. This method effectively reduces the processing time of the cleaning robot cluster by collaboratively optimizing task offloading and resource allocation. Within the alliance, tasks can be categorized based on computational and storage-intensive characteristics. By using computational preference values, nodes with the corresponding resources are prioritized for processing, ensuring high task execution success rate and efficiency. A service caching strategy dynamically adapts to task requirements, avoiding processing delays caused by cache misses. This distributed collaboration mechanism enables the system to maintain low-latency response even in complex multi-tasking scenarios, improving ship cleaning efficiency.

[0039] 3. This method achieves precise matching between cleaning tasks and robot characteristics through resource collaboration among nodes in the cleaning robot cluster and an improved bilateral matching algorithm. The dynamic alliance mechanism enables collaborative allocation of cleaning task data caching, cleaning path optimization, and robotic arm control resources, which can significantly improve hull cleaning efficiency and reduce cleaning energy consumption.

[0040] 4. In this invention, the service caching strategy is verified and optimized before the alliance is divided, thereby optimizing task allocation, which helps to reduce the number of unloaded objects and further improves task execution efficiency. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the process of the present invention;

[0042] Figure 2 This is a flowchart of the service caching strategy optimization process in this invention;

[0043] Figure 3 This is a flowchart of the alliance division and task execution process in this invention;

[0044] Figure 4 This illustrates the impact of the number of cleaning robots on performance indicators in the embodiments.

[0045] Figure 5 This illustrates the impact of the total number of cleaning service types on performance metrics in the embodiments.

[0046] Figure 6 This example illustrates the impact of the cleaning robot's storage capacity on performance indicators. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Reference Figure 1 This invention proposes a cooperative game theory-based resource allocation method for a cleaning robot cluster. First, all cleaning robots in the system are considered as nodes. With the goal of maximizing the total node revenue, a service caching strategy is determined, i.e., the cleaning service tasks (referred to as tasks) that each robot needs to execute are determined. The robot needs to cache the corresponding cleaning service program and complete the task through program execution. Then, cleaning tasks are allocated, and alliances are formed, making nodes in the same alliance neighbors. Neighboring nodes unload tasks (i.e., cleaning service tasks) from each other to ensure that each cleaning task can be completed smoothly and to guarantee the robot's task revenue.

[0049] Reference Figure 2 In this method, the service caching strategy is determined according to the following steps.

[0050] St1. Obtain the current service caching strategy and read the current cleaning service program cache status (hereinafter referred to as cache status) of each node. Calculate the revenue of each node by combining the current service caching strategy and cache status.

[0051] St2, generate a service caching strategy with the goal of maximizing the total revenue of nodes as the new caching strategy, and calculate the revenue of each node based on the cache status and the new caching strategy.

[0052] Specifically, in this step, the service caching strategy can be updated through Gibbs sampling iterations until the total node revenue is maximized.

[0053] Step 3: Compare the new caching strategy with the current service caching strategy, and calculate the fluctuation probability of each node. p And the changes in the total node revenue (the sum of revenue from all nodes);

[0054] If the floating probability of each node p If all values ​​are less than the set value, and the total benefit of the new caching strategy is greater, then the new caching strategy is adopted; otherwise, the current caching strategy is maintained. The set value can be set based on experience.

[0055] No. The revenue of each node The calculation formula is:

[0056] ;

[0057] In the formula, Indicates the first The revenue of each node; This indicates the total number of categories of cleaning service procedures; To determine the service caching strategy, the first Does the first node cache the first...? Binary numbers for cleaning service procedures. =1 indicates the first The node needs to cache the first one. Cleaning service procedures, and vice versa. =0; Indicates execution of the first Fees available for this type of cleaning service procedure; Indicates the first The node stores the execution of the first node. Data volume of cleaning service programs; To indicate the first Does the node cache the first one? The data of the cleaning service program is a binary number, which is 1 if it is true and 0 otherwise; Indicates execution of the first Delay weighting for cleaning service programs; Indicates the first The node is assigned to the first The optimal ratio of computing resources for a cleaning service program; This indicates processing 1 bit (bit) of the first... The CPU clock cycles required by the cleaning service program; Indicates the first The node receives the user-uploaded first... Data volume of cleaning service programs; Represents a node Computational power; This indicates the network transmission rate.

[0058] The data in the cleaning service procedure refers to the data of the cleaning service tasks corresponding to the cleaning service procedure.

