A method for scheduling a cleaning robot and a computer readable storage medium

Abstract

The application discloses a scheduling method of a cleaning robot and a computer readable storage medium, and relates to the field of cleaning robots, which comprises the following steps: encapsulating the cleaning task of each grid unit as a smart contract and publishing the smart contract to a blockchain network; each cleaning robot accessing the blockchain network responds to the smart contract, acquires local environment information of a target grid unit and self state information, and makes a bid based on the self state information, the local environment information and pheromone concentration perceived from the target grid unit; the blockchain network determines a target cleaning robot that wins the bid based on a preset consensus rule; the target cleaning robot executes the cleaning task and releases pheromones in the target grid unit during the execution of the task; and the blockchain network verifies the execution state of the cleaning task of the target grid unit, and after verification, the task reward is distributed to the virtual account of the target cleaning robot through the smart contract. The application can realize dynamic, credible and efficient distributed cleaning task scheduling.

Description

Technical Field

[0001] This application relates to the technical field of cleaning robots, and in particular to a scheduling method for cleaning robots and a computer-readable storage medium. Background Technology

[0002] With the rapid development of the Internet of Things and intelligent robot technology, cleaning robots such as sweepers are widely used in large parks, public places, and commercial facilities. Currently, the scheduling of multiple sweepers mainly relies on centralized scheduling systems; such systems typically have a central server or control center responsible for collecting environmental information and vehicle status, and uniformly planning and issuing tasks to each sweeper through preset algorithms or manual instructions.

[0003] However, this centralized scheduling model has the following inherent drawbacks: First, the system has a single point of failure risk. If the central server fails or communication is interrupted, the entire scheduling system will be paralyzed, causing cleaning operations to stop.

[0004] Secondly, the system has poor scalability. When the number of sweepers needs to be increased or decreased, the central scheduling algorithm often needs to be readjusted or globally optimized, making it difficult to achieve smooth elastic scaling.

[0005] Furthermore, in scenarios involving multiple operators (such as shared cleaning services), centralized systems struggle to guarantee the transparency and fairness of task allocation, execution records, and fee settlement, resulting in a lack of credible collaborative foundations among the parties.

[0006] Furthermore, centralized decision-making faces significant computational pressure and limited real-time response capabilities when dealing with dynamic and complex cleaning environments.

[0007] Therefore, how to design a decentralized, highly robust, scalable, and trust-building intelligent management and scheduling method for cleaning robots has become a pressing technical problem to be solved in this field. Summary of the Invention

[0008] The purpose of this application is to provide a scheduling method for cleaning robots and a computer-readable storage medium to solve the problems of single point of failure risk, poor system scalability, lack of trusted cooperation mechanism among multiple operators, and insufficient adaptability to dynamic environments in the centralized scheduling mode of the prior art.

[0009] Firstly, the scheduling method for a cleaning robot provided in this application adopts the following technical solution: A method for scheduling cleaning robots includes the following steps: S10. Divide the area to be cleaned into multiple grid units, then encapsulate the cleaning tasks of each grid unit into smart contracts and publish them to the blockchain network. S20. Each cleaning robot connected to the blockchain network responds to the smart contract, obtains the local environmental information and its own state information of the target grid unit, and bids based on its own state information, the local environmental information, and the pheromone concentration sensed from the target grid unit; wherein, the target grid unit is the grid unit corresponding to the cleaning task. S30. The blockchain network determines the winning target cleaning robot based on preset consensus rules; S40. The target cleaning robot performs the cleaning task and releases pheromones in the target grid cell during the task execution. S50. The blockchain network verifies the cleaning task execution status of the target grid unit. After successful verification, the task reward is distributed to the virtual account of the target cleaning robot through the smart contract.

[0010] In an alternative embodiment of the first aspect, step S10, dividing the area to be cleaned into multiple grid units, includes: Based on historical cleaning data and / or real-time environmental data, the granularity of the grid units in the park is dynamically adjusted to obtain multiple grid units.

[0011] In an optional implementation of the first aspect, the scheduling method further includes: Based on historical cleaning data and / or real-time environmental data, the task generation frequency is dynamically adjusted to dynamically publish smart contracts for each cleaning task.

[0012] In one optional embodiment of the first aspect, in step S20, the self-state information includes self-position information; The local environmental information includes the location information, area, and pedestrian density level of the target grid cell; The bidding process based on the self-state information, the local environment information, and the pheromone concentration sensed from the target grid cell includes: In response to the smart contract, each of the cleaning robots performs the following operations: Based on its own position information and the position information of the target grid cell, the basic distance from the cleaning robot to the target grid cell is calculated; The estimated cleaning time is determined based on the area of ​​the target grid cell; Obtain the rated moving speed, rated moving power, and rated cleaning power of the cleaning robot; The base distance and the rated moving speed are corrected based on the pedestrian density level to obtain the adjusted moving distance and the adjusted moving speed. Based on the adjusted moving speed and the adjusted moving distance, calculate the moving time required to move to the target grid cell; Calculate the mobile power consumption based on the rated mobile power and the mobile duration; Calculate the cleaning power consumption based on the rated cleaning power and the estimated cleaning time; Based on the mobile power consumption and the cleaning power consumption, calculate the total energy consumption required to perform the cleaning task; Obtain the task reward set in the smart contract; Obtain the pheromone concentration sensed from the target grid cell; The bidding price is determined based on the task reward, the base distance, the total energy consumption, and the pheromone concentration.

[0013] In one alternative implementation of the first aspect, determining the bidding price based on the task reward, the base distance, the total energy consumption, and the pheromone concentration includes: The bid price is determined using the following formula: ; in, This represents the bid offered by cleaning robot i for task j. This represents the reward for task j. This represents the basic distance from cleaning robot i to the target grid cell corresponding to task j. This represents the total energy consumption required for cleaning robot i to perform task j. This represents the pheromone concentration sensed from the target grid cell corresponding to task j. These represent the weighting coefficients corresponding to rewards, base distance, total energy consumption, and pheromone concentration, respectively.

[0014] In one alternative embodiment of the first aspect, the pheromone concentration is negatively correlated with the bidding price.

