Urban environment monitoring unmanned aerial vehicle control method and system

Abstract

The invention relates to the field of unmanned aerial vehicle control, in particular to an unmanned aerial vehicle control method and system for urban environment monitoring. The method comprises the following steps: acquiring large-scale activity event information, analyzing an activity subject and a radiation range taking the subject as a center according to the large-scale activity event information, and generating an activity radiation range information set; obtaining an unmanned aerial vehicle road surface information set, analyzing the gathering trend and trend contradiction of the monitoring target according to the unmanned aerial vehicle road surface information set and the activity radiation range information set, and determining a target gathering contradiction information set; and according to the target gathering contradiction information set, analyzing to obtain a multi-unmanned-aerial-vehicle cooperation guiding strategy, guiding a road surface target to a destination, and generating an unmanned-aerial-vehicle target control report. In the urban unmanned aerial vehicle control process, management and control are promoted to be changed from experience driving to data driving, and the large-scale activity road surface management and control efficiency and safety are improved.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) control, and in particular to a UAV control method and system for urban environmental monitoring. Background Technology

[0002] In the field of emergency command for large-scale urban events, real-time monitoring of pedestrian and vehicle traffic around venues and ensuring unobstructed emergency access are core aspects of safety work. These aspects directly relate to the efficiency of emergency response, public safety, and social order, and are key technological supports for building a smart city public safety system.

[0003] However, existing monitoring methods based on fixed cameras or single drone patrols lack a mechanism for multi-dimensional and systematic perception and real-time analysis of traffic congestion in the face of instantaneous, high-density, and tightly coupled traffic jams involving people and vehicles. This not only makes it difficult to quickly locate the causes and scope of congestion, but also fails to provide the command center with forward-looking, tiered traffic management decision support, severely restricting the effectiveness of emergency response. Summary of the Invention

[0004] This application provides a method and system for controlling unmanned aerial vehicles (UAVs) used for urban environmental monitoring, in order to solve the above-mentioned technical problems.

[0005] In a first aspect, this application provides a method for controlling an unmanned aerial vehicle (UAV) for urban environmental monitoring, the method comprising: Obtain information on large-scale events; based on the information on large-scale events, analyze the main body of the event and the radiation range centered on the main body, and generate an information set on the radiation range of the event; Acquire a set of UAV road surface information; based on the UAV road surface information set and the activity radiation range information set, analyze the aggregation trend and trend contradictions of the monitored targets, and determine the target aggregation contradiction information set. Based on the set of conflicting information about the target aggregation, a multi-UAV cooperative guidance strategy is obtained to guide the road target to its destination and generate a UAV target control report.

[0006] The above technical solution first acquires information about large-scale events, integrates organizer registration materials, government platform data, and social media promotional content, and delineates the event's coverage area using urban GIS data. Then, drone aerial photography supplements information on roads and facilities, generating an event coverage area information set. Next, multiple drones are deployed to monitor routes covering the entire coverage area, collecting real-time data on people, vehicles, and obstacles to form a road surface information set. Correlation analysis of these two types of information sets identifies conflicts such as facility throughput, path competition, and emergency lane blockage, determining the target aggregation conflict information set. Finally, based on conflict priority and drone performance, a multi-drone coordination and guidance strategy is developed, drones are scheduled to perform tasks, real-time monitoring and data collection are conducted, and a drone target control report is generated and output to drone operators. By accurately defining the event control area, blind spots in traditional manual surveys are avoided, making monitoring and guidance more targeted. Real-time identification of three core road surface conflicts allows for rapid mitigation through multi-drone collaborative guidance, reducing congestion and safety hazards, and ensuring the orderly conduct of events. The generated control report provides data support for subsequent event management, helping to optimize solutions and shifting management from experience-driven to data-driven approaches, improving the efficiency and safety of large-scale event road surface management, and reducing the impact on surrounding traffic and residents' lives.

[0007] Optionally, the step of analyzing the main body of the event and the radiation range centered on the main body based on the large-scale event information to generate an event radiation range information set includes: the large-scale event information includes the event main body type, the estimated scale level of the event, and the core geographic coordinates of the event; based on the core geographic coordinates, coupled with pre-collected urban road network topology data, performing radiation range analysis: taking the core geographic coordinates of the event as the center, prioritizing path radiation along the urban main road network, analyzing isochronous circles and equidistant circles based on road network accessibility, and generating the event radiation range information set by covering the main roads, key transportation hubs, and public places where people may gather.

[0008] Optionally, acquiring the UAV road surface information set includes: acquiring low-resolution image data using a wide-angle vision sensor mounted on the UAV, with a resolution not exceeding a preset privacy protection resolution threshold, so that individual facial features cannot be identified in the image and can only be used to extract the overall light and shadow contour for optical flow calculation; acquiring thermal radiation data using a thermal imaging sensor and generating a group heat map, wherein the group heat map uses different colors to identify different object types, including pedestrians, motor vehicles and non-motor vehicles; and performing real-time fusion processing on the acquired low-resolution image data and the group heat map to generate the UAV road surface information set that only contains group movement features and does not contain individual identity information.

[0009] Optionally, the step of analyzing the aggregation trend and trend contradictions of the monitored targets based on the UAV road surface information set and the activity radiation range information set, and determining the target aggregation contradiction information set, includes: based on the UAV road surface information set, analyzing the overall motion vector of the monitored targets in the corresponding sub-region of the activity radiation range information set to quantify the real-time aggregation trend; comparing the real-time aggregation trend with the preset aggregation trend predicted based on the activity event model, and identifying at least one of the following trend contradictions: identifying situations where pedestrian aggregation causes the instantaneous passage demand of key entrances and exits to exceed their design capacity, and marking it as a facility throughput contradiction; identifying situations where pedestrians, motor vehicles, and non-motor vehicles interfere with each other in a limited right-of-way space, and marking it as a path competition contradiction; identifying situations where the aggregation behavior of pedestrians or vehicles encroaches on or weakens the smoothness of the preset emergency support path, and marking it as an emergency passage blockage contradiction; and generating the target aggregation contradiction information set based on the facility throughput contradiction, the path competition contradiction, and the emergency passage blockage contradiction.

[0010] Optionally, the step of analyzing the target aggregation conflict information set to obtain a multi-UAV cooperative guidance strategy to guide road targets to their destination includes: based on the facility throughput conflict, analyzing the geographical boundaries, congestion intensity, and capacity of available diversion paths at congested entrances and exits, and generating a dynamic diversion guidance strategy for multi-UAV collaborative diversion guidance; based on the path competition conflict, analyzing the type, speed difference, and geometric characteristics of conflicting traffic flows, and generating a spatiotemporal rights allocation guidance strategy for allocating spatiotemporal rights for paths; based on the emergency passage blockage conflict, analyzing the length of the blocked section, the nature of the blockage, and the location of the emergency destination, and generating an emergency passage priority guidance strategy for rapid clearance and priority passage guidance; and synthesizing the multi-UAV cooperative guidance strategy based on the dynamic diversion guidance strategy, the spatiotemporal rights allocation guidance strategy, and the emergency passage priority guidance strategy.

[0011] Optionally, based on the facility throughput contradiction, the analysis of the geographical boundaries, congestion intensity, and capacity of surrounding available diversion paths of congested entrances and exits, and the generation of a dynamic diversion guidance strategy for multi-drone collaborative diversion guidance, includes: determining the geographical boundaries of multiple congested entrances and exits by analyzing the location data of congested nodes in the facility throughput contradiction, combined with a pre-stored electronic map of the venue, and calculating the real-time congestion intensity level of each entrance and exit based on the density and flow velocity of target aggregation in the drone road surface information; and analyzing the functional role of each entrance and exit in the overall road network based on the activity radiation range information set: entrances and exits directly connecting the core activity area with external main roads. The primary diversion exits are identified, and the pathways connecting to secondary roads or parking lots are identified as secondary diversion exits. By coupling the real-time congestion intensity level of each entrance / exit with its functional role in the road network, a diversion priority is dynamically assigned to each congested entrance / exit: nodes with high congestion intensity and that are primary diversion exits are assigned the highest diversion priority. Based on the diversion priority, high-priority congested nodes are preferentially matched with surrounding diversion paths that have sufficient traffic capacity and consistent destinations, and the dynamic diversion guidance strategy is generated accordingly: multiple drones are designated to fly to the corresponding entrance / exit, with the highest diversion priority node as the core, to perform a coordinated diversion guidance task.

