A heterogeneous swarm consensus decision-making and air-ground collaborative scheduling method, device and system thereof

Abstract

This invention discloses a heterogeneous crowdsourced intelligence consensus decision-making and air-ground collaborative scheduling method, device, and system for urban road network control. The method utilizes heterogeneous crowdsourced intelligence sensing information to achieve statistical consensus on multi-UAV control suggestion signals, employs a spatial weighted aggregation method to determine candidate control center locations, and schedules road control by ground execution units under control constraints. The device includes a processor and a terminal device with computer-readable storage media. The system includes a heterogeneous crowdsourced intelligence sensing information fusion module, a UAV swarm module, a consensus computing module, an aggregation computing module, a road network mapping module, a constraint control module, and an execution scheduling module. Compared with existing technologies, this invention improves the stability and collaborative efficiency of road network control decisions in complex urban environments, enhances environmental situational awareness in complex urban environments, and provides a reliable information foundation for subsequent consensus decision-making and collaborative execution.

Description

Technical Field

[0001] This invention relates to the field of urban public safety management and unmanned system collaborative control technology, and in particular to a heterogeneous crowd intelligence consensus decision-making and air-ground collaborative scheduling method, device and system for urban road network control. Background Technology

[0002] In scenarios such as urban public safety incident response, emergency pursuit, and dynamic target control, it is often necessary to continuously track multiple moving targets and combine this with road space constraints to limit the target's activity range, thereby improving the efficiency of target control and response. Especially in complex urban road network environments, target movement is characterized by diverse paths, rapid transfer, and strong evasion capabilities. Traditional methods relying solely on single perception or single-point decision-making are insufficient to meet the demands for real-time, stable, and efficient road closures.

[0003] Currently, target tracking and road control in urban environments typically employ the following methods: 1) Single-UAV tracking: This method utilizes a single UAV to track and monitor targets. However, due to limited sensing range, it is prone to blind spots and struggles to create effective spatial containment in complex urban road networks. 2) Fixed monitoring equipment: This method relies on existing urban monitoring equipment to monitor targets. However, it is limited by the deployment location and coverage of the equipment, lacking flexible dynamic response capabilities. 3) Manual road dispatching: This method involves manually deciding and dispatching ground roads. However, the overall response speed is slow, making it difficult to meet real-time dynamic control requirements. 4) Multi-UAV collaborative tracking: This method utilizes multiple UAVs for collaborative sensing and joint tracking of targets. However, if a single UAV independently issues a blockade command, it may lead to frequent road closures due to misjudgments of the local environment, thereby reducing the overall resource utilization efficiency of the system. Furthermore, while existing research has begun to focus on multi-source sensing fusion and heterogeneous agent collaboration, aiming to improve the reliability of target detection and state estimation in complex environments by fusing information from UAVs, ground monitoring facilities, and mobile sensing agents, most existing solutions still primarily remain at the level of information fusion or local collaboration. They focus more on continuous sensing, tracking, or pursuit strategies, with insufficient attention paid to how to actively shape the target's movement space, achieve real-time road closures, and air-ground collaborative constraint control. At the same time, existing solutions typically treat sensing results as passive input, failing to effectively establish a closed-loop collaborative mechanism between sensing, decision-making, scheduling, and execution.

[0004] In summary, existing lockdown technologies suffer from several shortcomings: a lack of a consensus-based decision-making constraint mechanism for lockdown decisions; a lack of reasonable spatial geometric optimization methods for determining lockdown locations; a lack of coordinated control mechanisms for resource and time constraints in the lockdown triggering process; and insufficient fusion and utilization of multi-source heterogeneous sensing information. At least the following deficiencies exist: 1) The lack of a consensus-based decision-making constraint mechanism for lockdown decisions makes them susceptible to single-point misjudgments; 2) The lack of reasonable spatial geometric optimization methods for determining lockdown locations makes it difficult to accurately select suitable target lockdown routes; 3) The lack of a coordinated constraint mechanism for the number of lockdowns, time intervals, and execution resource status in the lockdown triggering process easily leads to ineffective lockdowns or scheduling conflicts; 4) Existing urban public safety systems typically rely on single-type sensing devices and lack the fusion and utilization of multi-source heterogeneous sensing information such as drones, fixed monitoring equipment, and crowd information, making it difficult to form a complete urban environmental situational awareness capability, thus affecting the accuracy and efficiency of subsequent collaborative decision-making.

[0005] Therefore, it is necessary to propose a heterogeneous crowd-based consensus decision-making and air-ground collaborative scheduling method, device and system for urban road network control, so as to improve the stability, accuracy and collaborative execution efficiency of road network control decisions in complex urban environments. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by providing a heterogeneous crowd-based consensus decision-making and air-ground collaborative scheduling method for urban road network control. This method integrates heterogeneous crowd-based sensing information from drones, fixed monitoring equipment, and crowd data to achieve statistical consensus on multi-drone control suggestion signals. Road control decisions are triggered based on consensus strength, and spatial aggregation methods are used to determine candidate control center locations. Furthermore, a city road network model is used to map control boundaries and select target roads. Under constraints on the number of control points, time cooldown, and execution resources, ground execution units are scheduled to implement road control. This forms an air-ground collaborative closed-loop control mechanism for complex urban road network environments, improving the stability, accuracy, and collaborative execution efficiency of road control decisions. The method is simple and has promising application prospects.