[0059] In terms of computational resource allocation, the Lagrange duality method is used to solve the convex optimization problem, allocating a proportion of computational resources to each type of task, i.e. ,in, Indicates the first The set of cleaning service program categories that the current node needs to execute, i.e., the service caching strategy given to the node. The set of cleaning service program categories corresponding to the tasks (cleaning service tasks) assigned to each node.

[0060] make Indicates the first The change in revenue at the nth node, i.e., the change in revenue at the nth node The absolute value of the difference between the benefit of the nth node under the new caching strategy and the benefit under the current caching strategy; then the nth node... The floating probability of each node The calculation formula is:

[0061] ;

[0062] in, It is a natural number; The set temperature parameter is used to control the randomness of the probabilistic decision-making process.

[0063] Reference Figure 3 In this method, through the following steps S1-S6, a coalition is constructed with nodes as game participants, and cleaning tasks are assigned. Nodes in the same coalition can unload tasks from each other and are denoted as neighboring nodes.

[0064] S1. Assume all nodes are in the same federation. Use tasks that a node cannot complete as offloadable tasks, and calculate the preference value of each neighboring node for the offloadable task. The higher the preference value, the greater the probability that the corresponding neighboring node has stored the corresponding cleaning service program and can execute it successfully, i.e., the higher the success rate of completing the corresponding cleaning task. Tasks that a node cannot complete refer to tasks for which the node has been assigned cleaning service tasks but has not loaded the corresponding cleaning service program.

[0065] No. The cleaning service task is for the first The preference values ​​for each node are:

[0066] ;

[0067] No. The cleaning service procedure is the first... The execution program corresponding to each cleaning service task.

[0068] The resource ratio is the ratio of computing resources to storage resources; the cleaning service program resource ratio is the resource ratio of the cleaning service program; the node's remaining resource ratio is the ratio of the node's remaining resources, that is, the ratio of the node's remaining computing resources to its remaining storage resources.

[0069] S2. Construct a directed graph, with each edge of the graph corresponding to the unloading direction of each unloading task, and label the corresponding preference value. Randomly select a node in the directed graph as the target point, and search for neighboring nodes based on the edge with the largest preference value until a closed loop path is formed. Make the nodes corresponding to the closed loop path form an alliance, and remove the nodes in the alliance and the connected edges from the directed graph. Repeat the closed loop search until all nodes are assigned to an alliance.

[0070] This step S2 can be further divided into the following sub-steps S21-S24:

[0071] S21. On a directed graph, sort the directed edges according to the preference values ​​from largest to smallest to form a preference relationship graph;

[0072] S22. On the preference graph, select a target point as the first point of the path, then find the second point of the path based on edge 1 of the first point of the path, find the third point of the path based on edge 1 of the second point of the path, and so on, until a closed loop path is formed on the preference graph.

[0073] S23. Extract the nodes on the closed-loop path to form a union, and delete the nodes in the union and the edges connected to the nodes from the directed graph;

[0074] S24. Determine if there are any remaining nodes in the directed graph;

[0075] If yes, then return to step S21;

[0076] If not, then determine the alliance division and mark the uninstallation direction and preference value corresponding to each uninstallation object in the alliance.

[0077] S3. Perform predictive verification of alliance stability.

[0078] First, within the alliance, each unloaded object is unloaded to the neighbor node corresponding to the maximum preference value. After unloading is completed, the task reward of each node is calculated.

[0079] Iterate through each cross-alliance migration direction of each node in each alliance, construct the post-migration alliance corresponding to each migration direction, unload each unloaded object to the neighbor node corresponding to the maximum preference value within the alliance, and calculate the task reward of each node after unloading is completed.

[0080] Determine if there exists any migration direction for any node that satisfies the following condition: after migration, both the task reward of the migrating node and the total task reward of the alliance increase.

[0081] If it exists, return the original directed graph, i.e., the directed graph containing all nodes, and proceed to step S2;

[0082] If it does not exist, then the alliance is deemed robust, and step S4 is executed.

[0083] It's important to note that for each migration direction, only one node migrates. After a node migrates, only the unloading objects and unloading directions change in the alliance it migrated out of and the alliance it migrated into, thus affecting the task's rewards. Therefore, in this step, we only need to calculate the sum of the total rewards of the alliance that migrated out of and the total rewards of the node that migrated into. If this sum increases, it indicates that the total rewards of the alliance have increased.