[0015] In an alternative implementation of the first aspect, in step S30, the blockchain network determines the winning target cleaning robot based on preset consensus rules, including: The blockchain network verifies the bidding bids of all cleaning robots, determines the cleaning robot with the highest bid through a consensus mechanism, and selects the cleaning robot that has the right to undertake the cleaning task. The cleaning robot that has the right to undertake the cleaning task is designated as the target cleaning robot. The bidding bid information and the undertaking result of the target cleaning robot are then verified by multiple nodes and written into the blockchain.

[0016] In an alternative embodiment of the first aspect, in step S40, the target cleaning robot performs the cleaning task and releases pheromones in the target grid cell during the task execution, including: The target cleaning robot dynamically adjusts the pheromone release rate according to the real-time cleaning speed, so that the pheromone concentration changes with the release rate and time. Alternatively, the target cleaning robot releases pheromones according to the progress of the cleaning task, wherein the concentration of the pheromones is positively correlated with the progress of the task.

[0017] In an alternative implementation of the first aspect, in step S50, the blockchain network verifies the cleaning task execution status of the target grid cell, including: The blockchain network acquires images of the target grid cell through camera devices deployed near the target grid cell; If the target grid cell is determined to be clean based on the image, the task is considered to have been successfully executed.

[0018] Secondly, the computer-readable storage medium provided in this application adopts the following technical solution: A computer-readable storage medium includes a computer program that, when executed by a processor, performs the scheduling method for the cleaning robot provided in the first aspect.

[0019] Compared with the prior art, the beneficial effects of this application are as follows: The scheduling method proposed in this application integrates IoT sensing, AI prediction, and blockchain smart contracts to achieve dynamic, reliable, and efficient distributed cleaning task scheduling, significantly improving the overall efficiency and system adaptability of cleaning operations.

[0020] Meanwhile, by building a decentralized and trustworthy task distribution platform and eliminating the central scheduling node, the risk of single point of failure in the traditional centralized scheduling model is effectively avoided. Even if individual cleaning robots malfunction or exit the system, it will not affect the overall scheduling process, significantly improving the robustness and scalability of the system.

[0021] In addition, each cleaning robot bids autonomously based on its own status, task rewards, and the pheromone concentration perceived from the environment. Through blockchain smart contracts and the introduction of an incentive-based bidding mechanism, the autonomous decision-making ability of each cleaning robot is stimulated, enabling it to optimize the efficiency of global task allocation through local information interaction while pursuing individual benefits. Furthermore, leveraging the immutability, transparency, and traceability inherent in blockchain network technology ensures the openness, fairness, and trustworthiness of the entire process from task release, bidding, execution to reward settlement. Moreover, by using pre-set consensus rules, tasks are automatically, fairly, and transparently allocated, laying a reliable foundation of trust for shared cleaning scenarios involving multiple operators.

[0022] Secondly, by introducing pheromones as a distributed coordination mechanism, the robot can perceive the status of the work area in real time (such as whether other robots are already cleaning) and dynamically adjust its own behavior through bidding, thereby achieving autonomous collaboration and load balancing among multiple robots. Even if some robots or network nodes malfunction, the entire system can still operate continuously and stably. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the scheduling method for the cleaning robot according to an embodiment of this application. Detailed Implementation

[0024] The following is in conjunction with the appendix Figure 1 This application will be described in further detail.

[0025] Firstly, this application discloses a method for scheduling cleaning robots, which is applied to a system comprising multiple cleaning robots, an Internet of Things (IoT) sensing network, and a blockchain network. Furthermore, the cleaning robots in this application can include, but are not limited to, intelligent devices with autonomous movement and cleaning functions such as sweeping robots, floor scrubbing robots, and outdoor cleaning vehicles.

[0026] Specifically, refer to Figure 1 In this embodiment, the scheduling method includes the following steps: S10. Divide the area to be cleaned into multiple grid units, then encapsulate the cleaning tasks of each grid unit into smart contracts and publish them to the blockchain network.

[0027] Meanwhile, in step S10, the area to be cleaned is divided into multiple grid units, including: The granularity of the grid units in the park can be dynamically adjusted based on historical cleaning data and / or real-time environmental data to obtain multiple grid units. The areas of different grid units can be the same or different; and the park described above is the work area.

[0028] Specifically, historical cleaning data may include, but is not limited to: the number of grid units divided into the park each time, the area of ​​each grid unit, the degree of pollution in each grid unit, and the corresponding cleaning completion time. Real-time environmental data may include, but is not limited to: dynamic information such as the current overall pollution status of the park and pedestrian density.

[0029] More specifically, the key technical terms mentioned above are explained below: Blockchain refers to a distributed ledger technology maintained by multiple nodes. It achieves distributed storage, verification, and synchronization of data through a peer-to-peer network architecture, cryptographic algorithms, and consensus mechanisms, and features decentralization, tamper-proofing, and traceability. In this application, blockchain serves as the trusted infrastructure for task publishing, bidding confirmation, and reward settlement.

[0030] Smart contracts refer to computer program code deployed on a blockchain that can be automatically executed when preset conditions are met. In this application, smart contracts are used to define the content of cleaning tasks, the reward for the task, the verification conditions, and the settlement logic, thereby achieving automation and decentralization of task management.

[0031] Based on the above data, various implementation methods can be used to dynamically determine the mesh granularity: In the first implementation, grid division is performed based on historical cleaning data. Specifically, the average number of grid cells divided in each historical park cleaning is calculated, and the number of grid cells to be divided this time is determined based on this average.

[0032] In the second implementation, grid partitioning can be performed based on real-time environmental data. Specifically, real-time environmental information of each area of ​​the park is acquired through an IoT sensing network, and the overall pollution status of the park is assessed. The partitioning strategy is dynamically adjusted based on the assessment results of the pollution level: if the pollution level is identified as high (e.g., dense garbage distribution, wide dust coverage), a smaller partitioning granularity parameter k is used (i.e., generating more grid cells with smaller unit areas); if the pollution level is identified as low, a larger partitioning granularity parameter m is used (i.e., generating fewer grid cells with larger unit areas), thereby improving the operational efficiency of large-area clean areas. Here, m > k to ensure that contaminated areas are divided into more refined operational zones.

[0033] In the third implementation, grid partitioning is performed based on historical cleaning data and real-time environmental data, and may include the following steps a1 to a6: a1. By acquiring monitoring area image information collected by each camera in the park and dust detection information collected by each dust sensor through the Internet of Things sensing network, the park will be divided into multiple sub-areas according to the degree of pollution based on the monitoring area image information and dust detection information, so as to generate a pollution level distribution map of the park.