[0007] The specific technical solution to achieve the purpose of this invention is: a heterogeneous crowd-based consensus decision-making and air-ground collaborative scheduling method for urban road network control, characterized by the following steps:

[0008] Step S0: Construct a heterogeneous swarm intelligence sensing information fusion mechanism

[0009] The system integrates drone perception information, neighboring drone communication information, fixed monitoring equipment perception information, crowd perception information, and road boundary or closure status information to form comprehensive drone observation information, providing an information foundation for subsequent closure suggestion signal generation, crowd consistency statistics, and road closure decisions.

[0010] Step S1: Generation of Lockdown Recommendation Signal

[0011] Based on the comprehensive observation information, each UAV generates a road closure suggestion signal to determine whether road closures are necessary in the current area. The road closure suggestion signal is used to characterize the corresponding UAV's judgment result on implementing road closures in the current area.

[0012] Step S2: Group Consistency Statistics

[0013] Collective consensus statistics are performed on all the blockade suggestion signals generated by drones, and the consensus strength value is calculated to represent the number of drones that suggest implementing road blockade at the current moment.

[0014] Step S3: Lockdown Decision

[0015] When the consensus strength value is greater than or equal to a preset threshold, the road closure decision process is triggered; otherwise, the current closure process is terminated, thereby reducing the impact of local misjudgments by a single drone on the overall closure decision through group consistency constraints.

[0016] Step S4: Aggregation Calculation of Candidate Lockdown Center Points

[0017] Spatial aggregation calculations are performed on the locations of drones that propose lockdown suggestions to obtain candidate lockdown center points, which characterize the spatial concentration trend of multiple drones implementing road lockdowns in the target area.

[0018] Step S5: Candidate lockdown location mapping

[0019] The candidate lockdown center point is mapped to the corresponding grid cell in the urban road network model, and the set of boundary roads corresponding to the grid cell is obtained to establish the mapping relationship between the candidate lockdown location and the specific road boundary.

[0020] Step S6: Determine the target road

[0021] Calculate the geometric distance between the candidate lockdown center point and each boundary road in the set of boundary roads, and select the road boundary with the smallest distance as the target lockdown road, thereby realizing the transformation from the candidate lockdown area to the specific road lockdown object.

[0022] Step S7: Trigger constraint judgment

[0023] Determine whether the restrictions on the number of blocks, the cooldown time, and the availability of execution resources are met; if the triggering conditions are met, proceed with the execution scheduling process; if the triggering conditions are not met, terminate the current round of block execution to avoid frequent block triggering or resource conflicts.

[0024] Step S8: Ground Execution Unit Scheduling

[0025] Send a blockade command to the ground execution unit to put the target blockade road into a blockade state and update the blockade road set, thereby realizing the air-ground coordinated closed-loop control of the road blockade task.

[0026] Furthermore, in step S0, the construction of the heterogeneous swarm intelligence perception information fusion mechanism specifically includes: acquiring the UAV's own position and motion status information of the target; acquiring the position, motion status, or target observation information within the communication range of neighboring UAVs; acquiring the target status information provided by fixed monitoring equipment and crowds; acquiring the passable, blocked, or closed status of adjacent boundary roads of the current road network unit; and unifying and fusing the above information to form a comprehensive UAV observation vector, which is used to support the subsequent generation of closed-off suggestion signals and collaborative decision-making.

[0027] Furthermore, in step S2, when performing group consistency statistics on the blockade suggestion signals generated by all drones, a consensus strength value is obtained by uniformly counting the number of drones that suggest implementing blockade; in step S3, the consensus strength value is compared with a preset threshold, and when the consensus strength value reaches the preset threshold, the subsequent blockade decision process is entered, thereby forming a blockade triggering mechanism based on the consensus of multi-drone group perception.

[0028] Furthermore, in step S4, spatial aggregation calculations are performed on the locations of the drones that propose containment suggestions to obtain candidate containment center points; the candidate containment center points are used to characterize the comprehensive judgment center of multiple drones in space, so as to improve the rationality and stability of containment location selection.

[0029] Furthermore, in steps S5 and S6, by mapping the candidate lockdown center point to the corresponding grid cell in the urban road network model, the set of adjacent boundary roads is obtained, and the target lockdown road is determined according to the geometric distance between the candidate lockdown center point and each boundary road, thereby improving the accuracy of lockdown road selection.

[0030] Furthermore, in steps S7 and S8, the blockade execution conditions are determined by jointly judging the number of currently blocked roads, the time interval between the current time and the last blockade trigger time, and the idle state of the ground execution unit. When the trigger conditions are met, a blockade command is sent to the ground execution unit to make the target blocked road enter the blockade state and update the blockade road set to form a collaborative execution closed loop.

[0031] The present invention also provides a heterogeneous crowd-based consensus decision-making and air-ground coordinated scheduling device for urban road network control. The device is characterized in that it is a terminal device including a processor and a computer-readable storage medium. The computer-readable storage medium stores multiple instructions, which are loaded and executed by the processor to implement the above-mentioned heterogeneous crowd-based consensus decision-making and air-ground coordinated scheduling method for urban road network control.

[0032] This invention also provides a heterogeneous swarm intelligence consensus decision-making and air-ground collaborative scheduling system for urban road network control. The system comprises: a heterogeneous swarm intelligence sensing information fusion module, a drone swarm module, a consensus computing module, an aggregation computing module, a road network mapping module, a constraint control module, and an execution scheduling module. The heterogeneous swarm intelligence sensing information fusion module is used to fuse drone sensing information, neighboring drone communication information, fixed monitoring equipment sensing information, crowd sensing information, and road boundary or control status information to form comprehensive observation information. The drone swarm module is used to generate control suggestion signals based on the comprehensive observation information. The consensus calculation module receives the lockdown suggestion signals from each UAV and calculates the consensus strength value; the aggregation calculation module performs aggregation calculations on the locations of the UAVs that made lockdown suggestions when the lockdown triggering conditions are met, to obtain candidate lockdown center points; the road network mapping module maps the candidate lockdown center points to the urban road network model and determines the target lockdown road; the constraint control module determines whether the lockdown quantity limit, cooldown time limit, and execution resource availability constraint are met; the execution scheduling module sends a lockdown command to the ground execution unit when the triggering constraint conditions are met, so that the target lockdown road enters the lockdown state.