[0012] Optionally, the step of analyzing the types, speed differences, and geometric characteristics of conflicting traffic flows based on the path competition contradictions, and generating a spatiotemporal rights allocation guidance strategy for allocating path spatiotemporal rights, includes: identifying conflicts between pedestrians, motor vehicles, and non-motorized vehicles at the junctions of main roads and entrances / exits around the venue by analyzing the UAV road surface information set; marking high-speed difference conflicts between motor vehicles and pedestrians / non-motorized vehicles as first-level conflicts, and low-speed difference conflicts between pedestrians and non-motorized vehicles as second-level conflicts; and analyzing the geometric characteristics of the junctions where conflicts occur, focusing on identifying whether a funnel shape is formed at the junction of the venue exit and the main road. The analysis examines whether the diffusion area and road channelization facilities form bottleneck-like contraction areas during the event. It then performs a conflict correlation analysis between the geometric features and conflict types: determining that the funnel-shaped diffusion area interrupts traffic flow on the main road, directly triggering a high-risk primary conflict; while the bottleneck-like contraction area compresses the coexistence space of different traffic flows, exacerbating the interweaving and congestion of primary and secondary conflicts. Based on the conflict correlation analysis results, a spatiotemporal rights allocation guidance strategy is generated: drones are assigned to key conflict points to allocate independent passage time windows and light-guided passage zones for different traffic flows, thereby achieving the separation of spatiotemporal rights.

[0013] Optionally, based on the emergency passage blockage conflict, the analysis of the length of the blocked passage segment, the nature of the blockage, and the location of the emergency destination, to generate an emergency passage priority guidance strategy for rapid clearing and priority passage guidance, includes: calling a pre-stored venue-specific emergency passage map, comparing the blocked segments identified in the emergency passage blockage conflict with the preset emergency support paths marked on the map, accurately defining the start and end points of the blocked segments, and obtaining the length of the blocked passage segment with the core activity area as a reference; fusing and analyzing the group heat map and the low-resolution image optical flow data, and determining the nature of the blockage based on the heat density distribution pattern and the consistency of contour movement: areas exhibiting high heat density and chaotic movement vectors are identified as densely populated areas. Blockage is identified by identifying areas with low heat density and contours moving along the road as vehicle stagnation. The absolute location of the emergency destination is determined by correlating the core geographic coordinates of the activity's radiation range information with pre-set emergency support route terminals. Combined with real-time positioning data transmitted by drones, the spatial orientation of this destination relative to the blocked section is analyzed. Based on this spatial orientation analysis, an emergency passage priority guidance strategy is generated: for short-distance blockages caused by dense crowds, rapid one-way traffic management is employed, with drones providing directional warnings and lateral guidance; for long-distance blockages caused by stationary vehicles, a tiered clearing system based on vehicle size and remote voice warnings are used, with multiple drones working together to address the source of congestion and ensure unobstructed access to the emergency passage.

[0014] Optionally, generating the UAV target control report includes: compiling the multi-UAV cooperative guidance strategy into a flight command set and guidance task sequence that can be parsed by the UAV swarm; scheduling designated UAVs to fly to corresponding spatial locations according to the flight command set and task sequence, and performing coordinated diversion, space-time priority allocation, and emergency channel priority guidance operations; collecting real-time status data of the UAV execution process and changes in road surface information after guidance, and generating the UAV target control report containing strategy execution effectiveness assessment and residual conflict identification. Through the above technical solution, the standardized flight command set clarifies the UAV's flight parameters, actions, and cooperative timing; the guidance task sequence clarifies the task stages and triggering conditions, making UAV execution based on evidence and multi-UAV cooperation orderly; the full-process data recorded in the UAV target control report provides reusable experience for the management of similar activities in the future and also provides direction for technical optimization.

[0015] Secondly, this application provides a UAV control system for urban environmental monitoring. The system includes: a range confirmation module, used to acquire information on large-scale events, analyze the main body of the event and the radiation range centered on the main body based on the event information, and generate an event radiation range information set; a conflict analysis module, used to acquire a UAV road surface information set, analyze the aggregation trend and trend conflicts of monitored targets based on the UAV road surface information set and the event radiation range information set, and determine a target aggregation conflict information set; and a control report module, used to analyze and obtain a multi-UAV cooperative guidance strategy based on the target aggregation conflict information set, guide road targets to their destination, and generate a UAV target control report.

[0016] Optionally, the scope confirmation module is specifically used for: the large-scale event information including the event subject type, the estimated scale level of the event, and the core geographic coordinates of the event; based on the core geographic coordinates, coupled with pre-collected urban road network topology data, performing a radiation range analysis: taking the core geographic coordinates of the event as the center, prioritizing path radiation along the urban main road network, analyzing isochronous circles and equidistant circles based on road network accessibility, and generating the event radiation range information set by covering the main roads, key transportation hubs, and public places where people may gather.

[0017] Optionally, the contradiction analysis module, when acquiring the UAV road surface information set, is specifically used for: acquiring low-resolution image data using a wide-angle vision sensor mounted on the UAV, with a resolution not exceeding a preset privacy protection resolution threshold, so that individual facial features cannot be identified in the image, and it can only be used to extract the overall light and shadow contour for optical flow calculation; acquiring thermal radiation data using a thermal imaging sensor and generating a group heat map, wherein the group heat map uses different colors to identify different object types, including pedestrians, motor vehicles and non-motor vehicles; and performing real-time fusion processing on the acquired low-resolution image data and the group heat map to generate the UAV road surface information set that only contains group movement characteristics and does not contain individual identity information.

[0018] Optionally, when the conflict analysis module analyzes the aggregation trend and trend conflicts of the monitored targets based on the UAV road surface information set and the activity radiation range information set to determine the target aggregation conflict information set, it is specifically used to: analyze the overall motion vector of the monitored targets in the corresponding sub-region of the activity radiation range information set based on the UAV road surface information set to quantify the real-time aggregation trend; compare the real-time aggregation trend with the preset aggregation trend predicted based on the activity event model, and identify at least one of the following trend conflicts: identify the situation where pedestrian aggregation causes the instantaneous passage demand of key entrances and exits and other nodes to exceed their design capacity, and mark it as a facility throughput conflict; identify the situation where pedestrians, motor vehicles and non-motor vehicles interfere with each other in the limited right-of-way space, and mark it as a path competition conflict; identify the situation where the aggregation behavior of pedestrians or vehicles encroaches on or weakens the smoothness of the preset emergency support path, and mark it as an emergency passage blockage conflict; and generate the target aggregation conflict information set based on the facility throughput conflict, the path competition conflict and the emergency passage blockage conflict.

[0019] Optionally, when the control reporting module analyzes and obtains a multi-UAV cooperative guidance strategy based on the target aggregation conflict information set to guide road targets to their destinations, it specifically performs the following: based on the facility throughput conflict, it analyzes the geographical boundaries, congestion intensity, and capacity of available diversion paths at congested entrances and exits to generate a dynamic diversion guidance strategy for multi-UAV collaborative diversion guidance; based on the path competition conflict, it analyzes the type, speed difference, and geometric characteristics of conflicting traffic flows to generate a spatiotemporal rights allocation guidance strategy for allocating spatiotemporal rights for paths; based on the emergency passage blockage conflict, it analyzes the length of the blocked passage segment, the nature of the blockage, and the location of the emergency destination to generate an emergency passage priority guidance strategy for rapid clearance and priority passage guidance; and based on the dynamic diversion guidance strategy, the spatiotemporal rights allocation guidance strategy, and the emergency passage priority guidance strategy, it synthesizes the multi-UAV cooperative guidance strategy.

[0020] Optionally, when the control reporting module analyzes the geographical boundaries, congestion intensity, and capacity of available diversion paths at congested entrances and exits based on the facility throughput imbalance, and generates a dynamic diversion guidance strategy for multi-drone collaborative diversion guidance, it specifically performs the following: by analyzing the location data of congested nodes in the facility throughput imbalance, combined with a pre-stored electronic map of the venue, it determines the geographical boundaries of multiple entrances and exits where congestion occurs, and calculates the real-time congestion intensity level of each entrance and exit based on the density and flow velocity of target aggregation in the drone road surface information; based on the activity radiation range information set, it analyzes the functional role of each entrance and exit in the overall road network: directly connecting the core activity area with the external main road. The entrances and exits of roads are identified as primary diversion exits, and pathways connecting to secondary roads or parking lots are identified as secondary diversion exits. By coupling the real-time congestion intensity level of each entrance and exit with its functional role in the road network, a diversion priority is dynamically assigned to each congested entrance and exit: nodes with high congestion intensity and that are primary diversion exits are assigned the highest diversion priority. Based on the diversion priority, surrounding diversion paths with sufficient traffic capacity and consistent destinations are preferentially matched to high-priority congested nodes, and the dynamic diversion guidance strategy is generated accordingly: multiple drones are designated to fly to the corresponding entrances and exits, with the highest diversion priority node as the core, to perform collaborative diversion guidance tasks.