[0034] Specifically, at least one sensor (such as a camera and a dust sensor; multiple sensors at different angles can be deployed at the same location) is set up at different locations within the park to ensure that the sensor combination at each location can cover an independent sub-area without blind spots. After collecting image information and dust detection information for the corresponding sub-area through these sensors, the image data and dust detection data of all sub-areas are aggregated, and the image data and dust detection data of each sub-area are input one by one into a pre-trained pollution detection model. The pollution detection model analyzes the characteristics of the sub-area, such as garbage density and dust coverage, to output the pollution level. The pollution level is divided into three levels: low pollution (light pollution), medium pollution (moderate pollution), and high pollution (severe pollution).

[0035] The training process of the pollution detection model is as follows: training samples carrying low / medium / high pollution sample labels are obtained and input into the machine learning model to obtain prediction results; the deviation between the prediction results and the sample labels is calculated. If the deviation exceeds a preset threshold, the model is iteratively trained based on the deviation until the deviation is less than or equal to the preset deviation threshold or the number of model training times reaches the preset maximum number of iterations.

[0036] Finally, based on the model output, the park is divided into multiple sub-regions and labeled with their corresponding pollution levels, generating a pollution level distribution map of the park.

[0037] a2. Based on the current and historical pedestrian traffic data of each sub-area, and combined with the park's scheduling data, the traffic density level of each sub-area within the prediction time window is obtained through a prediction model; where the prediction time window is the preset maximum cleaning duration after the cleaning start plan time, that is, the prediction time window is [cleaning start plan time, cleaning start plan time + preset maximum cleaning duration T_max].

[0038] Specifically, a2 above includes the following steps: 1. Collect real-time pedestrian flow data: During the preset time period before the cleaning start time (e.g., the first 10 minutes), the current pedestrian flow data of each sub-area is collected in real time through the park's IoT sensing system (cameras, infrared sensors, access control system, etc.) at a preset update frequency (e.g., every 5 minutes), and the current pedestrian flow trend is generated. 2. Extract structured activity scheduling data for each sub-region within the forecast time window from the park management system. The core fields include: sub-region number (which needs to be mapped one-to-one with the sub-region numbers classified by pollution level in step a1), whether an activity is being held, activity type, expected number of participants, and historical records of similar activities. 3. Extract historical pedestrian traffic data: Extract pedestrian traffic data for each sub-area within a preset adjacent time period (e.g., the last 3-12 months) from the historical database of the park's IoT sensing system. Specifically, this includes: pedestrian traffic statistics for each sub-area, and pedestrian traffic change curves corresponding to similar activities in the same historical period. 4. Perform unified preprocessing on the above-mentioned real-time pedestrian flow data, event schedule data, and historical pedestrian flow data, including: (1) Standardize timestamps: Standardize the timestamps of all data to the format “YYYY-MM-DDHH:MM”; (2) Missing / outlier handling: A small number of missing data are supplemented by linear interpolation. Outliers such as sensor failure and extreme peak values ​​are replaced with the average value of the historical data of the same period in the corresponding sub-region. 5. Input the preprocessed real-time pedestrian flow data, event schedule data, and historical pedestrian flow data into the encoder to generate a structured multi-dimensional feature vector. The specific vector composition is as follows: Real-time pedestrian flow characteristics: average pedestrian flow in sub-areas (normalized) 10 minutes before the prediction window starts, and real-time pedestrian flow trend; Scheduled event characteristics: whether there is an event in the sub-region within the prediction window (yes=1, no=0), event type (integer code: e.g., conference=1 / exhibition=2, etc.), time overlap rate between the event and the window, and the expected number of participants in the event (after normalization). Historical reference features: average pedestrian flow in sub-areas within a preset time period, peak pedestrian flow of similar activities in the same historical period, and pedestrian flow fluctuation rate of similar activities in the same historical period.

[0039] 6. Input the multi-dimensional feature vectors into the trained temporal machine learning model (such as the LSTM model, adapted to the temporal dependence of pedestrian flow), and output the average pedestrian flow density prediction results (unit: people / square meter) for each sub-region within the prediction time window. 7. Classify traffic density levels: Based on the preset traffic density classification standards, such as high traffic density ≥ 0.5 people / m², medium traffic density 0.2-0.5 people / m², and low traffic density < 0.2 people / m², the average traffic density prediction results of each sub-region are mapped to the corresponding traffic density level, and finally the traffic density level of each sub-region within the prediction time window is output.

[0040] Understandably, park events (such as exhibitions and conferences) are the core trigger for significant fluctuations in pedestrian traffic in localized areas within a short period. Traditional forecasting methods that rely solely on historical and real-time data are ill-suited to anticipating such sudden changes in pedestrian flow and are prone to prediction errors. This solution introduces event scheduling data and leverages key information such as event type and expected number of participants to quantify the impact of different events on pedestrian flow (e.g., exhibitions typically attract more participants than conferences, resulting in higher peak pedestrian density). This allows for the early identification of high-risk areas and, by utilizing the overlap between event start and end times and the forecast window, accurately pinpoints the timeframe for sudden changes in pedestrian flow. This avoids prediction lags caused by overlaps between event and cleaning periods. Finally, by combining historical records of similar events and reusing past pedestrian flow curves, the solution further reduces prediction errors for event-related pedestrian flow fluctuations, making the average pedestrian density prediction more closely reflect the actual scenario.

[0041] a3. Determine the cleaning rate correction coefficient for each sub-region based on the flow density level of each sub-region.

[0042] Specifically, a preset cleaning rate correction coefficient is obtained to adjust the baseline cleaning rate to reflect the actual cleaning difficulty. The correction coefficient is less than 1 for high flow density levels, equal to 1 for medium flow density levels, and greater than 1 for low flow density levels. The cleaning rate correction coefficient can be set through historical data regression analysis or expert experience and is dynamically optimized as data accumulates.