[0033] Compared with the prior art, the present invention has the following beneficial technical effects and represents a current technological advancement:

[0034] 1) By introducing a group consensus statistical mechanism, the local judgment results of multiple drones are statistically analyzed and integrated for decision-making, which effectively reduces the impact of single-point misjudgment on the control decision and improves the overall decision-making stability of the system;

[0035] 2) By combining spatial aggregation calculation with road network mapping, the accuracy of candidate lockdown location determination and target lockdown road selection is improved, thereby enhancing the spatial rationality of lockdown decisions;

[0036] 3) By setting constraints on the number of blocks, time cooldown constraints, and execution resource constraints, the block triggering process is jointly controlled to reduce invalid block behaviors and improve system resource utilization efficiency;

[0037] 4) By coordinating and scheduling between UAV swarms and ground execution units, air-ground closed-loop control of road closure tasks can be achieved, improving the execution efficiency of multi-objective spatial constraints in complex urban environments;

[0038] 5) By introducing a heterogeneous swarm intelligence sensing information fusion mechanism, the collaborative integration of UAV sensing, fixed monitoring equipment information and crowd information can be achieved, thereby improving the environmental situational awareness capability in complex urban environments and providing a reliable information foundation for subsequent consensus decision-making and collaborative execution. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of a city road network control and coordination scenario in Example 1;

[0040] Figure 2 This is a schematic diagram of heterogeneous swarm intelligence sensing information fusion in Example 1;

[0041] Figure 3 This is a schematic diagram of multi-drone collaborative voting and consensus decision-making in Example 1;

[0042] Figure 4 This is a schematic diagram of the aggregation calculation of candidate lockdown center points in Example 1;

[0043] Figure 5 This is a schematic diagram of the candidate lockdown locations mapped to the road network boundary in Example 1. Detailed Implementation

[0044] The present invention specifically includes the following steps:

[0045] Step S0: Construct a heterogeneous swarm intelligence sensing information fusion mechanism

[0046] The system integrates UAV perception information, neighboring UAV communication information, and road boundary status information to form a comprehensive observation vector for the UAV. The observation information from the drone is expressed by the following formula:

[0047]

[0048] .

[0049] in, For the first The comprehensive observation vector of the UAV; For structured splicing; To aggregate sensory information for heterogeneous crowd intelligence, For communication information of nearby drones; Information on road boundaries or road closure status; For the first The adjacent grid cell where the drone is currently located is the The status of the boundary roads, where 1 indicates blocked or closed, and 0 indicates passable; This represents the number of adjacent boundary roads of the current grid cell.

[0050] By integrating the aforementioned information, a comprehensive environmental observation system is constructed, which is then used to generate blocking suggestion signals. Through the heterogeneous swarm intelligence sensing information fusion mechanism, drones can acquire more complete information about the urban environmental situation, thereby improving the accuracy and stability of subsequent group consensus decision-making.

[0051] Step S1: Generation of Lockdown Recommendation Signal

[0052] Each drone generates a road closure recommendation signal based on its own local perception information to determine whether road closures are necessary in its current area. ,in 0 indicates that a lockdown is not recommended, and 1 indicates that a lockdown is recommended.

[0053] Step S2: Perform group consensus statistics on the containment recommendation signals of all drones and calculate the consensus strength value: ,in For indicator functions, This refers to the number of drones.

[0054] Step S3: When consensus strength Greater than or equal to the preset threshold If the conditions are met, the lockdown decision-making process is triggered; otherwise, the current lockdown process is terminated. .

[0055] Step S4: Perform weighted aggregation of the locations of all drones that proposed lockdown suggestions, and calculate the candidate lockdown center point:

[0056] .

[0057] in, Indicates the candidate lockdown center point; For the first Spatial coordinates of the drone when it made the containment proposal (can be taken as the current location); This represents the consensus strength value.

[0058] Step S5: Map the candidate lockdown center point to the corresponding grid cell in the urban road network model, and calculate the boundary set of the grid cell using the following formula:

[0059] .

[0060] Step S6: Calculate the distance from the candidate lockdown center point to each boundary road:

[0061] .

[0062] Take the road boundary with the smallest distance As a target of lockdown Let be the geometric distance function from the point to the road boundary.

[0063] Step S7: Execute a blockade when the following triggering constraints are met: the current number of blockades is less than the maximum number of schedulable blockades, the time interval between the last blockade and the last blockade is greater than the preset cooldown time, and the ground execution unit has idle resources; this can be formally expressed as: the ground execution unit has idle resources.

[0064] .

[0065] in, For the current moment, This refers to the time when the last lockdown was triggered. Cooling time, To ensure timely assembly on roads under lockdown, This is the maximum number that can be sealed off at the same time. The number of drones that proposed lockdown recommendations in this round. This is the minimum recommended number threshold.

[0066] Step S8: Send a lockdown command to the ground execution unit, the target road enters a lockdown state, and the lockdown set is updated:

[0067] .