[0021] Optionally, when the control reporting module analyzes the type, speed difference, and geometric characteristics of the conflicting traffic flows based on the path competition conflict to generate a spatiotemporal weight allocation guidance strategy for allocating path spatiotemporal weights, it specifically uses the following methods: By analyzing the UAV road surface information set, it identifies conflicts between pedestrians, motor vehicles, and non-motorized vehicles at the junctions of the main roads and entrances / exits around the venue; it marks high-speed difference conflicts between motor vehicles and pedestrians / non-motorized vehicles as Level 1 conflicts, and low-speed difference conflicts between pedestrians and non-motorized vehicles as Level 2 conflicts; for the junctions where conflicts occur, it analyzes their geometric characteristics; it focuses on identifying whether the junction between the venue exit and the main road... The system identifies whether a funnel-shaped diffusion area is formed and whether road channelization facilities form a bottleneck-like contraction area during the activity. It then performs a conflict correlation analysis between the geometric features and conflict types: determining that the funnel-shaped diffusion area interrupts traffic flow on the main road, directly triggering a high-risk primary conflict; while the bottleneck-like contraction area compresses the coexistence space of different traffic flows, exacerbating the interweaving and congestion of primary and secondary conflicts. Based on the conflict correlation analysis results, a spatiotemporal rights allocation guidance strategy is generated: drones are assigned to key conflict points to allocate independent passage time windows and light-guided passage zones for different traffic flows, thereby achieving the separation of spatiotemporal rights.

[0022] Optionally, when the control reporting module analyzes the length of the blocked passage, the nature of the obstruction, and the location of the emergency destination based on the emergency passage blockage conflict, and generates an emergency passage priority guidance strategy for rapid clearing and priority passage guidance, it specifically performs the following: It calls a pre-stored venue-specific emergency passage map, compares the blocked passage identified in the emergency passage blockage conflict with the preset emergency support path marked on the map, accurately defines the start and end points of the blocked passage, and obtains the length of the blocked passage with the activity core area as a reference; it fuses and analyzes the group heat map and the low-resolution image optical flow data, and determines the nature of the obstruction based on the heat density distribution pattern and the consistency of contour movement: areas exhibiting high heat density and chaotic movement vectors are identified. For dense crowd congestion, areas with low heat density and whose outlines move along the road direction are identified as vehicle standstill congestion. By associating the core geographic coordinates of the activity's radiation range with the preset emergency support route terminals, the absolute location of the emergency destination is determined. Combined with the positioning data transmitted back by drones in real time, the spatial orientation relationship of this destination relative to the obstructed section is analyzed. Based on the spatial orientation relationship analysis results, the emergency passage priority guidance strategy is generated: for short-distance congestion caused by dense crowds, rapid one-way diversion is adopted, with drones providing directional warnings and lateral guidance; for long-distance congestion caused by stationary vehicles, graded clearing based on vehicle size and remote early warning via voice broadcast are adopted, with multiple drones working together to handle the source of congestion and ensure the smooth flow of emergency passage entrances.

[0023] Optionally, when generating the UAV target control report, the control report module is specifically used for: compiling the multi-UAV cooperative guidance strategy into a flight instruction set and guidance task sequence that can be parsed by the UAV swarm; scheduling designated UAVs to fly to the corresponding spatial location according to the flight instruction set and task sequence, and performing coordinated diversion, space-time rights allocation, and emergency channel priority guidance operations; collecting real-time status data of the UAV execution process and changes in road surface information after guidance, and generating the UAV target control report that includes strategy execution effectiveness evaluation and residual conflict identification. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram illustrating an application scenario provided in one embodiment of this application; Figure 2A flowchart illustrating a method for controlling an unmanned aerial vehicle (UAV) for urban environmental monitoring, provided as an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an unmanned aerial vehicle (UAV) control system for urban environmental monitoring, provided as an embodiment of this application. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0027] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0028] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0029] Existing monitoring methods based on fixed cameras or single drone patrols lack a mechanism for multi-dimensional and systematic perception and real-time analysis of traffic congestion in the face of instantaneous, high-density, and tightly coupled traffic jams involving people and vehicles. This not only makes it difficult to quickly locate the causes and scope of congestion, but also fails to provide the command center with forward-looking, tiered traffic management decision support, severely restricting the effectiveness of emergency response.

[0030] Based on this, this application provides a method and system for controlling unmanned aerial vehicles (UAVs) for urban environmental monitoring. First, information on large-scale events is acquired, integrating organizer registration documents, government platform data, and social media promotional content. The event's coverage area is then delineated using urban GIS data. UAV aerial photography is then used to supplement information on roads, facilities, etc., generating an event coverage area information set. Next, multiple UAVs are deployed to monitor along routes covering the entire coverage area, collecting real-time data on people, vehicles, obstacles, etc., to form a road surface information set. The two types of information sets are correlated and analyzed to identify conflicts such as facility throughput, path competition, and emergency lane blockage, determining a target aggregation conflict information set. Finally, based on conflict priority and UAV performance, a multi-UAV coordination and guidance strategy is formulated, scheduling UAVs to perform tasks, real-time monitoring and data collection, and the integration of these data to generate a UAV target control report, which is then output to the UAV control personnel. By accurately defining the scope of event control, blind spots in traditional manual surveys are avoided, making monitoring and guidance more targeted; three types of core road surface problems are identified in real time, and rapid relief is achieved through multi-drone collaborative guidance, reducing congestion and safety hazards and ensuring the orderly conduct of events; the generated control reports provide data support for subsequent event control, helping to optimize plans, promoting the shift from experience-driven to data-driven control, improving the efficiency and safety of road control for large-scale events, and reducing the impact on surrounding traffic and residents' lives.

[0031] Figure 1 This diagram illustrates an application scenario provided by this application. In the process of drone control, the method provided in this application promotes a shift in management from experience-driven to data-driven approaches, improving the efficiency and safety of roadside management for large-scale events.

[0032] Specifically, the method of this application is applied to any server that communicates with the city event management system and the video aggregation and analysis platform. The server obtains large-scale event information provided by the city event management system and drone road surface information sets provided by the video aggregation and analysis platform. First, it acquires large-scale event information, integrates organizer registration materials, government platform data, and social media promotional content, and delineates the event's coverage area using city GIS data. Then, it supplements this with drone aerial photography to obtain information on roads, facilities, etc., generating an event coverage area information set. Next, it deploys multiple drones to monitor routes covering the entire coverage area, collecting real-time data on people, vehicles, obstacles, etc., to form a road surface information set. It then correlates and analyzes the two types of information sets to identify conflicts such as facility throughput, path competition, and emergency lane blockage, determining a target aggregation conflict information set. Finally, based on conflict priority and drone performance, it formulates a multi-drone coordination guidance strategy, schedules drones to perform tasks, monitors and collects data in real time, integrates and generates a drone target control report, and outputs it to the drone controller. Specific implementation methods can be found in the following embodiments.

[0033] Figure 2This is a flowchart illustrating a method for controlling an unmanned aerial vehicle (UAV) for urban environmental monitoring, provided as an embodiment of this application. The method of this embodiment can be applied to servers in the above-described scenarios. Figure 2 As shown, the method includes: S201. Obtain information on large-scale events. Based on the information on large-scale events, analyze the main body of the event and the radiation range centered on the main body, and generate an information set on the radiation range of the event.

[0034] Large-scale event information can refer to a set of information that can characterize the core attributes of a large-scale event, including the core geographical coordinates of the event (such as the city center sports center), the main type of the event (such as music festivals, sports events, exhibitions), and the estimated scale level of the event (such as an estimated 50,000 participants). The data comes from the city event management system.

[0035] The event radiation range information set can be a set of spatial and impact range data derived from the analysis of information on large-scale events. It is used to characterize the boundaries and attributes of the event subject (such as the main venue and core area) and its surrounding affected areas (such as traffic congestion areas, crowd gathering areas, and security control areas).

[0036] Specifically, in the environmental monitoring and control process of large-scale events, the event's radiation range information set is the "foundation" of the entire control work. Its necessity directly determines the effectiveness of all subsequent links. The core pain point solution lies in the comprehensive innovation of the traditional manual survey model. In the traditional preparation stage of large-scale events, determining the control range relies on manual on-site investigation, which requires a large amount of manpower to visit different areas. This is not only extremely time-consuming (e.g., a survey of a large music festival takes 2-3 days), but also limited by the personnel's field of vision and energy, making it easy to miss key areas—such as hidden locations like alley entrances and underground passage exits around the venue. These blind spots may become "dead zones" for crowds to gather during the event, causing congestion or even safety risks. At the same time, manual surveys can only record static information (such as road names) and cannot dynamically adjust the range according to the scale and type of the event. For example, the impact range of a sports event with an estimated 50,000 participants is completely different from that of an exhibition with 10,000 participants on the surrounding traffic and facilities. Manual methods are difficult to accurately match this difference. By acquiring information about large-scale events and analyzing the main body and radiation range of the event, an event radiation range information set is generated, providing accurate spatial positioning and data foundation for subsequent drone monitoring. By clearly defining control boundaries and core areas, drone monitoring is not blind and guidance is not lacking. This avoids the waste of control resources and safety hazards caused by unclear scope definition from the source, and lays a solid data foundation for the entire control process.

[0037] S202. Obtain the UAV road surface information set. Based on the UAV road surface information set and the activity radiation range information set, analyze the aggregation trend and trend contradictions of the monitored targets, and determine the target aggregation contradiction information set.

[0038] The UAV road surface information set can be a collection of road surface environment data collected in real time by sensors (such as high-definition cameras, infrared thermal imagers, and lidar) carried by UAV swarms. It includes low-resolution image data and group heat maps, and the data comes from a video aggregation and analysis platform.