[0043] a4. Calculate the expected cleaning rate v_actual for each sub-region, including: 1. Statistical benchmark cleaning rate: Based on historical cleaning data, for each pollution level (low / medium / high pollution), multiple sets of key data of historical cleaning tasks (including area area and actual robot completion time) are selected for the area corresponding to that level. The cleaning rate of each task is calculated (cleaning rate = area area / actual robot completion time). Then, the average cleaning rate of all task groups is taken as the benchmark cleaning rate for the area of ​​that pollution level. 2. Correcting the expected cleaning rate v_actual: Combining the cleaning rate correction coefficients for each sub-area determined in step a3, the baseline cleaning rate for the corresponding pollution level is corrected to obtain the expected cleaning rate for each sub-area. The correction formula is: Expected cleaning rate v_actual = Baseline cleaning rate for the corresponding pollution level × Correction coefficient for the corresponding sub-area cleaning rate.

[0044] a5. Based on the expected cleaning rate v_actual and the preset maximum cleaning time T_max, calculate the maximum allowable area of ​​the grid cell in each sub-region: A_max = v_actual × T_max, where A_max is the maximum allowable area of ​​the grid cell. Where T_max ≥ preset maximum cleaning time + moving reference time, the specific definitions and explanations are as follows: Preset maximum cleaning time: refers to the longest time allowed by the system when the cleaning robot completes the cleaning task of a single grid cell with an area of ​​A_max; its value is determined based on the upper limit of the power of the cleaning robot on a single charge, ensuring that the robot can complete the cleaning task of a single grid cell without charging in the middle. Mobile baseline duration: refers to the fixed redundancy duration calculated based on historical data. Specifically, it is the average time for the cleaning robot to move from a representative starting point (such as a charging station, main road intersection, or rest area) to the center of its neighboring grid cell. The purpose of setting this duration is to reserve time for the robot to arrive at the task area (or travel back and forth between the starting point and the task area) and to avoid the total duration exceeding T_max due to movement time.

[0045] This solution introduces a correction coefficient to quantify the impact of different flow density levels on cleaning rates. For high flow density areas, the baseline cleaning rate is reduced, while for low flow density areas, the baseline cleaning rate is appropriately increased. This differentiated adjustment is deeply aligned with the actual pedestrian traffic patterns in the area, significantly improving the rationality of the rate settings.

[0046] a6. Based on the maximum allowable area of ​​the grid cells in each sub-region, an adaptive mesh generation algorithm is used to generate the grid cells for each sub-region. The specific steps are as follows: Differentiated initial partitioning: For each sub-region, first call the maximum allowable area A_max of the corresponding sub-region calculated in step a5 to determine the maximum size threshold of the grid cell as A_max. Then, prioritize the partitioning of rectangular grids. If the sub-region has irregular boundaries (such as park corners or green belts), then use quadtree partitioning or dynamic grid refinement algorithm to process them. After that, recursively partition the current grid cell. If the area of ​​the existing grid cell is greater than A_max, continue to split it until the area of ​​all grid cells is ≤ A_max.

[0047] Merging optimization: Traverse the divided grid cells. If adjacent grid cells simultaneously meet the following two conditions: (1) the total area is less than or equal to the A_max of the corresponding sub-region; (2) the pollution level is the same, then merge them into one grid cell.

[0048] Complete Grid Attribute Recording: Core attributes are entered for each final generated grid cell, including but not limited to: grid cell area, geographic coordinates, pollution level, pedestrian density level, and estimated cleaning time (calculation formula: estimated cleaning time = grid cell area / expected cleaning rate of sub-area v_actual). Then, based on the unique identifier of each grid cell, a data index is generated for its attribute information, and this attribute information is uploaded to the park's distributed storage node (such as IPFS or a private distributed storage cluster). After storage, the unique access address of the grid cell's attribute information in the distributed storage node is recorded, and the grid cell's identifier, corresponding data index, and corresponding distributed storage access address are bound together to form an association mapping table. This table is also synchronized to the park management system for subsequent queries and calls.

[0049] By constructing a differentiated initial partitioning logic using the maximum allowable area corresponding to each sub-region as the core threshold, the maximum allowable area is correlated with the pollution level and flow density. This allows high-pollution, high-flow-density sub-regions to form fine-grained grids, ensuring that the robot can concentrate resources to complete fine cleaning and avoid incomplete cleaning due to severe pollution or human interference. At the same time, the small area of ​​each individual grid ensures that the cleaning time is completed within the maximum allowable cleaning time. Low-pollution, low-flow-density sub-regions form coarse-grained grids, reducing the total number of grids and the frequency of robot task switching, increasing the cleaning coverage area per unit time, and avoiding resource waste. This design of allocating granularity on demand allows the grid partitioning depth to adapt to the cleaning difficulty of each sub-region.

[0050] In some implementations, the scheduling method further includes: dynamically adjusting the task generation frequency based on historical cleaning data and / or real-time environmental data to dynamically publish smart contracts for each cleaning task.

[0051] Specifically, in the first method, historical cleaning data of the target sub-area (or sub-areas of the same type, such as those with the same pollution level or the same functional zone) is retrieved from the distributed storage nodes of the park, and key data is filtered: after each cleaning task is completed, the historical time intervals in which the area rises from "clean state" (pollution level below the preset cleaning threshold) to "needs cleaning state" (pollution level reaches medium pollution) are recorded as T1, T2, ..., Tn; then the average value of all historical time intervals is calculated as T_avg=(T1+T2+...+Tn) / n; after that, the task generation base frequency F_base=1 / T_avg is calculated; Optionally, T_avg can be fine-tuned by combining the historical traffic density patterns of the same period: if the proportion of time periods with high traffic density levels in the same period of history is greater than the preset first threshold (e.g., 60%), then T_avg is lowered by k1%~k2% (e.g., 10%~20%) to obtain the calibrated base interval T_avg', and the adjusted task generation frequency F_base'=1 / T_avg'.

[0052] In the second approach, real-time pollution monitoring of each sub-area is conducted via the park's IoT sensing network (such as cameras and dust sensors), or periodic monitoring is performed at preset intervals (e.g., once every 10 minutes). Then, a pollution detection model (refer to step a1) is invoked to determine the pollution level of the sub-area in real time. If any of the following conditions are met, a task is immediately triggered: (1. The pollution level has reached the high pollution level;) (2) The pollution level remains at the medium pollution level for m monitoring cycles. As an example, m can be 3.