[0068] in, This represents the set of roads under lockdown prior to the implementation of the lockdown. This represents the set of roads under lockdown after the lockdown is implemented. This indicates the target roads to be closed in this round.

[0069] In step S0, the heterogeneous swarm intelligence sensing information fusion mechanism is constructed, which specifically includes the following steps:

[0070] Step S0-1: Obtain the first... The self-perception information of the drone, wherein the self-perception information includes at least the first The location and motion information of targets within the sensing range of the UAV; wherein, the first The drone determines its location within the road network grid cell based on its current position and forms the corresponding local observation status.

[0071] Step S0-2: Obtain communication information of neighboring drones. When other drones are within a preset communication range, receive the location information, motion status information, target observation information, or blockade suggestion information of the other drones, and use the communication information of the neighboring drones as part of the current drone observation information.

[0072] Step S0-3: Obtain perception information from fixed monitoring equipment and crowd perception information. When fixed monitoring equipment or individuals in the crowd meet the preset perception access conditions, extract their corresponding target location, target status, or regional situation information, and introduce it as heterogeneous swarm intelligence aggregated perception information into the current observation construction process of the UAV to improve the completeness of situation perception in complex urban environments.

[0073] Step S0-4: Obtain road boundary or closure status information. This information is used to characterize the blocking, passable, or closed status of adjacent boundary roads within the grid cell where the UAV is located. The status of adjacent boundary roads can be represented as a binary state, with blocking or closure recorded as 1 and passable as 0.

[0074] Steps S0-5: The self-perception information, nearby drone communication information, fixed monitoring equipment perception information, crowd perception information, and road boundary or lockdown status information are uniformly integrated to form the first... The integrated observation vector of the UAV The integrated observation vector consists of at least heterogeneous swarm intelligence aggregated perception information, nearby UAV communication information, and road boundary or closure status information.

[0075] Steps S0-6: Based on the comprehensive observation vector, generate local observation inputs for subsequent lockdown decisions, so that each UAV can make a judgment on whether the current area needs to be locked down in step S1 and output the corresponding lockdown suggestion signal.

[0076] In steps S2 and S3, the consensus statistics of all the lockdown suggestion signals generated by the drones are statistically analyzed and a lockdown decision is triggered. This specifically includes the following steps:

[0077] Step S2-1: Obtain all lockdown suggestion signals generated by drones for the current area. Among them, the first The control recommendation signal output by the drone meets the requirements. .in, It is recommended that the corresponding roads in the current area be closed. They indicated that they do not recommend implementing a lockdown.

[0078] Step S2-2: Perform unified statistics on the blockade suggestion signals of all drones and calculate the consensus strength value corresponding to the group consensus. The consensus strength value is used to characterize the number of drones that are recommended for lockdown at the current moment; the consensus strength value is calculated by the following formula:

[0079] .

[0080] in, For indicator functions, The total number of drones involved in the decision-making process.

[0081] Step S2-3: Set the consensus strength value With preset consensus threshold When comparing, When the current area is determined to meet the group consistency constraint, the subsequent lockdown decision-making process begins; when If this occurs, the current lockdown process will be terminated, and road closures will not be triggered.

[0082] Step S2-4: Use the statistical results of the group consistency as the triggering basis for transitioning from local judgment of the UAV group to ground-based lockdown execution decision, so as to reduce the impact of local misjudgment of a single UAV on the overall lockdown decision and improve the stability of lockdown triggering in complex urban environments.

[0083] In step S4, the locations of the drones that proposed the containment suggestions are spatially weighted and aggregated to calculate the candidate containment center points. This specifically includes the following steps:

[0084] Step S4-1: Select drones from all drones that meet the lockdown recommendation signal. The set of drones is denoted as the candidate containment suggestion drone set. The spatial coordinates... The coordinates can be taken as the current position coordinates of the drone.

[0085] Step S4-2: Obtain the spatial coordinates of each drone in the candidate lockdown suggestion drone set when it made a lockdown suggestion. .

[0086] Step S4-3: Based on the spatial coordinates of the candidate containment suggestion drone set, aggregate the positions of all drones that have proposed containment suggestions to obtain the candidate containment center point. The candidate lockdown center point satisfies:

[0087] .

[0088] in, The number of drones used to propose lockdown measures.

[0089] Step S4-4: Use the candidate lockdown center point as the spatial reference position for subsequent road network mapping and target lockdown road selection, so as to characterize the comprehensive and concentrated area of ​​multiple UAV local perception results in space, thereby improving the rationality and stability of lockdown position determination.

[0090] In steps S5 and S6, mapping the candidate lockdown center point to the urban road network model and determining the target lockdown road specifically includes the following steps:

[0091] Step S5-1: The candidate containment center points obtained in step S4... Mapped to the corresponding grid cells in the urban road network model, which is used to represent the urban road topology and road boundary connection relationships.

[0092] Step S5-2: Obtain the set of boundary roads corresponding to the grid cell where the candidate control center point is located, which is used to characterize the connection relationship between the grid cell and the surrounding road boundaries.

[0093] Step S6-1: Calculate the candidate lockdown center point The geometric distance to each boundary road in the set of boundary roads, wherein the geometric distance is used to characterize the spatial proximity between the candidate lockdown center and the corresponding road boundary.

[0094] Step S6-2: Select the road boundary with the smallest geometric distance to the candidate lockdown center point from the set of boundary roads, and use it as the target lockdown road. .

[0095] Step S6-3: The target road to be blocked is used as the execution object for subsequent ground execution units to implement road blockade, so as to map the high-level blockade proposal of the UAV swarm into the constraint control at the specific road level.