[0084] In this step, the alliance uses task offloading preferences as a guide. Even if a task assigned to any node in the alliance does not have a corresponding service program on that node, it can find a corresponding service program within the alliance. Therefore, the task revenue calculation method for each node in the alliance is as follows:

[0085] ;

[0086] In the formula, Indicates the first The task rewards of each node; Represents a node (i.e., the first) The total number of cleaning task categories assigned to the node, let the nth node be the number of cleaning task categories assigned to the node. The cleaning service program corresponding to a cleaning task is referred to as the cleaning service program. ; Indicates the first Does each node cache the cleaning service program? If yes, it is 1; otherwise, it is 0. Indicates the execution of cleaning service procedures Available fees; Indicates the first Each node stores the execution cleaning service program. The amount of data; Indicates the first Does each node cache the cleaning service program? The data is 1 if it is true and 0 otherwise; Indicates the execution of cleaning service procedures Delay weighting; Indicates the first Each node is assigned to the cleaning service program. The optimal ratio of computing resources can be calculated by referring to... ; This indicates a 1-bit cleaning service program. Required CPU clock cycles; Indicates the first Each node receives user-uploaded cleaning service programs. The amount of data; Represents a node Computational power; This indicates the network transmission rate.

[0087] S4. Within the alliance, unload each unloaded object to the neighbor node corresponding to the maximum preference value and execute the cleaning service task.

[0088] It is worth noting that the unloading action in step S3 is a hypothetical operation used to calculate the performance of the alliance. Only after the alliance has been verified to be stable is it necessary to perform task unloading in step S4. This avoids the easy calculation and data damage caused by frequent unloading, and improves the calculation efficiency and execution success rate.

[0089] It is worth noting that during the alliance iteration process, the system records the target point selection each time to avoid repeated operations; if steps S2-S3 are repeated multiple times, for example, reaching a set number of times... If the task allocation fails, all tasks will be uploaded to the central server for allocation.

[0090] To verify the feasibility and effectiveness of the cooperative game-based resource allocation method for cleaning robot clusters proposed in this invention in the cleaning scenario of bulk carriers, the following simulation experiment scenario was constructed.

[0091] In the experimental scenario, the surface of the bulk carrier hull was divided into multiple cleaning areas, each handled by a single cleaning robot. Each robot possessed a certain level of computing and storage capabilities to perform tasks such as task scheduling, path planning, and data processing. The central controller collected task requirements and acted as a global coordinating node, but did not directly participate in the execution of specific cleaning tasks.

[0092] The method of this invention constructs a cooperative game model of computing and storage resources with cleaning robots as alliance members. Within the alliance, it realizes the joint optimization allocation of cleaning task service cache, cleaning task data cache and computing resources, and continuously adjusts the cooperative relationship between robots through the alliance formation mechanism until a stable alliance structure is achieved.

[0093] During the experiment, the method of this invention was used to schedule a cluster of cleaning robots. System performance indicators, including task execution latency and robot resource utilization, were recorded under different robot numbers, task scales, and resource configurations. The experimental results are as follows: Figure 4 , Figure 5 , Figure 6 As shown.

[0094] from Figure 4 It can be seen that the more cleaning robots there are, the more data can be processed (i.e., the total amount of data for performing cleaning service tasks), and the amount of data that can be processed maintains a linear increasing trend relative to the number of cleaning robots; when the number of cleaning robots reaches 40, the total latency basically converges; indicating that the method of the present invention has more outstanding performance in scenarios with more cleaning robots.

[0095] from Figure 5 As can be seen, the more cleaning services there are, the lower the overall latency becomes. This is because the more cleaning services there are, the more cleaning service programs each robot caches, and the less task unloading occurs between robots, resulting in better task execution timeliness.

[0096] from Figure 6 As you can see, the amount of data processed increases linearly with the storage capacity of the cleaning robot, further demonstrating that the present invention can maintain good working performance even in scenarios with large amounts of data, and is applicable to complex scenarios.

[0097] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A resource allocation method based on cooperative game theory in a cleaning robot cluster, characterized in that: All robots participate in alliance games as nodes. Nodes in the same alliance are neighbors, and neighboring nodes can unload tasks from each other. The alliance is constructed in the following way: First, initialize all nodes to be in the same federation, take the tasks that the nodes cannot complete as the offload objects, and calculate the preference values ​​of the offload objects for each neighbor node; Construct a directed graph, where each edge corresponds to the unloading direction of each unloading task, and label the corresponding preference value. Randomly select a node as the target point in the directed graph, search for neighboring nodes based on the edge with the largest preference value until a closed loop path is formed, make the nodes on the closed loop path form an alliance, and remove the nodes in the alliance and the connected edges from the directed graph. Repeat the closed-loop path search until all nodes are assigned to their corresponding alliances; After the alliance is divided, the following steps are required for verification: First, assume that each unloaded object is unloaded to the neighbor node corresponding to the maximum preference value within the alliance. After the unloading is completed, calculate the task reward of each node. Select any node in any alliance as the migration node to perform cross-alliance migration. Assume that after the migration, each unloaded object within the alliance will be unloaded to the neighbor node corresponding to the maximum preference value. After the unloading is completed, calculate the task revenue of each node. If the task revenue of the migrated node and the total task revenue of the alliance both increase after the migration, it indicates that the alliance is unstable and needs to be reorganized. If, after traversing every cross-alliance migration direction for every node in each alliance, it is impossible to satisfy the condition that the task revenue of the migrated node and the total task revenue of the alliance both increase after migration, then the alliance is considered stable. Within the alliance, the unloading target selects the neighbor node corresponding to the highest preference value for unloading.