[0053] Furthermore, tasks triggered by high pollution levels have higher priority than those triggered by medium pollution levels. Priority tags are used when smart contracts are published to ensure that bots respond to high-urgency tasks first. Additionally, a minimum interval of at least 30 minutes can be set between trigger-based publications within the same sub-region to avoid repeated publications in a short period and wasting resources.

[0054] In the third implementation method, the first and second methods described above can be combined to form a hybrid scheduling mechanism, which ensures the regularity of cleaning in normal scenarios while also taking into account the timeliness of cleaning in emergency scenarios. The specific implementation steps are as follows: Base frequency preset: First, determine the base frequency (F_base or F_base') for task generation in the target sub-region using the first method, and start the countdown for scheduled task release at this frequency (for example, if F_base' = 1 time / 2 hours, then a scheduled release judgment will be triggered every 2 hours). Parallel real-time monitoring: During the timed countdown, the pollution level of the sub-area is monitored synchronously according to the monitoring rules of the second method (real-time / preset cycle), and the monitoring frequency is consistent with the second method (e.g., once / 10 minutes). Trigger priority determination includes: (1) If the sub-region meets any of the triggering conditions of the second method before the timed countdown ends, the current countdown will be terminated immediately, and the cleaning task smart contract will be published according to the rules of the second method (marking the corresponding priority); after the task is published, the next round of basic frequency countdown will be restarted; if the sub-region does not trigger any of the conditions of the second method after the timed countdown ends, the cleaning task smart contract will be published according to the basic frequency of the first method (priority set to normal level); after publication, the next round of basic frequency countdown will be started directly. (2) If the pollution level of the sub-area has not dropped to a clean state when the countdown of the next round of basic frequency starts after the second method of issuing the task is triggered, the interval T_avg (or T_avg') of the next round of basic frequency will be reduced by 20%~30% (further increase the frequency) until the pollution level of the area is restored to clean. (3) The minimum interval between “trigger-based release” and “timed release” in the same sub-region still follows the constraint of the preset time interval of the second method, that is, the minimum release interval ≥ the preset time interval. If the timed countdown ends and the timed release does not meet the constraint of the preset time interval from the last trigger-based release, the timed task will be released after the interval requirement is met.

[0055] S20. Each cleaning robot connected to the blockchain network responds to the smart contract, obtains local environmental information and its own state information of the target grid cell, and bids based on its own state information, local environmental information, and the pheromone concentration sensed from the target grid cell; where the target grid cell is the grid cell corresponding to the cleaning task. Furthermore, the pheromone concentration is used to indicate the cleaning urgency of the target grid cell, allowing other cleaning robots to adjust their bidding strategies based on the pheromone concentration.

[0056] Specifically, the key technical term "pheromone" will be explained below: Pheromones: Derived from the concept of pheromones in ant colony algorithms, they are used in this application to quantify the attractiveness of cleaning tasks to cleaning robots. Their core characteristic is that "concentration is negatively correlated with attractiveness": the lower the pheromone concentration, the more the task is not covered and the higher the urgency of cleaning, and the stronger the attraction to the cleaning robot; the higher the concentration, the more or less the task is covered and the lower the urgency of cleaning, and the weaker the attraction to the cleaning robot, thereby achieving implicit coordination and load balancing among cleaning robots.

[0057] Meanwhile, in step S20, the local environmental information may include, but is not limited to: the location information, area, pollution level, and population density level of the target grid cell.

[0058] Furthermore, in this embodiment, the self-state information includes self-location information, which at least includes: (1) Real-time positioning parameters: real-time GPS coordinates, which are collected in real time by the GPS module carried by the robot; (2) Energy status parameters: The current remaining battery power E_battery is monitored in real time by the robot battery management system and used for subsequent energy consumption safety verification; (3) Performance calibration parameters: rated moving power P_move, rated cleaning power P_clean, and rated moving speed v_move, used to accurately calculate moving energy consumption and cleaning energy consumption.

[0059] After the robot responds to the smart contract, it obtains the access address of the distributed storage node corresponding to the target grid cell from the smart contract (this address is bound to the smart contract by the grid attribute storage process in step a6); it accesses the distributed storage node of the park through this access address and retrieves the attribute information of the target grid cell (including pollution level, traffic density level, and expected cleaning rate).

[0060] If the distributed storage node access is abnormal (such as network interruption or data lag), the local real-time computing logic is triggered: through the real-time interface of the park's IoT sensing system (such as infrared sensor data gateway, access control statistics interface, and robot-mounted camera), real-time environmental data (image information, dust detection information, and current pedestrian flow statistics) of the target grid unit is collected, and then local computing is performed: based on the pre-agreed computing rules in the smart contract (fully reusing the pollution detection logic of step a1 and the traffic density level classification logic of step a2), a lightweight algorithm model is locally called to quickly calculate the pollution level and pedestrian flow density level of the target grid unit, ensuring that the subsequent process is not interrupted.

[0061] Simultaneously, in step S20, bidding is conducted based on its own state information, local environment information, and the pheromone concentration sensed from the target grid cell, including: Each cleaning robot responds to the smart contract and performs the following operations: Based on its own location information and the location information of the target grid cell, the basic distance from the cleaning robot to the target grid cell is calculated. The estimated cleaning time is determined based on the area of ​​the target grid cell; Obtain the rated moving speed, rated moving power, and rated cleaning power of the cleaning robot; The base distance and rated moving speed are adjusted based on the pedestrian density level to obtain the adjusted moving distance and adjusted moving speed. Based on the adjusted moving speed and adjusted moving distance, calculate the moving time required to move to the target grid cell; Calculate mobile power consumption based on rated mobile power and mobile duration; Calculate the cleaning power consumption based on the rated cleaning power and the estimated cleaning duration; Calculate the total energy consumption required to perform the cleaning task based on the moving power consumption and the cleaning power consumption; Obtain the task reward set in the smart contract; Obtain the pheromone concentration sensed from the target grid cell; Determine the bidding price based on the task reward, the basic distance, the total energy consumption, and the pheromone concentration.

[0062] Specifically, in the above steps, when calculating the basic distance, the Haversine formula is used to calculate the straight-line distance from the location of the cleaning robot to the center of the target grid cell to obtain the basic distance.