[0096] In steps S7 and S8, road closure is implemented by the ground execution unit when the triggering constraint conditions are met, specifically including the following steps:

[0097] Step S7-1: Determine the set of roads under closure at the current time. Check if the number of roads currently under lockdown is less than the maximum number that can be locked down simultaneously; if the number of currently locked-down roads does not exceed the maximum number that can be locked down simultaneously, then the lockdown number constraint condition is met.

[0098] Step S7-2: Determine the current time. Compared to the last time the lockdown was triggered Does the time interval exceed the preset cooldown time? If it does... If this condition is met, the time-cooling constraint will be satisfied to avoid triggering road closures too frequently.

[0099] Step S7-3: Determine whether there are any idle ground execution units in the current system; if there are idle ground execution units, the execution resource constraints are met.

[0100] Step S7-4: If and only if the constraints on the number of road blockades, the time cooldown constraint, and the execution resource constraint are all satisfied at the same time, it is determined that the current target road blockade meets the trigger execution conditions and enters the ground execution scheduling process.

[0101] Step S8-1: Send a blockade command for the target blockade road to the idle ground execution unit, so that the target blockade road enters the blockade state.

[0102] Step S8-2: Update the set of blocked roads by adding the target blocked road to the set of blocked roads at the current time. The system records the moment when the lockdown was triggered, for use in the next round of lockdown constraint judgment.

[0103] Step S8-3: The ground execution unit physically blocks the target road, preventing the target from passing through the blocked road, thereby creating a road-level constraint on the target's activity space and realizing air-ground collaborative closed-loop control between the UAV swarm and the ground execution unit.

[0104] A heterogeneous crowdsourced consensus decision-making and air-ground coordinated scheduling device for urban road network control includes: a processor and an Internet terminal device with a computer-readable storage medium; the processor is used to call and execute multiple instructions stored in the computer-readable storage medium; the computer-readable storage medium is used to store multiple instructions; the instructions are used by the processor to load and execute a heterogeneous crowdsourced consensus decision-making and air-ground coordinated scheduling method for urban road network control.

[0105] A heterogeneous crowd-based consensus decision-making and air-ground collaborative scheduling system for urban road network control includes: a heterogeneous crowd-based sensing information fusion module, a drone swarm module, a consensus computing module, an aggregation computing module, a road network mapping module, a constraint control module, and an execution scheduling module. The heterogeneous crowd-based sensing information fusion module integrates drone sensing information, neighboring drone communication information, fixed monitoring equipment sensing information, crowd sensing information, and road boundary or control status information to form comprehensive observation information. The drone swarm module generates control suggestion signals based on the comprehensive observation information.

[0106] The consensus calculation module receives the lockdown suggestion signals from each UAV and calculates the consensus strength value to determine whether a preset lockdown threshold has been reached. The aggregation calculation module performs aggregation calculations on the locations of the UAVs that made lockdown suggestions when the lockdown triggering conditions are met to obtain candidate lockdown center points. The road network mapping module maps the candidate lockdown center points to the urban road network model and determines the target lockdown road. The constraint control module determines whether the lockdown quantity limit, cooldown time limit, and execution resource availability constraints are met. The execution scheduling module sends a lockdown command to the ground execution unit when the triggering constraint conditions are met, causing the target road to enter the lockdown state.

[0107] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0108] Example 1

[0109] In this embodiment, the system is deployed within the urban road network area.

[0110] See Figure 1 The scenario includes: a target, a drone, a monitoring point, and ground personnel. The target moves continuously along road boundaries or road sections in the urban road network; the drone dynamically tracks and monitors the target from the air; the monitoring point provides fixed perception support for local road areas; and the ground personnel are used to implement physical lockdown of the target road after receiving a lockdown order.

[0111] Through the coordinated efforts of aerial sensing, fixed monitoring, and ground execution, a dynamic spatial constraint control process is formed for complex urban road network scenarios. The system operates cyclically within discrete decision-making cycles. For each decision-making cycle, the system first collects the UAV's own perception data, neighboring UAV communication data, monitoring point perception data, crowd feedback data, and road boundary status data. Based on this, it generates the judgment results of each UAV on whether road closures should be implemented in the current area. Subsequently, the system performs consistency statistics on the judgment results of multiple UAVs and triggers the road closure decision process when a preset consensus threshold is reached. Further, the system determines the location of the candidate closure center and maps this location to a specific boundary road in the urban road network model. Finally, when the triggering constraint conditions are met, idle ground execution units are dispatched to the target road to execute the closure, thereby restricting the target's reachable activity range in the road network.

[0112] The number of drones in the system can be set to... The number of ground execution units can be set to [specified value]. Each drone perceives the target's status within its corresponding observation range and receives information from neighboring drones within its communication range. Fixed monitoring points continuously monitor road nodes or key areas, while individuals can provide auxiliary target information through mobile sensing or reporting. Through the joint participation of these multiple heterogeneous entities, the system can form a more complete situational awareness capability in complex urban environments.

[0113] The heterogeneous crowd-based consensus decision-making and air-ground coordinated scheduling for urban road network control in this embodiment specifically includes the following steps:

[0114] Step S0: Heterogeneous Swarm Intelligence Sensing Information Fusion

[0115] See Figure 2 Construct a heterogeneous swarm intelligence sensing information fusion mechanism, the first The comprehensive observation information of the drones consists of heterogeneous swarm intelligence aggregated perception information, neighboring drone communication information, and road boundary or closure status information. Specifically, the heterogeneous swarm intelligence aggregated perception information includes the drone's own position and motion status information of the target, target status information from fixed monitoring equipment, and auxiliary perception information from the crowd; the neighboring drone communication information is used to reflect the observation results, position status, or collaborative information of adjacent drones; the road boundary or closure status information is used to characterize the passability, congestion, or closure status of adjacent boundary roads of the current road network unit.