2. The resource allocation method based on cooperative game theory in a cleaning robot cluster according to claim 1, characterized in that, The task reward calculation method for a node is as follows: In the formula, Indicates the first The task rewards of each node; Indicates the first Let the total number of cleaning task categories assigned to node i be the number of nodes in the nth node. The cleaning service program corresponding to a cleaning task is referred to as the cleaning service program. ; To indicate the first Individual node cache cleaning service program A binary number; Indicates the execution of cleaning service procedures Available fees; Indicates the first Each node stores the execution cleaning service program. The amount of data; To indicate the first Individual node cache cleaning service program The binary number of the data; Indicates the execution of cleaning service procedures Delay weighting; Indicates the first Each node is assigned to the cleaning service program. The optimal ratio of computing resources; This indicates a 1-bit cleaning service program. Required CPU clock cycles; Indicates the first Each node receives user-uploaded cleaning service programs. The amount of data; Represents a node Computational power; This indicates the network transmission rate.

3. The resource allocation method based on cooperative game theory in a cleaning robot cluster according to claim 1, characterized in that, No. Cleaning service tasks for the first The preference values ​​for each node are: Among them, the The cleaning service procedure is the first The execution program corresponding to the cleaning service task; the resource ratio is the ratio of computing resources to storage resources.

4. A resource allocation method based on cooperative game theory in a cleaning robot cluster according to any one of claims 1-3, characterized in that, Before building the alliance, the service caching strategy for the robot was also verified and optimized. First, read the current cleaning service program cache status and current service caching strategy of each node; then generate a new service caching strategy with the goal of maximizing the total node revenue. Calculate the total node revenue under the current service caching strategy and the new service caching strategy, respectively, and the probability of fluctuation for each node after implementing the new service caching strategy; fluctuation probability. The calculation formula is: in, It is a natural number; For the set temperature parameters, The difference in node revenue between the new service caching strategy and the current service caching strategy; If the total revenue of nodes increases after implementing the new service caching strategy, and the fluctuation probability of each node... p If all values ​​are less than the set value, the current service caching strategy will be maintained; otherwise, a new service caching strategy will be implemented.

5. The resource allocation method based on cooperative game theory in a cleaning robot cluster according to claim 4, characterized in that, The formula for calculating node revenue is: In the formula, Indicates the first The revenue of each node; This indicates the total number of categories of cleaning service procedures; To determine the service caching strategy, the first Does the first node cache the first...? A binary number representing a cleaning service procedure; Indicates execution of the first Fees available for this type of cleaning service procedure; Indicates the first The node stores the execution of the first node. Data volume of cleaning service programs; To indicate the first Does the node cache the first one? The binary number of data corresponding to the task in the cleaning service program; Indicates execution of the first Delay weighting for cleaning service programs; Indicates the first The node is assigned to the first The optimal ratio of computing resources for a cleaning service program; Indicates processing 1 bit of the first bit The CPU clock cycles required by the cleaning service program; Indicates the first The node receives the user-uploaded first... Data volume of cleaning service programs; Represents a node Computational power; This indicates the network transmission rate.

6. The resource allocation method based on cooperative game theory in a cleaning robot cluster according to claim 5, characterized in that, The optimal computing resource ratio is calculated as follows: in, Indicates the service caching strategy for the first The set of cleaning service program categories corresponding to the tasks assigned to each node.

7. A resource allocation system based on cooperative game theory in a cleaning robot cluster, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program and the processor executes the computer program to implement a cooperative game-based resource allocation method for a cleaning robot cluster as described in any one of claims 1-6.

8. A storage medium, characterized in that, The system contains a computer program, which, when executed, is used to implement a cooperative game-theoretic resource allocation method for a cleaning robot cluster as described in any one of claims 1-6.

Images