[0063] In the above steps, when calculating the estimated cleaning duration, the cleaning robot obtains the access address of the distributed storage node corresponding to the target grid cell from the smart contract (this address is bound to the smart contract by the grid attribute storage process in step a6), accesses the园区 distributed storage node through this access address, and retrieves the attribute information of the target grid cell, including the expected cleaning rate v_actual, the pollution level, and the pedestrian flow density level of the target grid cell. If the access to the distributed storage node is abnormal (such as network interruption, data lag), the local real-time calculation logic is triggered: through the real-time interface of the园区 Internet of Things perception system (such as the infrared sensor data gateway, the access control statistics interface, the camera carried by the robot), the real-time environmental data of the target grid cell is supplemented, and then based on the pre-agreed calculation rules in the smart contract (fully reusing the pollution detection logic in step a1 and the traffic density level division logic in step a2), the lightweight algorithm model is locally called to quickly calculate the pollution level and the pedestrian flow density level of the target grid cell.

[0064] After determining the expected cleaning rate v_actual, use the formula t_clean = A_j / v_actual. Where A_j represents the area of the target grid cell, and t_clean represents the estimated cleaning duration.

[0065] In the above steps, when correcting the basic distance in combination with the pedestrian flow density level, the distance correction coefficient q is set as follows: If the pedestrian flow density level is high flow, then q = q1 (q1 > 1, for example, 1.2 - 1.5, to adapt to the need for detouring around dense pedestrians); If the pedestrian flow density level is medium flow, then q = q2 (q2 > 1 and q2 < q1, for example, 1.05 - 1.1, to adapt to the need for detouring around a small number of pedestrians); If the pedestrian flow density level is low flow, then q = q3 = 1 (no detour, drive in a straight line).

[0066] The adjusted moving distance \(d_{adjust} = q\times d_{ij}\), where \(d_{ij}\) represents the base distance.

[0067] In the above steps, when correcting the moving speed of the robot in combination with the crowd flow density level, the speed correction coefficient \(k\) is set as follows: If the crowd flow density level is high flow, then \(k = k1\) (\(0 < k1 < k2\), for example \(0.5 - 0.7\), adapting to the low-speed avoidance requirement); If the crowd flow density level is medium flow, then \(k = k2\) (\(0 < k2 < k3\), for example \(0.8 - 0.9\), adapting to the normal driving requirement); If the crowd flow density level is low flow, then \(k = k3 = 1\) (driving at the rated speed, adapting to the interference-free scenario); The adjusted moving speed \(v_{adjust} = k\times v_{move}\), where \(v_{move}\) represents the rated moving speed and \(v_{adjust}\) represents the adjusted moving speed.

[0068] Then, calculate the moving duration through the formula \(t_{move} = d_{adjust} / v_{adjust}\). Here, \(t_{move}\) represents the moving duration and \(t_{clean}\) represents the estimated cleaning duration.

[0069] In the above solution, the moving power consumption \(E_{move} = P_{move}\times t_{move}\), the cleaning power consumption \(E_{clean} = P_{clean}\times t_{clean}\), and the total energy consumption \(E_j = E_{move} + E_{clean}\). Here, \(P_{move}\) represents the rated moving power and \(P_{clean}\) represents the rated cleaning power.

[0070] In the above embodiment, the real-time pheromone concentration \(P_j\) sensed from the target grid cell, this concentration will increase with time or the progress of task completion, the higher the concentration indicates the higher the degree of task coverage, and the attraction to other cleaning robots should be correspondingly reduced, that is, the pheromone concentration and the bidding price are negatively correlated.

[0071] More specifically, based on the task reward, base distance, total energy consumption, and pheromone concentration, determine the bidding price, including: Determine the bidding price through the following formula: , Where, represents the bidding price of cleaning robot \(i\) for task \(j\), represents the reward of task \(j\), represents the base distance from cleaning robot \(i\) to the target grid cell corresponding to task \(j\), represents the total energy consumption required for cleaning robot \(i\) to execute task \(j\), represents the pheromone concentration sensed from the target grid cell corresponding to task \(j\), These represent the weighting coefficients corresponding to rewards, base distance, total energy consumption, and pheromone concentration, respectively.

[0072] This model achieves multi-objective optimization. And the task reward... Incentivize robot participation, basic distance Total energy consumption This leads to tasks being assigned to robots that are closer and consume less energy (system efficiency optimization), while pheromone concentration... This achieves implicit coordination; as the concentration increases, it automatically suppresses bids from other cleaning robots, thereby reducing redundant clustering of robots in the same area and dynamically balancing the load. Ultimately, the smart contract typically assigns the task to the cleaning robot that bids the highest (representing the most efficient).

[0073] In the above scheme, through the blockchain distributed bidding and pheromone coordination mechanism, the robot autonomously bids based on distance, energy consumption, and real-time task urgency, so as to achieve task allocation nearby, optimal energy consumption, and automatic avoidance of multi-robot aggregation, thereby improving the overall collaborative efficiency of the multi-robot system.

[0074] In some optional implementations, the robot's own state information may also include the current remaining battery power, E_battery. After calculating the total energy consumption required to perform the cleaning task of the target grid cell, a power safety check is performed: a safety factor k_safe is set (e.g., 0.9, reserving emergency power). If the current remaining battery power E_battery × k_safe ≥ E_j, then the robot has the power to complete the task and can continue to participate in subsequent bidding; otherwise, it abandons the bidding to avoid interruption of the task. By performing energy consumption estimation and power verification before the robot bids, it ensures that the task is executed within a safe power range. Combined with a movement time reservation mechanism, this effectively prevents the battery from running out during task execution and ensures the continuity of operations.

[0075] S30, the blockchain network determines the winning target cleaning robot based on preset consensus rules.

[0076] In some implementations, the process includes: a blockchain network verifying the bidding bids of all cleaning robots, determining the cleaning robot with the highest bid through a consensus mechanism to obtain the right to undertake the cleaning task, designating the cleaning robot that obtains the right to undertake the cleaning task as the target cleaning robot, and writing the bidding bid information and undertaking result of the target cleaning robot into the blockchain after verification by multiple nodes.

[0077] More specifically, the verification nodes in the blockchain network validate all received bidding information. This validation includes, but is not limited to: the legality of the digital signature of the bidding information, the validity of the bidding robot's identity, the conformity of the bidding format with the smart contract specifications, and whether the associated task is still available for bidding. After successful verification, the network adjudicates all valid bids according to the consensus rules preset in the smart contract, confirming the sole bidder with the highest bid. The cleaning robot confirmed as having the highest bid wins the right to undertake the cleaning task and is designated as the "target cleaning robot."