[0039] The target aggregation conflict information set can be a set of abnormal aggregation behavior patterns of monitoring targets (pedestrians, motor vehicles and non-motor vehicles) obtained by analyzing the fusion of UAV road surface information set and activity radiation range information set, including facility throughput conflict, path competition conflict and emergency channel blockage conflict.

[0040] Specifically, during large-scale events, traditional manual monitoring employs a "fixed-point duty + patrol" model. Each staff member can only cover 1-2 small areas, making it impossible to grasp the dynamics of the entire coverage area in real time. It's even more difficult to detect the aggregation conflicts between different targets. Furthermore, traditional monitoring has three major blind spots: First, there's the conflict of facility capacity. During peak hours, important entrances and exits within the event's coverage area can experience a surge of people far exceeding the facility's capacity, which traditional manual monitoring struggles to detect in time, ultimately leading to widespread congestion around the facilities. Second, there's the conflict of path competition. Pedestrians, motor vehicles, and non-motorized vehicles often clash due to overlapping paths, such as pedestrians blocking narrow roads and electric vehicles needing to pass. Manual patrols struggle to capture these dynamic conflicts in real time, easily leading to road disorder. Finally, there's the conflict of blocked emergency lanes. Some sections of road are blocked by people or vehicles randomly stopping, and traditional monitoring easily overlooks these hidden risks. In the event of an emergency, this can severely hinder the passage of rescue forces. By acquiring drone-based road surface information and combining it with information from the event's coverage area, aggregation trends and trend conflicts can be analyzed to determine the target aggregation conflict information set. Establishing a crucial bridge between monitoring and guidance allows for timely and comprehensive understanding of conflict points, enabling precise matching of needs to subsequent multi-drone guidance strategies. This avoids escalation of conflicts due to information lag and provides a basis for efficient management, directly determining the timeliness and effectiveness of control and response.

[0041] S203. Based on the target aggregation conflict information set, analyze and obtain a multi-UAV cooperative guidance strategy to guide the ground target to the destination and generate a UAV target control report.

[0042] Multi-drone guidance strategies can be drone action plans generated by collaborative decision-making algorithms based on a set of conflicting information about the target aggregation. These strategies include dynamic diversion guidance strategies, temporal and spatial rights allocation guidance strategies, and emergency passage priority guidance strategies. The aim is to guide road targets (such as pedestrians or vehicles) safely and efficiently to designated destinations (such as evacuation points, parking areas, or event entrances).

[0043] The drone target control report can be a summary document of the guidance process, including an assessment of the effectiveness of strategy execution and identification of remaining contradictions, for subsequent optimization and manual review.

[0044] Specifically, after the traditional management of large-scale events, the summary work relies on the memory and simple notes of staff, resulting in fragmented and subjective content. For example, it may only record that congestion in a certain area has been alleviated, but cannot specify the exact duration of traffic control, the number of participants, or quantitative data on road improvements. When problems arise (such as poor traffic control in a certain area), it is difficult to trace the causes, let alone develop targeted optimization plans. This "no data, no review" model makes each event management operation feel like a "restart," preventing the accumulation of experience and hindering the improvement of management capabilities. Furthermore, it fails to provide objective evidence of work results to urban management departments, which is detrimental to subsequent resource allocation and policy formulation. Based on the target-aggregated conflict information set, this new approach analyzes the types and spatial distribution of conflicts inherent in the target-aggregated conflict information set, generates multi-drone coordination and guidance strategies, and ultimately outputs a drone target control report. The drone target control report summarizes each management operation and serves as the core basis for subsequent optimization. By analyzing the report, signals and enhancement equipment can be deployed in advance, continuously improving the level of management precision and shifting large-scale event management from "experience-driven" to "data-driven," ensuring continuous iteration and upgrading of management capabilities.

[0045] The method provided in this embodiment first obtains information about large-scale events, integrates the organizer's registration information, government platform data, and social media promotional content, and delineates the event's coverage area using urban GIS data. Then, it supplements information such as roads and facilities through drone aerial photography to generate an event coverage area information set. Next, it deploys multiple drones to monitor routes covering the entire coverage area, collecting real-time data on people, vehicles, obstacles, etc., to form a road surface information set. It then analyzes the correlation between the two types of information sets to identify conflicts such as facility throughput, path competition, and emergency passage blockage, and determines the target aggregation conflict information set. Finally, based on the conflict priority and drone performance, it formulates a multi-drone coordination guidance strategy, schedules drones to perform tasks, monitors and collects data in real time, integrates and generates a drone target control report, and outputs it to the drone control personnel. By accurately defining the scope of event control, blind spots in traditional manual surveys are avoided, making monitoring and guidance more targeted; three types of core road surface problems are identified in real time, and rapid relief is achieved through multi-drone collaborative guidance, reducing congestion and safety hazards and ensuring the orderly conduct of events; the generated control reports provide data support for subsequent event control, helping to optimize plans, promoting the shift from experience-driven to data-driven control, improving the efficiency and safety of road control for large-scale events, and reducing the impact on surrounding traffic and residents' lives.

[0046] In some embodiments, large-scale event information includes the event subject type, the estimated scale level of the event, and the core geographic coordinates of the event. Based on the core geographic coordinates, coupled with pre-collected urban road network topology data, a radiation range analysis is performed: taking the core geographic coordinates of the event as the center, the path is preferentially radiated along the urban main road network, and the isochronous circles and equidistant circles based on road network accessibility are analyzed to generate an event radiation range information set covering the main roads, key transportation hubs, and public places where people may gather.

[0047] The subject type of an event can be a classification and definition of the core carrier of a large-scale event, reflecting the essential attributes and characteristics of the event. For example, concerts, marathon events, international auto shows, and public welfare lectures all belong to different subject types.

[0048] The estimated scale level of an event can be a classification of the scale of a large-scale event based on factors such as the expected number of participants, venue capacity, and social impact.

[0049] The core geographic coordinates of an event can be the precise spatial location of the main venue for a large-scale event, presented in latitude and longitude coordinates (such as 30°15′N, 120°05′E) or urban plane rectangular coordinates, serving as the spatial reference point for radiation range analysis.

[0050] Radiation range analysis can be a process that starts from the core geographic coordinates of the event, combines urban road network topology data, estimated event scale and level, and analyzes the accessibility and flow patterns of people and vehicles in the road network to determine the spatial range in which large-scale events will affect the surrounding areas and require key monitoring and control.

[0051] An isochronous circle can be a closed spatial circle formed by taking the core geographic coordinates of an activity as the center and allowing people or vehicles to reach the same area in the same amount of time using the same mode of transportation. Examples include a "10-minute walking isochronous circle" and a "20-minute driving isochronous circle".

[0052] An equidistant circle can be a circular spatial layer formed with the core geographic coordinates of the activity as the center and a fixed distance as the radius, such as a "1-kilometer equidistant circle" or a "3-kilometer equidistant circle". It is the basis for defining the radiation range of the activity from the dimension of spatial distance and can intuitively reflect the spatial distance relationship between the core area of ​​the activity and the surrounding areas.

[0053] Specifically, in the traditional management of large-scale events, the coverage area is often subjectively determined based on the experience of staff, such as "extending 1-2 kilometers outward from the event venue." This does not take into account the actual characteristics of the event and the road network, resulting in the coverage area being either too large or too small. If the coverage area is too large, drone monitoring needs to cover unnecessary areas, resulting in a waste of drone resources (such as battery life consumption and computing power occupation), and may also miss real risk points due to the scattered monitoring areas. If the coverage area is too small, the impact of the event on the surrounding areas (such as crowd gathering and traffic congestion) exceeds the control range, forming a monitoring blind spot and causing safety hazards. For example, for a medium-sized concert, the traditional experience defines a coverage area of ​​1.5 kilometers. However, because the event venue is adjacent to the city's main road and subway station (a key transportation hub), the actual crowd gathering and traffic impact extend to a range of 2.5 kilometers. The traditional definition results in traffic congestion in the 2-2.5-kilometer area not being monitored, ultimately causing road paralysis. To address the above issues, this step first obtains information on large-scale events from the municipal database or event management platform, including the event's main type (e.g., sports events), estimated event scale level (e.g., large-scale - 50,000 people), and the core geographic coordinates of the event (e.g., the coordinates of a stadium (116.4°E, 39.9°N)). It also loads pre-collected urban road network topology data (e.g., including main road networks, transportation hub locations, and public space distribution maps). Then, using the core geographic coordinates of the event as the center, it prioritizes radiating paths along the urban main road network, using path planning algorithms to calculate the actual travel time from the core point to each area, generating isochronous circles. (e.g., 5-minute, 10-minute, 15-minute drive circles) and simultaneously calculate the actual distance of the road network to generate equidistant circles (e.g., 1 km, 3 km, 5 km road network distance circles). At the same time, identify key areas that overlap between isochronous circles and equidistant circles, including main roads (e.g., major urban expressways connecting stadiums), key transportation hubs (e.g., the nearest subway station, bus terminal) and public places where people may gather (e.g., surrounding squares, parking lot entrances). Finally, integrate the boundaries of isochronous circles, the range of equidistant circles, the list of main roads, hub coordinates, and public place polygons to generate a structured set of activity radiation range information (e.g., GIS layers or JSON format data).