[0116] When a monitoring point or individual within a group is within the effective sensing range of the current drone, the target location or motion information it provides is written into the current drone's integrated observation vector. When a neighboring drone is within the current drone's communication range, the local observation results and position status of the neighboring drone are sent to the current drone and participate in observation fusion. In this way, a single drone no longer relies solely on its own local field of view to make judgments, but can integrate information from multiple sources to form a more complete environmental representation. This process helps reduce information gaps caused by single-source sensing and improves the accuracy of subsequent control and containment recommendation signals.

[0117] The road boundary or closure status information can be used to represent the status of adjacent boundary roads in the road network unit where the UAV is located. For example, when a road boundary is already closed, the corresponding state variable is recorded as closed; when a road boundary is not yet closed and is passable, the corresponding state variable is recorded as passable. By incorporating boundary status information into comprehensive observation information, the system can avoid repeatedly selecting already closed roads when selecting target closed roads in subsequent operations, thus improving the effectiveness of road decision-making.

[0118] II. Lockdown Recommendation Signal Generation, Group Consensus Statistics, and Lockdown Decision-Making

[0119] Step S1: Generation of Lockdown Recommendation Signal

[0120] Based on comprehensive observation information, each UAV generates a road closure recommendation signal to determine whether road closures are necessary in the current area. Specifically, the i-th UAV determines, based on its current comprehensive observation results, whether a target is about to pass through a critical road section or whether it may cross a boundary road of the current road network unit, and then outputs a road closure recommendation signal. , in, It is suggested that the relevant roads in the current area be closed. The statement indicated that a lockdown was not recommended.

[0121] Step 2: Group Consistency Statistics

[0122] See Figure 3 The system performs consensus statistics on all lockdown suggestion signals generated by drones. It receives all lockdown suggestion signals generated by drones and obtains the consensus strength value by counting the number of drones that suggest implementing lockdowns. The consensus strength value is used to characterize the number of drones that are considered to warrant road closures within the current decision-making cycle. The consensus strength value is calculated using the following formula:

[0123] .

[0124] Where I(⋅) is the indicator function, and N is the number of drones. The consensus strength value is used to characterize the number of drones that are considered to trigger road closures within the current decision-making cycle.

[0125] Step S3: Lockdown Decision

[0126] The consensus strength value is compared with a preset threshold. When the consensus strength value reaches or exceeds the preset threshold, it is determined that the current area has met the group consensus constraint, and the system enters the subsequent lockdown decision-making process; otherwise, the current lockdown process is terminated, and road closures are not triggered. Through this group consensus statistical mechanism, the problem of false triggering caused by individual drones due to limited local field of view, temporary obstruction, or local misjudgment can be effectively reduced, thereby improving the stability of the overall lockdown decision-making of the system.

[0127] The preset threshold can be set according to the number of drones, scene density, road complexity, or mission urgency. When the mission has high requirements for response timeliness, the threshold can be appropriately lowered to improve trigger sensitivity; when the scene noise is large or the cost of false alarms is high, the threshold can be appropriately increased to enhance system stability.

[0128] III. Aggregation Calculation of Candidate Lockdown Center Points

[0129] Step S4: Perform spatial aggregation calculations on the locations of the drones that proposed the lockdown recommendations.

[0130] See Figure 4 The system first filters out all that meet the requirements. The drones were monitored, and their spatial coordinates at the time the lockdown recommendations were made were obtained. Then, the candidate lockdown center point is calculated based on the spatial coordinates. This characterizes the spatial concentration trend of multiple drones' assessments of the controlled area.

[0131] The candidate containment center points are obtained by weighted aggregation of the locations of drones that proposed containment suggestions. The aggregation method can be either an average method or a weighted average method, and is expressed by the following formula:

[0132] .

[0133] in, The number of drones used to propose lockdown measures, For the first The spatial coordinates of the drone that proposed the lockdown.

[0134] By aggregating multiple spatial locations, the system can avoid directly selecting road closures based solely on a single drone location, thereby improving the stability and spatial rationality of candidate closure locations.

[0135] The candidate lockdown center point is not the final location for implementing lockdown measures, but rather an intermediate decision-making reference location. This location describes the spatial convergence area of ​​multiple UAV judgments and provides a geometric basis for subsequent road network mapping and target road selection.

[0136] IV. Candidate Blockade Location Mapping and Target Road Determination

[0137] Step S5: Candidate lockdown location mapping

[0138] The system maps candidate lockdown center points to corresponding grid cells in the urban road network model and obtains the set of boundary roads corresponding to each grid cell. The urban road network model can be used to represent road nodes, road edges, and the adjacency relationships between grid cells. By assigning candidate lockdown center points to their corresponding road network cells, the system can determine the set of candidate road boundaries most relevant to that point.

[0139] Step S6: Determine the target road

[0140] See Figure 5The system calculates the geometric distances between candidate lockdown centers and each boundary road, and selects the road boundary with the smallest distance as the target lockdown road. When a candidate lockdown center is located within a grid cell, the system calculates the distance relationship between it and each adjacent boundary road of that cell; combining the current lockdown and passability status of each boundary road, the target road for lockdown is finally determined. This process can be described as follows: the road boundary with the smallest distance is selected as the lockdown target, where the distance from the center point to the road boundary is determined by a geometric distance function, and the target lockdown road is expressed by the following formula:

[0141] .