[0078] Furthermore, the blockchain network generates a "confirmation transaction" containing the task identifier, the target cleaning robot identifier, the winning bid, and a timestamp. This transaction is broadcast throughout the blockchain network and, after consensus verification by multiple nodes, is formally written into the blockchain's distributed ledger as a new, immutable block. Thus, the task allocation result is public, transparent, and permanently documented.

[0079] Simultaneously, the smart contract status is automatically updated, marking the corresponding grid cell task as "assigned" or "in execution," and locking the information of the winning target cleaning robot to prevent duplicate assignment. The blockchain network publishes the bidding results in the form of event notifications, allowing the target cleaning robot and other cleaning robots in the blockchain network to know the allocation results in real time, ensuring system status synchronization.

[0080] Through the above process, task allocation is driven entirely by a decentralized consensus mechanism, achieving fair, automated, and auditable scheduling decisions, and ensuring the consistency and reliability of the entire system state.

[0081] S40. The target cleaning robot performs cleaning tasks and releases pheromones in the target grid cells during the task execution.

[0082] Specifically, this includes: the target cleaning robot dynamically adjusts the pheromone release rate according to the real-time cleaning speed, so that the pheromone concentration changes with the release rate and time; or, the target cleaning robot releases pheromones according to the completion progress of the cleaning task, wherein the pheromone concentration is positively correlated with the completion progress.

[0083] More specifically, the target cleaning robot uses its onboard sensors (such as an odometer and a cleaning brush speed sensor) to collect the real-time cleaning rate v_real of the current cleaning operation; based on a preset mapping relationship between real-time cleaning rate and pheromone release rate, it dynamically adjusts the pheromone release rate r_release. The pheromone concentration P_j changes dynamically with the release rate and release time, satisfying the formula: Where t is the release duration, which is accumulated from the start of the task to ensure that the concentration change truly reflects the cleaning efficiency.

[0084] Alternatively, the target cleaning robot calculates the completion progress S of the cleaning task in real time, where S = cleaned area A_completed / total area of ​​the target grid cell A_j. The pheromone concentration P_j is positively correlated with the completion progress S. The specific correlation rules are as follows: When the task starts (S=0), the pheromone concentration P_j = P_init, where P_init represents the initial concentration, which is a threshold value that approaches 0, indicating the highest urgency of cleaning. During the task execution (0 < S < 1), the pheromone concentration increases linearly or non-linearly. For example, the linear relationship is: P_j = P_init + (P_sat - P_init) × S, where P_sat is the saturation concentration, which is a preset value that indicates that the cleaning is basically completed. When the task is completed (S=1), the pheromone concentration reaches P_sat and is maintained for a preset duration (e.g., 5~10 minutes) to ensure that other cleaning robots clearly understand that the target grid cell has been cleaned.

[0085] S50: The blockchain network verifies the cleaning task execution status of the target grid unit. Once verification is successful, the task reward is distributed to the virtual account of the target cleaning robot via a smart contract. Specifically, tokens or points can be issued as rewards through the digital wallet of the target cleaning robot's virtual account.

[0086] Meanwhile, in step S50, the blockchain network verifies the cleaning task execution status of the target grid cell, including: the blockchain network acquires images of the target grid cell through a camera device (such as a high-definition camera) deployed near the target grid cell; if it is determined from the images that the target grid cell has been cleaned, the task is determined to have been successfully executed.

[0087] More specifically, upon receiving the task completion report request from the target cleaning robot, the blockchain network sends image acquisition instructions and dust detection instructions to the camera devices and dust sensors deployed at preset locations around the target grid unit (ensuring no blind spots in the coverage of the grid unit, coordinating with the sensor deployment in step a1 above). The camera devices respond to the image acquisition instructions by acquiring image data of the target grid unit in real time (acquiring at least 3 angles, covering all areas of the grid unit), and encrypt and upload the image data to the blockchain network (using hash value verification to ensure the data has not been tampered with). Similarly, the dust sensors respond to the dust detection instructions by acquiring dust data of the target grid unit in real time and encrypting and uploading it to the blockchain network. The blockchain network calls a preset cleaning effect judgment model (with the same implementation principle as the pollution detection model in step a1 above) to analyze the acquired image and dust data. If the model determines that the pollution level of the target grid unit is lower than the preset cleaning threshold, it outputs a verification result of "task successfully executed"; if the model determines that the pollution level is higher than or equal to the preset cleaning threshold, it outputs a verification result of "task failed," and sends a "cleaning unqualified" message to the target cleaning robot via a smart contract.

[0088] Furthermore, to improve the accuracy of verification, auxiliary verification can be performed by combining the task execution data of the target cleaning robot (such as cleaning path coverage, the matching degree between actual cleaning energy consumption and estimated energy consumption, and pheromone release saturation state). Only when the image verification result is consistent with the auxiliary verification result can the final judgment be "task execution successful".

[0089] Therefore, the present application provides a scheduling method for cleaning robots. This scheduling method integrates IoT sensing, AI prediction, and blockchain smart contracts to achieve dynamic, reliable, and efficient distributed cleaning task scheduling, significantly improving the overall efficiency and system adaptability of cleaning operations.

[0090] Meanwhile, by building a decentralized and trustworthy task distribution platform and eliminating the central scheduling node, the risk of single point of failure in the traditional centralized scheduling model is effectively avoided. Even if individual cleaning robots malfunction or exit the system, it will not affect the overall scheduling process, significantly improving the robustness and scalability of the system.

[0091] In addition, each cleaning robot bids autonomously based on its own status, task rewards, and the pheromone concentration perceived from the environment. Through blockchain smart contracts and the introduction of an incentive-based bidding mechanism, the autonomous decision-making ability of each cleaning robot is stimulated, enabling it to optimize the efficiency of global task allocation through local information interaction while pursuing individual benefits. Furthermore, leveraging the immutability, transparency, and traceability inherent in blockchain network technology ensures the openness, fairness, and trustworthiness of the entire process from task release, bidding, execution to reward settlement. Moreover, by using pre-set consensus rules, tasks are automatically, fairly, and transparently allocated, laying a reliable foundation of trust for shared cleaning scenarios involving multiple operators.