[0054] The method provided in this embodiment combines urban road network topology data and road network accessibility analysis to determine the radiation range, avoiding blind spots and resource waste caused by traditional subjective delineation, and enabling the drone monitoring area to accurately match the actual impact range of the activity. Based on the isochronous circle and equidistant circle analysis of road network accessibility, it perfectly matches the actual travel paths of the participants in the activity. Key transportation hubs and public places within the radiation range are highlighted, and the drone can then specifically monitor the population gathering and traffic congestion in these areas.

[0055] In some embodiments, low-resolution image data is acquired using a wide-angle vision sensor mounted on the drone. The resolution of the image is no higher than a preset privacy protection resolution threshold, making it impossible to identify individual facial features in the image. The image can only be used to extract the overall light and shadow contour for optical flow calculation. Thermal radiation data is acquired using a thermal imaging sensor to generate a group heat map. The group heat map uses different colors to identify different object types, including pedestrians, motor vehicles, and non-motor vehicles. The acquired low-resolution image data and the group heat map are fused in real time to generate a drone road surface information set that contains only group movement characteristics and no individual identity information.

[0056] Low-resolution image data can be image data collected by a wide-angle vision sensor with a resolution not higher than a preset privacy protection resolution threshold. The image can only present the overall light and shadow outline of the road surface group (such as "areas of pedestrians" or "areas of vehicle clusters"), and cannot identify individual facial features, clothing details, vehicle license plates and other privacy information.

[0057] The privacy protection resolution threshold can be a critical value used to limit the resolution of images acquired by wide-angle vision sensors. It is a core parameter for balancing "monitoring clarity requirements" and "privacy protection requirements".

[0058] A group heat map can be a visualization image generated based on thermal radiation data, which uses color gradients to identify the types and densities of road targets. Different colors correspond to different target types (e.g., green represents pedestrians, red represents motor vehicles, and orange represents non-motor vehicles), and the shade of the same color represents the target density (the darker the color, the higher the density).

[0059] Specifically, in traditional large-scale event road monitoring, drones often use high-definition visual sensors to collect images. While these sensors can clearly capture road details, they also inevitably record private information such as individual facial features, vehicle license plates, and clothing details. For example, in the monitoring of a concert, the 1080P high-definition images collected by drones can clearly identify the faces of the audience. If this data is stored or transmitted directly without filtering, it may be illegally used for facial recognition and personal tracking, violating the relevant regulations that "the processing of personal information shall follow the principles of legality, legitimacy, and necessity, and shall not be excessive." This also raises public concerns about privacy and security, leading to compliance risks and public opinion pressure for event management. To address the above issues, this step first involves flying to the target area (e.g., the area around a stadium) and acquiring low-resolution image data (e.g., resolution not exceeding 640×480 pixels, frame rate 15fps) using a wide-angle vision sensor. Simultaneously, a thermal imaging sensor collects thermal radiation data and generates a group heat map (e.g., red indicates areas with temperatures above 35℃ corresponding to pedestrians, and blue indicates areas with temperatures below 20℃ corresponding to vehicles). Then, optical flow calculations are performed on consecutive low-resolution image frames to generate motion vector fields (e.g., arrow length indicates velocity, and direction indicates flow direction), and the heat map is used for object classification (e.g., setting temperature thresholds). 28℃ is set as the lower limit for pedestrians, and motor vehicles and non-motor vehicles are distinguished by the aspect ratio of the outline. Next, the optical flow vector is registered with the object position in the heat map to establish a motion-object association mapping, and fusion features are extracted (such as statistically analyzing the optical flow vector in the pedestrian area marked by the heat map, calculating the average movement speed of 1.5 m / s and the density of 0.4 people / m² in the area). Finally, a structured UAV road surface information set is generated, which includes timestamps, GPS coordinates, object type statistics, and motion parameters. The feature filtering algorithm is used to filter out any residual identity information (such as clothing texture) to ensure that the output contains only group motion features.

[0060] The method provided in this embodiment, based on low-resolution image acquisition, extracting only the overall light and shadow contours, and designing thermal maps that do not present individual details, completely eliminates personal privacy information concentrated in road surface information, fully complies with relevant legal compliance requirements, and avoids legal disputes and public opinion pressure caused by privacy leaks; the fusion design of "wide-angle vision + thermal imaging" is not affected by lighting conditions or target occlusion, can accurately distinguish between pedestrians, motor vehicles, and non-motor vehicles, and can simultaneously acquire group movement direction and density information, avoiding the type misjudgment problem of traditional single sensors.

[0061] In some embodiments, based on the UAV road surface information set, within the corresponding sub-region of the activity radiation range information set, the overall motion vector of the monitored target is analyzed to quantify the real-time aggregation trend; the real-time aggregation trend is compared with the preset aggregation trend predicted based on the activity event model, and at least one of the following trend contradictions is identified: the situation where pedestrian aggregation causes the instantaneous passage demand of key entrances and exits to exceed their design capacity is identified and marked as facility throughput contradiction; the situation where pedestrians, motor vehicles and non-motor vehicles interfere with each other in the limited right-of-way space is identified and marked as path competition contradiction; the situation where the aggregation behavior of pedestrians or vehicles encroaches on or weakens the smoothness of the preset emergency support path is identified and marked as emergency passage blockage contradiction; a target aggregation contradiction information set is generated based on the facility throughput contradiction, path competition contradiction and emergency passage blockage contradiction.

[0062] The overall motion vector can be the vector data of the overall motion direction and speed of the monitored targets (pedestrians, motor vehicles, and non-motor vehicles) by quantifying and analyzing the group light and shadow contours and group heat maps of the road surface information collected by UAVs within the corresponding sub-region of the activity radiation range. It includes two core dimensions: direction (e.g., 30° east of north) and speed (e.g., 5 m / s).

[0063] The preset aggregation trend can be the predicted data of the aggregation pattern of monitoring targets in different sub-regions and at different time nodes of the activity radiation range, output by the activity event model. It includes preset aggregation speed, preset aggregation density threshold, preset movement direction distribution, etc.

[0064] The contradiction between facility throughput and capacity can be caused by the gathering of pedestrians leading to an instantaneous passage demand at key entrances and exits within the activity's coverage area (such as the main entrance of the venue or the subway entrance near the venue) that exceeds the facility's design capacity (such as "the maximum passage capacity of the main entrance is 300 people per minute"), thus creating a trend contradiction.

[0065] Path competition conflicts can occur within the limited right-of-way space of the activity's coverage area (such as two-way two-lane roads or narrow pedestrian streets), where the movement directions or traffic demands of the three types of monitoring targets—pedestrians, motor vehicles, and non-motor vehicles—interfere with each other, leading to a trend of declining road traffic efficiency.

[0066] Emergency access blockage can be caused by the gathering of pedestrians or vehicles encroaching on pre-designated emergency access routes (such as fire lanes and ambulance lanes) within the activity's coverage area, leading to a trend where the accessibility of these routes is weakened or completely blocked.

[0067] Specifically, in traditional large-scale event road management, staff often rely on "visual observation + experience-based judgment" to identify gathering trends. For example, seeing a large number of people at the entrance is considered congestion. This lack of quantitative analysis of gathering trends easily leads to "misjudgment" or "missed judgment." Even if congestion is found on the road, it can only be described in general terms as "a large number of people and vehicles in a certain area," without accurately distinguishing the type of conflict, such as "insufficient facility capacity," "path competition interference," or "emergency lane blockage." This results in subsequent diversion strategies being ineffective. For example, when a concert ended, congestion occurred on the west side of the venue. Traditional judgment could only confirm the congestion but failed to identify that it was caused by the superposition of "pedestrians gathering towards the subway entrance (facility capacity conflict)" and "motor vehicles driving towards the parking lot (path competition conflict)." Diversion efforts only dispatched staff to guide pedestrians without controlling vehicles, and the congestion persisted for an hour without relief. In another example, during a plaza event, traditional management failed to detect the conflict of "emergency lanes being occupied by pedestrians," until a sudden medical incident occurred, and ambulances could not enter quickly, delaying rescue opportunities. To address the above issues, this step divides the coverage area into several sub-regions based on the activity radiation range information set (e.g., a 100-meter radius around each stadium entrance / exit). Within each sub-region, the overall motion vector from the UAV road surface information set is extracted, and the average speed of the monitored targets (e.g., pedestrian flow speed 0.8 m / s) and flow direction consistency (e.g., vector angle variance) are calculated to generate a real-time aggregation trend map (e.g., heat density distribution). Then, a preset aggregation trend model (e.g., a spatiotemporal distribution template based on historical activities) is invoked, and the real-time aggregation trend is compared point-by-point with the preset values ​​to identify facility throughput conflicts (e.g., detecting a real-time throughput at the south gate exceeding the design capacity of 200 people / minute), path competition conflicts (e.g., analyzing intersections where vehicle and pedestrian motion vectors intersect and the speed difference exceeds a threshold, such as vehicle speed 15 km / h and pedestrian speed 5 km / h), and emergency passage blockage conflicts (e.g., real-time aggregation hotspots covering an emergency passage exceeding 50 meters in length). Finally, all conflicts are integrated to generate a target aggregation conflict information set, where each record includes conflict type, geographic coordinates, and intensity level (e.g., facility throughput conflict - Level). 3) Add timestamps and prioritize the conflicts (e.g., emergency channel blockage conflicts take precedence over path competition conflicts), and output in structured JSON format.