[0142] in, Indicates the target road to be blocked; Show the set of boundary roads corresponding to the grid cell where the candidate lockdown center is located; Indicates the candidate lockdown center point Road to the border The geometric distance.

[0143] When the nearest boundary road is already under lockdown or does not meet the execution conditions, the system can continue to evaluate the remaining boundary roads and select the next best candidate boundary road as an alternative lockdown target. In this way, the system can further transform high-level lockdown suggestions at the drone swarm level into specific execution targets at the road level, improving the feasibility and implementability of the road selection process.

[0144] V. Trigger Constraint Judgment and Ground Execution Unit Scheduling

[0145] Step S7: Trigger constraint judgment

[0146] The system jointly assesses the limits on the number of roads under lockdown, the cooldown time limit, and the availability of execution resources. Before implementing a lockdown, the system must determine at least the following conditions: whether the number of roads currently under lockdown is less than the maximum number of schedulable lockdowns; whether the time interval between the current moment and the last lockdown trigger moment is greater than the preset cooldown time; and whether there are currently any idle ground execution units. Only when all of the above conditions are met will the system proceed with the lockdown execution process.

[0147] Specifically, the limit on the number of road closures is used to prevent the system from triggering too many road closure tasks at the same time, avoiding excessive occupation of execution resources due to an excessively large closure area; the cooldown time limit is used to prevent the system from repeatedly triggering closure actions in a short period of time, thereby reducing invalid closures and frequent switching; the execution resource availability judgment is used to ensure that there are ground execution units that can respond immediately, so as to avoid the system making idle decisions that cannot be implemented.

[0148] Step S8: Ground Execution Unit Scheduling

[0149] When the triggering conditions are met, the system sends a lockdown command to the ground execution unit, putting the target road into a lockdown state and updating the lockdown road set. Upon receiving the lockdown command, the ground execution unit moves to the target road location to implement physical lockdown, such as setting up roadblocks, assigning personnel to guard the road, or taking other road restriction measures. After the road lockdown is implemented, the target cannot pass through the road boundary, thereby spatially reducing its reachable area and effectively constraining the target's activity range. After completing the road lockdown, the system records the trigger time of this round of lockdown and updates the current lockdown road set for continued use in the next decision cycle. In this way, the system can continuously execute a closed-loop control process of "perception-judgment-lockdown-update" in a dynamic environment. Example Description

[0150] In one specific embodiment, this system can be deployed in an urban public safety management and control system, an emergency response system, or an intelligent transportation system. Within each decision-making cycle, the UAV judges the target's motion state and trend based on its own perception model. When the system predicts that the target is about to pass through a critical road section, each UAV judges whether road closures are necessary in the current area and generates a closure suggestion signal. Simultaneously, the system calculates the consensus strength value. When a preset threshold condition is met, the system enters the closure decision-making process, aggregating the positions of all UAVs that have proposed closures to obtain candidate closure center points. These candidate closure center points are then mapped to corresponding grid cells in the urban road network model, and the geometric distance between the grid cell and adjacent boundary roads is calculated to determine the target closure road. If the closure quantity limit, cooldown time limit, and execution resource availability constraint are met, the system schedules the nearest idle ground execution unit to the target road to implement road closure and updates the closure road set. After road closure is implemented, the target cannot pass through the target road, thus effectively constraining the target's activity space.

[0151] In practical applications, the method of this invention achieves dynamic road closure and spatial constraint control in complex urban road network environments through collaborative perception by UAV swarms and collaborative scheduling by ground execution units. It is applicable to scenarios such as urban public safety incident handling, emergency management and control, and road network closure management in key areas.

[0152] Example 2

[0153] This embodiment is a heterogeneous crowd-based consensus decision-making and air-ground coordinated scheduling device for urban road network control. The device includes a terminal device, which includes a processor and a computer-readable storage medium. The computer-readable storage medium stores multiple instructions, which are used by the processor to load and execute the above method steps.

[0154] Example 3

[0155] This embodiment describes a heterogeneous swarm intelligence consensus decision-making and air-ground collaborative scheduling system for urban road network lockdown. The system includes: a drone swarm module, a consensus calculation module, an aggregation calculation module, a road network mapping module, a constraint control module, and an execution scheduling module. Specifically, the drone swarm module generates lockdown suggestion signals based on local environmental perception information; the consensus calculation module receives lockdown suggestion signals from each drone and calculates the consensus strength value; the aggregation calculation module calculates candidate lockdown center points; the road network mapping module determines the target lockdown road; the constraint control module determines whether lockdown triggering constraints are met; and the execution scheduling module sends lockdown commands to the ground execution unit when the triggering constraints are met, causing the target road to enter a lockdown state.

[0156] Those skilled in the art will understand that various modifications, equivalent substitutions, and improvements can be made to the technical solutions of this invention without departing from the spirit and scope of protection of this invention, and all such modifications, equivalent substitutions, and improvements should fall within the scope of protection of this invention.