[0092] Secondly, by introducing pheromones as a distributed coordination mechanism, the robot can perceive the status of the work area in real time (such as whether other robots are already cleaning) and dynamically adjust its own behavior through bidding, thereby achieving autonomous collaboration and load balancing among multiple robots. Even if some robots or network nodes malfunction, the entire system can still operate continuously and stably.

[0093] Secondly, embodiments of this application disclose a computer-readable storage medium storing a computer program that, when executed by a processor, performs the scheduling method for the cleaning robot provided in the first aspect.

[0094] Specifically, in this embodiment, the computer-readable storage medium has computer-readable program instructions (i.e., computer programs) stored thereon, which are used to execute the scheduling method of the cleaning robot in the above embodiment.

[0095] More specifically, the computer-readable storage medium provided in this application may be a USB flash drive, but is not limited to an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or any combination thereof. More specific examples of such a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0096] In this embodiment, the computer-readable storage medium can also be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. Furthermore, the program code contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof. The aforementioned computer-readable storage medium can also be included in an electronic device; or it can exist independently without being assembled into an electronic device.

[0097] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0098] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions. Modules described in the embodiments of this application can be implemented in software or hardware. The names of modules do not, in some cases, constitute a limitation on the unit itself.

[0099] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., a computer program) for executing the scheduling method of the cleaning robot described above. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those provided in the above embodiments, and will not be repeated here.

[0100] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A scheduling method for a cleaning robot, characterized in that: Includes the following steps: S10. Divide the area to be cleaned into multiple grid units, then encapsulate the cleaning tasks of each grid unit into smart contracts and publish them to the blockchain network. S20. Each cleaning robot connected to the blockchain network responds to the smart contract, obtains the local environmental information and its own state information of the target grid unit, and bids based on its own state information, the local environmental information, and the pheromone concentration sensed from the target grid unit; wherein, the target grid unit is the grid unit corresponding to the cleaning task. S30. The blockchain network determines the winning target cleaning robot based on preset consensus rules; S40. The target cleaning robot performs the cleaning task and releases pheromones in the target grid cell during the task execution. S50. The blockchain network verifies the cleaning task execution status of the target grid unit. After successful verification, the task reward is distributed to the virtual account of the target cleaning robot through the smart contract.

2. The scheduling method for a cleaning robot according to claim 1, characterized in that: In step S10, dividing the area to be cleaned into multiple grid units includes: Based on historical cleaning data and / or real-time environmental data, the granularity of the grid units in the park is dynamically adjusted to obtain multiple grid units.

3. The scheduling method for a cleaning robot according to claim 1, characterized in that: The scheduling method further includes: Based on historical cleaning data and / or real-time environmental data, the task generation frequency is dynamically adjusted to dynamically publish smart contracts for each cleaning task.

4. The scheduling method for a cleaning robot according to claim 1, characterized in that: In step S20, the self-state information includes its own position information; The local environmental information includes the location information, area, and pedestrian density level of the target grid cell; The bidding process based on the self-state information, the local environment information, and the pheromone concentration sensed from the target grid cell includes: In response to the smart contract, each of the cleaning robots performs the following operations: Based on its own position information and the position information of the target grid cell, the basic distance from the cleaning robot to the target grid cell is calculated; The estimated cleaning time is determined based on the area of ​​the target grid cell; Obtain the rated moving speed, rated moving power, and rated cleaning power of the cleaning robot; The base distance and the rated moving speed are corrected based on the pedestrian density level to obtain the adjusted moving distance and the adjusted moving speed. Based on the adjusted moving speed and the adjusted moving distance, calculate the moving time required to move to the target grid cell; Calculate the mobile power consumption based on the rated mobile power and the mobile duration; Calculate the cleaning power consumption based on the rated cleaning power and the estimated cleaning time; Based on the mobile power consumption and the cleaning power consumption, calculate the total energy consumption required to perform the cleaning task; Obtain the task reward set in the smart contract; Obtain the pheromone concentration sensed from the target grid cell; The bidding price is determined based on the task reward, the base distance, the total energy consumption, and the pheromone concentration.

5. The scheduling method for a cleaning robot according to claim 4, characterized in that: The process of determining the bidding price based on the task reward, the base distance, the total energy consumption, and the pheromone concentration includes: The bid price is determined using the following formula: ； in, This represents the bid offered by cleaning robot i for task j. This represents the reward for task j. This represents the basic distance from cleaning robot i to the target grid cell corresponding to task j. This represents the total energy consumption required for cleaning robot i to perform task j. This represents the pheromone concentration sensed from the target grid cell corresponding to task j. These represent the weighting coefficients corresponding to rewards, base distance, total energy consumption, and pheromone concentration, respectively.

6. The scheduling method for a cleaning robot according to claim 5, characterized in that: The pheromone concentration is negatively correlated with the bidding price.

7. The scheduling method for a cleaning robot according to claim 1, characterized in that: In step S30, the blockchain network determines the winning target cleaning robot based on preset consensus rules, including: The blockchain network verifies the bidding bids of all cleaning robots, determines the cleaning robot with the highest bid through a consensus mechanism, and selects the cleaning robot that has the right to undertake the cleaning task. The cleaning robot that has the right to undertake the cleaning task is designated as the target cleaning robot. The bidding bid information and the undertaking result of the target cleaning robot are then verified by multiple nodes and written into the blockchain.

8. The scheduling method for a cleaning robot according to claim 1, characterized in that: In step S40, the target cleaning robot performs the cleaning task and releases pheromones in the target grid cell during the task execution, including: The target cleaning robot dynamically adjusts the pheromone release rate according to the real-time cleaning speed, so that the pheromone concentration changes with the release rate and time. Alternatively, the target cleaning robot releases pheromones according to the progress of the cleaning task, wherein the concentration of the pheromones is positively correlated with the progress of the task.

9. The scheduling method for a cleaning robot according to claim 1, characterized in that: In step S50, the blockchain network verifies the cleaning task execution status of the target grid cell, including: The blockchain network acquires images of the target grid cell through camera devices deployed near the target grid cell; If the target grid cell is determined to be clean based on the image, the task is considered to have been successfully executed.

10. A computer-readable storage medium, characterized in that, The system includes a computer program that, when executed by a processor, performs a scheduling method for a cleaning robot as described in any one of claims 1 to 9.

Images