[0068] The method provided in this embodiment, through the identification logic of "quantitative comparison + type subdivision", transforms traditional subjective experience judgment into objective data-driven analysis. It can accurately identify three types of core contradictions and avoid misjudging normal population flow as congestion or omitting the occupation of emergency channels. After contradiction identification, the system can continuously track the development dynamics of contradictions based on real-time aggregation trends, assess changes in risk levels, and provide a basis for adjusting guidance strategies.

[0069] In some embodiments, based on the contradiction between facility throughput and traffic volume, the geographical boundaries, congestion intensity, and capacity of available diversion paths at congested entrances and exits are analyzed to generate a dynamic diversion guidance strategy for multi-UAV collaborative diversion guidance. Based on the contradiction between path competition and traffic volume, the types, speed differences, and geometric characteristics of conflicting traffic flows are analyzed to generate a spatiotemporal weight allocation guidance strategy for allocating spatiotemporal weights for paths. Based on the contradiction between emergency passage blockage and traffic congestion, the length of the blocked passage segment, the nature of the blockage, and the location of the emergency destination are analyzed to generate an emergency passage priority guidance strategy for rapid clearance and priority passage guidance. Based on the dynamic diversion guidance strategy, the spatiotemporal weight allocation guidance strategy, and the emergency passage priority guidance strategy, a multi-UAV collaborative guidance strategy is synthesized.

[0070] Dynamic diversion and guidance strategies can be used to address the imbalance between facility throughput and to alleviate congestion at entrances and exits. This involves planning multiple drones to collaboratively guide people to available diversion routes in the surrounding area, such as "guiding people at the main entrance to divert to secondary entrances 2 and 3".

[0071] The strategy for guiding the allocation of time and space rights can be a strategy that addresses the conflict of path competition by allocating specific time or space rights of passage to different types of traffic flows (pedestrians, motor vehicles, and non-motor vehicles), such as "7:00-7:30, the east lane is given priority for pedestrians; 7:30-8:00, the east lane is given priority for motor vehicles" and "the 1.5-meter-wide area on the south side is designated as a dedicated lane for non-motor vehicles".

[0072] Emergency access priority guidance strategy can be a strategy to address the problem of emergency access blockage by dispatching drones to quickly guide pedestrians or vehicles gathered in the access lane to evacuate, and to ensure the priority passage of emergency vehicles (ambulances, fire trucks). It includes the order of clearing obstacles (such as "first guide pedestrians in the access lane, then guide temporarily parked vehicles"), the division of drone clearing areas, and the guidance of emergency vehicle passage routes.

[0073] Specifically, in traditional large-scale events, traffic management relies mainly on manual guidance (such as staff using loudspeakers), but this is limited by the small coverage area and low efficiency of instruction transmission, making it completely inadequate when dealing with large-scale gatherings and conflicts. Furthermore, traditional guidance methods lack specificity, employing a "general guidance" approach (such as "everyone go this way") regardless of whether the conflict involves "facility throughput," "path competition," or "emergency lane blockage," resulting in poor guidance effectiveness. For example, in a section of road where "pedestrians and electric vehicles compete for paths," the traditional approach only dispatches one staff member to guide pedestrians without managing electric vehicles, resulting in electric vehicles congesting the road as soon as pedestrians disperse. Similarly, when an emergency lane is blocked by pedestrians, the traditional approach only tells pedestrians to "hurry up" without specifying the evacuation direction, leading to chaos within the lane and delaying emergency rescue. To address the above issues, this step first generates a dynamic diversion guidance strategy based on the contradiction between facility throughput and capacity. It extracts the geographical boundaries of congested entrances and exits (e.g., the width of the stadium's east gate is 20 meters) and congestion intensity (e.g., real-time density exceeding 5 people / square meter), analyzes the capacity of surrounding diversion paths (e.g., the capacity of the western auxiliary road is 300 people / minute), and prioritizes paths according to congestion intensity and path capacity. High-capacity paths are matched to high-intensity congestion nodes, and multiple drones are instructed to guide diversion through light projection. Next, based on the contradiction of path competition, a spatiotemporal rights allocation guidance strategy is generated. It identifies conflicting traffic flow types (e.g., motor vehicles and pedestrians), speed differences (e.g., motor vehicles at 20 km / h, pedestrians at 5 km / h), and geometric characteristics of intersections (e.g., trumpet-shaped intersections), and classifies them according to conflict risk level. Equipped with independent passage time windows (e.g., 3-minute vehicle restriction) and light guidance strips, drones are assigned to hover and perform spatiotemporal separation. Next, based on the conflict of emergency passage blockage, an emergency passage priority guidance strategy is generated. The length of the obstructed section is measured (e.g., 80 meters), the nature of the obstruction is determined (e.g., high heat density and chaotic movement indicate crowd obstruction), and the location of the emergency destination (e.g., a medical point on the northwest side) is combined to generate a clearance priority. For crowd obstruction, directional sound wave warnings are used for guidance; for vehicle obstruction, multi-drone collaborative voice clearance is used. Finally, sub-strategies are integrated according to the priority order of "emergency passage priority > dynamic diversion > spatiotemporal rights allocation," and compiled into a command sequence (e.g., first dispatching drones A / B for obstruction clearance, then dispatching C / D for diversion, and finally E / F for spatiotemporal rights allocation), achieving seamless task connection.

[0074] The dynamic diversion and guidance strategy provided in this embodiment avoids overcrowding at a single exit and reduces the risk of stampedes; the time and space rights allocation guidance strategy clarifies the time and space rights of traffic flow and reduces the risk of collisions between pedestrians and vehicles; the emergency lane priority guidance strategy ensures that emergency lanes are unobstructed and avoids delays in emergency rescue; the generation and synthesis logic of the three types of sub-strategies can be quickly adapted to different activity types and conflict characteristics to form standardized strategy templates.

[0075] In some embodiments, by analyzing the location data of congested nodes in the facility throughput conflict and combining it with a pre-stored electronic map of the venue, the geographical boundaries of multiple entrances and exits where congestion occurs are determined, and the real-time congestion intensity level of each entrance and exit is calculated based on the density and flow velocity of target aggregation in the UAV road surface information. Based on the activity radiation range information set, the functional role of each entrance and exit in the overall road network is analyzed: entrances and exits directly connecting the activity core area and external main roads are identified as primary diversion exits, and passages connecting to secondary roads or parking lots are identified as secondary diversion exits. By coupling the real-time congestion intensity level of each entrance and exit with its functional role in the road network, diversion priority is dynamically assigned to each congested entrance and exit: nodes with high congestion intensity and that are primary diversion exits are assigned the highest diversion priority. Based on the diversion priority, high-priority congested nodes are preferentially matched with surrounding diversion paths with sufficient traffic capacity and consistent path endpoints, and a dynamic diversion guidance strategy is generated accordingly: multiple UAVs are designated to fly to the corresponding entrances and exits, with the highest diversion priority node as the core, to perform collaborative diversion guidance tasks.

[0076] Real-time congestion intensity level can be an indicator to quantify the degree of congestion at entrances and exits. Based on the comprehensive judgment of "target aggregation density" and "traffic flow speed" gathered from drone road information, it can be divided into three levels: mild, moderate and severe.

[0077] The highest diversion priority can be the highest level of diversion order assigned in a dynamic diversion guidance strategy after coupling analysis of the real-time congestion intensity level of congested entrances and exits and their functional roles in the road network. It is the core basis for prioritizing the matching of drone resources and diversion paths.

[0078] Collaborative diversion and guidance tasks can be diversion tasks based on the diversion needs of the highest priority nodes, executed by multiple drones in a manner of "regional division of labor + time sequence coordination + command coordination". The core objective is to quickly alleviate the capacity constraints of the highest priority nodes and guide people to orderly evacuate to suitable diversion paths.