Claims

1. A heterogeneous swarm consensus decision and air-ground collaborative scheduling method, characterized in that, The method includes the following steps: Step S0: Construct a heterogeneous swarm intelligence sensing information fusion mechanism The system integrates drone perception information, neighboring drone communication information, fixed monitoring equipment perception information, crowd perception information, and road boundary or closure status information to form comprehensive drone observation information; Step S1: Generation of Lockdown Recommendation Signal Based on comprehensive observation information, each UAV generates a road closure suggestion signal to determine whether road closures are necessary in the current area. The road closure suggestion signal is used to characterize the judgment result of the corresponding UAV on implementing road closures in the current area. Step S2: Group Consistency Statistics Collective consensus statistics are performed on all the blockade suggestion signals generated by drones, and the consensus strength value is calculated to characterize the number of drones that suggest implementing road blockade at the current moment; Step S3: Lockdown Decision The road closure decision is as follows: when the consensus strength value is greater than or equal to a preset threshold, the road closure decision process is triggered. Otherwise, the current lockdown process will be terminated; Step S4: Aggregation Calculation of Candidate Lockdown Center Points Spatial aggregation calculations are performed on the locations of drones that propose lockdown suggestions to obtain candidate lockdown center points, which characterize the spatial concentration trend of multiple drones implementing road lockdowns in the target area; Step S5: Candidate lockdown location mapping The candidate lockdown center point is mapped to the corresponding grid cell in the urban road network model, and the set of boundary roads corresponding to the grid cell is obtained to establish the mapping relationship between the candidate lockdown location and the specific road boundary. Step S6: Determine the target road Calculate the geometric distance between the candidate lockdown center point and each boundary road in the set of boundary roads, and select the road boundary with the smallest distance as the target lockdown road to realize the conversion from the candidate lockdown area to the specific road lockdown object; Step S7: Trigger constraint judgment Determine whether the restrictions on the number of people subject to lockdown, the cooldown time limit, and the availability of execution resources are met; if the triggering conditions are met, proceed with the execution scheduling process; if the triggering conditions are not met, terminate the current round of lockdown execution. Step S8: Ground Execution Unit Scheduling Send a lockdown command to the ground execution unit to put the target road under lockdown, and update the set of roads under lockdown to achieve air-ground coordinated closed-loop control of the road lockdown task.

2. The heterogeneous swarm consensus decision-making and air-ground collaborative scheduling method according to claim 1, characterized in that, Step S0 specifically includes: Step S0-1: Obtain the UAV's own position and motion information relative to the target; Step S0-2: Obtain the position, motion status, or target observation information within the communication range of a nearby UAV; Step S0-3: Obtain target status information provided by fixed monitoring equipment and crowds; Step S0-4: Obtain the passability, congestion, or closure status of the adjacent boundary roads of the current road network unit; Step S0-5: Integrate the above information to form a comprehensive UAV observation vector.

3. The heterogeneous swarm consensus decision-making and air-ground collaborative scheduling method according to claim 1, characterized in that, Step S2 involves a unified statistical analysis of the number of drones recommended for lockdown to obtain a consensus strength value.

4. The heterogeneous swarm consensus decision and air-ground collaborative scheduling method of claim 1, wherein, Step S3 compares the consensus strength value with a preset threshold. When the consensus strength value reaches the preset threshold, the subsequent blockade decision process is initiated, thereby forming a blockade triggering mechanism based on the consensus of multiple drone groups.

5. The heterogeneous swarm consensus decision and air-ground collaborative scheduling method of claim 1, wherein, Step S4 involves spatial aggregation calculation of the locations of the drones that propose the lockdown suggestions to obtain candidate lockdown center points, which are then characterized as the comprehensive judgment center of multiple drones in space.

6. The heterogeneous swarm consensus decision and air-ground collaborative scheduling method of claim 1, wherein, Step S6 maps the candidate lockdown center point to the corresponding grid cell in the urban road network model, obtains the set of adjacent boundary roads, and determines the target lockdown road based on the geometric distance between the candidate lockdown center point and each boundary road.

7. The heterogeneous swarm consensus decision-making and air-ground collaborative scheduling method according to claim 1, characterized in that, Steps S7 and S8 determine whether the lockdown execution conditions are met by jointly judging the number of currently locked roads, the time interval between the current time and the last lockdown trigger time, and the idle state of the ground execution unit. When the trigger conditions are met, a lockdown command is sent to the ground execution unit to put the target locked road into a locked state and update the locked road set to form a collaborative execution closed loop.

8. The scheduling device constructed by the heterogeneous swarm consensus decision and air-ground collaborative scheduling method of claim 1, characterized in that, The scheduling device is a terminal device consisting of a processor and a computer-readable storage medium. The computer-readable storage medium stores multiple instructions, which are used by the processor to load and execute the above-mentioned heterogeneous crowd intelligence consensus decision-making and air-ground collaborative scheduling method for urban road network control.

9. The scheduling system constructed by the heterogeneous swarm consensus decision and air-ground collaborative scheduling method of claim 1, characterized in that, The system includes: a heterogeneous swarm intelligence sensing information fusion module, a drone swarm module, a consensus computing module, an aggregation computing module, a road network mapping module, a constraint control module, and an execution scheduling module. The heterogeneous swarm intelligence sensing information fusion module integrates drone sensing information, neighboring drone communication information, fixed monitoring equipment sensing information, crowd sensing information, and road boundary or lockdown status information to form comprehensive observation information. The drone swarm module generates lockdown suggestion signals based on the comprehensive observation information. The consensus computing module receives lockdown suggestion signals from each drone and calculates the consensus strength value. The aggregation computing module aggregates the locations of drones that proposed lockdown suggestions when lockdown triggering conditions are met, obtaining candidate lockdown center points. The road network mapping module maps candidate lockdown center points to an urban road network model and determines the target lockdown road. The constraint control module determines whether lockdown quantity limits, cooldown time limits, and execution resource availability constraints are met. The execution scheduling module sends lockdown commands to the ground execution unit when triggering constraint conditions are met, causing the target lockdown road to enter a lockdown state.

Images