[0079] Specifically, traditional diversion and guidance methods have four major shortcomings: First, the geographical boundaries of congested entrances and exits are not clearly defined, and the areas guided by drones are arbitrary, which easily leads to overlap, waste, or blind spots. For example, when the main entrance is congested, drones guide indiscriminately, which wastes resources and misses out on gathering crowds. Second, the functional roles of entrances and exits are misjudged, and main and secondary diversion exits are treated equally, resulting in insufficient resources for core channels. For example, if only one drone is deployed when the main exit is congested, it is ineffective in diversion. Third, there is no priority for diversion, and diversion is carried out in a random order, with high-risk nodes being dealt with late, which leads to escalation of congestion. Fourth, the diversion paths are not properly adapted, either the destination is not consistent or the capacity is insufficient, resulting in crowd resistance or congestion on new routes. To address the above issues, this step first extracts the GPS coordinate set of congested nodes from the facility throughput conflict data, and uses a pre-stored electronic map of the venue to generate an analysis grid of a certain size (e.g., 50m x 50m) centered on the coordinates. Based on the UAV road surface information set, the density per unit area (e.g., people / square meter) and average movement speed (e.g., meters / second) of targets within each grid are calculated. Then, the density and speed are substituted into the congestion intensity model to output the real-time level (e.g., level 1-4) of each entrance / exit. Subsequently, all paths connecting to the core activity area are marked in the activity radiation range information set. Graph theory algorithms are used to calculate the betweenness centrality of the paths, defining paths above a certain threshold (e.g., 0.7) as main road connections. Entrances / exits directly connecting to main roads are marked as primary diversion exits (e.g., marked in red), while the rest are marked as secondary diversion exits (e.g., marked in yellow). Next, a priority decision matrix is ​​established, where rows represent congestion levels (e.g., levels 1-4) and columns represent functional roles (e.g., primary / secondary). A priority score is calculated for each congested node, which is composed of the congestion level weight (e.g., 0.6) multiplied by the level value and the functional role weight (e.g., 0.4) multiplied by the role value. Then, priority queues are generated by sorting the nodes from high to low scores, and nodes with the same score are further sorted by considering historical congestion frequency (e.g., the number of times congestion occurred in the past hour). Finally, surrounding diversion paths are matched for the highest priority node, and paths with a capacity greater than the current congestion flow (e.g., 1000 people per hour) are selected. After verifying the consistency of the path endpoints, a drone instruction set containing target coordinates and broadcast content (e.g., "Please evacuate to the subway station to the southeast") is generated, and differentiated task packages are issued to multiple drones through the central control platform.

[0080] The method provided in this embodiment, based on the precise definition of the geographical boundaries of congested entrances and exits, clearly defines the responsibility area for each drone. This avoids the waste of resources caused by multiple drones repeatedly guiding in the same area, and also prevents the omission of crowds due to unclear boundaries, ensuring that the diversion guidance is "fully covered and without overlap". By identifying the functional roles of entrances and exits and determining the diversion priority, drone resources are preferentially allocated to the main diversion exits with high importance and high risk, avoiding the insufficient core channel resources caused by the traditional "average effort".

[0081] In some embodiments, by analyzing the UAV road surface information set, conflicts between pedestrians, motor vehicles, and non-motorized vehicles at the junctions of main roads and entrances / exits around the venue are identified: high-speed differences between motor vehicles and pedestrians / non-motorized vehicles are marked as Level 1 conflicts, and low-speed differences between pedestrians and non-motorized vehicles are marked as Level 2 conflicts; for the intersections where conflicts occur, their geometric characteristics are analyzed: the focus is on identifying whether a funnel-shaped diffusion area is formed at the junction of the venue exit and the main road, and whether the road channelization facilities form a bottleneck-like contraction area during the event; the geometric characteristics are correlated with the conflict type to determine that a funnel-shaped diffusion area will interrupt the traffic flow on the main road, directly triggering a high-risk Level 1 conflict; while a bottleneck-like contraction area will compress the coexistence space of different traffic flows, exacerbating the interweaving and congestion of Level 1 and Level 2 conflicts; based on the conflict correlation analysis results, a temporal and spatial rights allocation guidance strategy is generated: UAVs are assigned to key conflict points to allocate independent passage time windows and light-guided passage zones for different traffic flows to achieve the separation of temporal and spatial rights.

[0082] The trumpet-shaped diffusion area can be a space that unfolds in a fan shape at the junction of the venue exit and the main urban road. Because the road width suddenly expands from the narrow passage of the venue exit to the wide road surface of the main road, it can easily lead to the interruption of motor vehicle traffic and pedestrians crossing at will, which can then trigger a primary conflict.

[0083] Bottleneck-like contraction areas can be areas where the effective passage width of the road is significantly reduced during the event due to temporary facilities (such as security checkpoints and barriers) or road channelization facilities (such as isolation islands) occupying the road surface. This will compress the space for traffic flow to coexist and exacerbate the intertwining of primary and secondary conflicts.

[0084] A passage time window can be a specific passage period allocated exclusively to a certain type of traffic flow to ensure that different traffic flows are staggered in time and avoid conflicts caused by occupying the same area at the same time. The duration is set according to the intensity of the conflict and the traffic flow (such as usually 10-30 minutes).

[0085] Traffic guidance strips can be designated traffic zones projected onto the road surface using directional lighting equipment mounted on drones (e.g., red light strips represent motor vehicle lanes, and green light strips represent pedestrian lanes). These strips are used to clearly define the spatial range of traffic flow and prevent conflicts caused by crossing boundaries.

[0086] Specifically, traditional traffic management methods are significantly inadequate for managing large-scale events, struggling to address complex conflicts involving pedestrians, motor vehicles, and non-motorized vehicles. They suffer from three main flaws: First, they fail to accurately differentiate conflict types, often lumping high-risk (high-speed, accident-prone) conflicts between motor vehicles and pedestrians (Level 1) and low-risk (Level 2) conflicts between pedestrians and non-motorized vehicles together under the umbrella of "congestion." This leads to an underestimation of high-risk conflicts, misallocation of traffic management resources, and a high risk of accidents. Second, they neglect the impact of intersection geometry on conflicts, lacking specific analysis of key scenarios such as "trumpet-shaped diffusion areas" (which easily disrupt traffic flow and trigger Level 1 conflicts) and "bottleneck-shaped contraction areas" (which compress space and exacerbate conflict). They only provide superficial traffic management, failing to address the root causes of conflict and leading to recurring issues. Third, their "one-size-fits-all" approach wastes time and space resources, failing to allocate time and space based on conflict intensity and traffic flow characteristics. They either prioritize off-peak traffic during peak hours or leave wide roads idle, resulting in low traffic efficiency. To address the above issues, this step first analyzes the motion vectors of each traffic flow to calculate the speed difference between vehicular and pedestrian / non-motorized vehicle flows. Based on a preset threshold (e.g., 30 km / h), conflicts are marked as either high-risk Level 1 conflicts or low-risk Level 2 conflicts. Next, high-precision electronic map data is retrieved and registered with real-time aerial images transmitted from drones. Two key geometric features are identified: first, the naturally formed funnel-shaped diffusion area at the junction of the venue exit and the main road; and second, the bottleneck-shaped contraction area formed by temporary isolation facilities required for event security. Subsequently, conflict correlation analysis is performed, overlaying the identified conflict points with the geometric feature layer to determine that the funnel-shaped area primarily amplifies Level 1 conflicts. The risk of bottleneck areas is that they exacerbate the intertwining of primary and secondary conflicts. Based on this analysis, the system generates a temporal and spatial rights allocation guidance strategy: For trumpet-shaped areas, the strategy instructs drones to allocate a temporary pause time window (e.g., lasting 45 seconds) for the empty traffic flow upstream of the area, while simultaneously projecting a dedicated light guide strip for the dense pedestrian flow, achieving complete separation of people and vehicles in time and space; For bottleneck areas, the strategy instructs multiple drones to work collaboratively, alternately projecting multiple narrow light strips to allocate alternating passage micro-time windows for pedestrians, non-motorized vehicles, and motorized vehicles, thereby achieving orderly shuttle within a limited space; Finally, these strategies are encapsulated into an executable instruction set and issued to the drone swarm.

[0087] The method provided in this embodiment clearly distinguishes between first-level and second-level conflicts through speed difference calculation, prioritizes the allocation of resources to resolve high-risk first-level conflicts, and avoids the accident hazards caused by traditional vague traffic management. By analyzing the conflict correlation, the impact of geometric features on the conflict is clarified, and targeted traffic management plans are designed to eliminate the root causes of "traffic interruption" and "space compression" and avoid the recurrence of conflicts caused by traditional "treating the symptoms but not the root cause".

[0088] In some embodiments, a pre-stored emergency passage map for the venue is invoked, and the blocked sections identified in the emergency passage blockage conflict are compared with the preset emergency support paths marked on the map to accurately define the start and end points of the blocked sections, obtaining the length of the blocked passage section with the core activity area as a reference; the group heat map and low-resolution image optical flow data are fused and analyzed, and the nature of the blockage is determined based on the consistency of heat density distribution and contour movement: areas with high heat density and chaotic movement vectors are identified as dense crowd blockages, and areas with low heat density and contours moving along the road direction are identified as vehicle standby blockages; through correlation The activity's core geographic coordinates and pre-set emergency support route terminals, based on the activity's coverage area information, determine the absolute location of the emergency destination. Combined with real-time location data transmitted by drones, the spatial orientation relationship of this destination relative to the obstructed section is analyzed. Based on the spatial orientation analysis results, an emergency passage priority guidance strategy is generated: for short-distance congestion caused by dense crowds, rapid one-way diversion is adopted, with drones providing directional warnings and lateral guidance; for long-distance congestion caused by stationary vehicles, graded clearing based on vehicle size and remote early warning via voice broadcast are adopted, with multiple drones working together to address the source of congestion and ensure unobstructed access to emergency passages.

[0089] The heat density distribution pattern can be the heat intensity distribution characteristics of different areas in a group heat map, represented by color density. For example, dense crowds appear as "large areas of dark green blocks", while vehicles appear as "dispersed red / orange blocks", which can intuitively distinguish the types of obstructions.

[0090] Contour motion consistency can be the consistency of the motion pattern of the overall contour of an obstruction in a low-resolution image. In dense crowds, the individual movement directions are chaotic, and the overall contour has no unified movement trend (motion vectors are scattered); when a vehicle is stationary, the overall contour is completely still or moves only slightly (motion vectors approach 0).

[0091] Specifically, traditional emergency lane clearing methods have significant shortcomings in large-scale event scenarios, making it difficult to guarantee the timeliness of emergency rescue. These shortcomings include four major flaws: First, the definition of the obstruction area is vague. Relying solely on visual observation to determine the obstructed area often results in clearing areas that are too large or too small, either wasting resources clearing clear sections or overlooking parts of the obstructed area, thus hindering the passage of emergency vehicles. Second, the nature of the obstruction is easily misjudged, mistaking crowd obstruction for vehicle obstruction, or vice versa, leading to incorrect clearing methods, such as using tow trucks to guide crowds or relying solely on manpower to guide vehicles. This not only fails to solve the problem but may also exacerbate congestion. Third, guidance directions are confused. The spatial relationship between the emergency destination and the obstructed section is not clearly defined, leading crowds or vehicles to evacuate in the direction of emergency vehicles, causing path conflicts and delaying rescue. Fourth, clearing efficiency is low. There is a lack of prioritization and coordinated planning. Long-distance vehicle obstructions are cleared in a random order, and mixed crowd and vehicle obstructions are not cleared in a distinguishing order, making it difficult to quickly open up the passage. To address the above issues, this step automatically retrieves a pre-stored emergency passage map for the venue and spatially overlays it with the real-time monitored congestion areas to precisely define the start and end points of the obstructed sections, calculating the length of the obstructed passage section with the core activity area as a reference. Next, the system integrates and analyzes real-time crowd heat maps and low-resolution image optical flow data transmitted from drones: by analyzing the heat density distribution pattern, high heat density areas are identified and marked as potential crowd congestion points; by analyzing the consistency of contour movement, areas with chaotic movement vectors are identified as dense crowd congestion, while areas with consistent movement directions and low heat density are identified as vehicle congestion; simultaneously, the system correlates the core geographical coordinates of the activity's radiation range with the pre-set emergency support routes. The terminal determines the absolute location of the emergency destination and, combined with UAV GPS data, analyzes the spatial orientation of the destination relative to the obstructed section (e.g., whether the blockage point is upstream or downstream of the destination). Based on the above analysis results, the system generates an emergency passage priority guidance strategy: For short-distance dense crowd blockages, the strategy instructs UAVs to use a rapid one-way diversion method, guiding the crowd to a safe direction by playing directional warning sounds and projecting lateral light strips; for long-distance vehicle blockages, the strategy initiates a tiered clearance procedure based on vehicle size, with small vehicles receiving remote warnings via UAV voice broadcasts urging them to leave, and large vehicles being handled by multiple UAVs collaboratively marking their location and guiding clearance vehicles to the scene, ensuring that the emergency passage entrance is quickly restored to unobstructed access.

[0092] The method provided in this embodiment accurately defines the start, end, and length of obstructed sections by comparing pre-stored emergency passage maps with real-time images, ensuring that drones only guide the obstructed areas and do not involve unobstructed sections, thus avoiding the waste of resources caused by traditional large-scale blind obstacle clearing. Based on the fusion analysis of group heat maps and optical flow data, it accurately identifies "crowd obstruction" or "vehicle obstruction" and adopts targeted "rapid one-way diversion" or "tiered obstacle clearing" to avoid obstacle clearing failure caused by misjudgment of traditional nature.

[0093] In some embodiments, based on a multi-UAV cooperative guidance strategy, the strategy is compiled into a flight instruction set and guidance task sequence that can be parsed by the UAV swarm. According to the flight instruction set and task sequence, designated UAVs are scheduled to fly to the corresponding spatial location to perform cooperative diversion, time and space rights allocation and emergency channel priority guidance operations. Real-time collection of UAV execution process status data and road surface information set change data after guidance is used to generate a UAV target control report that includes strategy execution effectiveness assessment and residual contradiction identification.

[0094] A flight instruction set can be a set of machine instructions that can be parsed and executed by the hardware of a drone swarm, which is transformed from a multi-drone coordination and guidance strategy.

[0095] The guiding task sequence can be a list of drone tasks ordered chronologically and logically, clearly defining the specific tasks of each drone at different stages.

[0096] Specifically, after traditional large-scale event management, due to the lack of systematic records and evaluation data, the management can only rely on the memories of staff to summarize, making it impossible to accurately trace the causes of problems or provide reusable experience for subsequent events. This results in the management of each event starting from scratch. For example, a city holds a music festival every year, but because there is no data to compare with previous years' guidance, it is necessary to retest the number of drones and guidance methods each time, which is inefficient. Another example is that a certain area has repeatedly experienced conflicts over routes that have been ineffective in terms of guidance. Because there is no data record, it is impossible to analyze whether the problem is due to command issues, insufficient number of drones, or inappropriate guidance methods, and the problem remains unresolved. To address the above issues, this step first compiles the synthesized dynamic diversion guidance strategy, spatiotemporal weight allocation guidance strategy, and emergency channel priority guidance strategy into a structured data format that can be parsed by the UAV swarm, forming a flight instruction set and guidance task sequence. Subsequently, the central dispatch system dynamically assigns tasks to designated UAVs based on the instruction set and task sequence, controlling them to fly to corresponding spatial locations (such as congested entrances / exits, conflict intersections, or emergency channel blockages). During execution, the UAVs collect real-time data on their own status (such as battery level and flight stability) and changes in road surface information after guidance (such as reduced crowd density and increased traffic flow speed). Finally, the system integrates this data, evaluates the effectiveness of the strategy execution by comparing changes in contradictory indicators before and after guidance, identifies remaining contradictions (such as incompletely cleared nodes or newly emerging gathering points), and generates a structured UAV target control report.

[0097] The standardized flight instruction set provided in this embodiment clarifies the flight parameters, actions, and coordination sequence of the UAV, guides the task sequence to clarify the task stages and triggering conditions, and makes the execution of the UAV based on evidence and the coordination of multiple UAVs in an orderly manner. The full-process data recorded in the UAV target control report provides reusable experience for the management and control of similar activities in the future, and also provides direction for technology optimization.

[0098] Figure 3 This application provides a schematic diagram of the structure of an unmanned aerial vehicle (UAV) control system for urban environmental monitoring, as shown in one embodiment. Figure 3 As shown, a UAV control system 300 for urban environmental monitoring in this embodiment includes: a range confirmation module 301, a conflict analysis module 302, and a control report module 303; The scope confirmation module 301 is used to acquire information about large-scale events, analyze the main body of the event and the radiation range centered on the main body, and generate an event radiation range information set based on the event information. The conflict analysis module 302 is used to acquire a set of UAV road surface information, analyze the aggregation trend and trend conflicts of monitored targets based on the UAV road surface information set and the event radiation range information set, and determine a target aggregation conflict information set. The control report module 303 is used to analyze and obtain a multi-UAV cooperative guidance strategy based on the target aggregation conflict information set, guide road targets to their destination, and generate a UAV target control report.

[0099] Optionally, the range confirmation module 301 is specifically used for: the large-scale event information including the event subject type, the estimated scale level of the event, and the core geographic coordinates of the event; based on the core geographic coordinates, coupled with pre-collected urban road network topology data, performing radiation range analysis: taking the core geographic coordinates of the event as the center, prioritizing path radiation along the urban main road network, analyzing isochronous circles and equidistant circles based on road network accessibility, and generating the event radiation range information set by covering the main roads, key transportation hubs, and public places where people may gather.

[0100] Optionally, when acquiring the UAV road surface information set, the contradiction analysis module 302 is specifically used to: collect low-resolution image data using a wide-angle vision sensor mounted on the UAV, with a resolution not exceeding a preset privacy protection resolution threshold, so that individual facial features cannot be identified in the image, and it can only be used to extract the overall light and shadow contour for optical flow calculation; collect thermal radiation data using a thermal imaging sensor and generate a group heat map, wherein the group heat map uses different colors to identify different object types, including pedestrians, motor vehicles and non-motor vehicles; and perform real-time fusion processing on the collected low-resolution image data and the group heat map to generate the UAV road surface information set that only contains group movement features and does not contain personal identity information.

[0101] Optionally, when the contradiction analysis module 302 analyzes the aggregation trend and trend contradictions of the monitored targets based on the UAV road surface information set and the activity radiation range information set to determine the target aggregation contradiction information set, it is specifically used to: analyze the overall motion vector of the monitored targets in the corresponding sub-region of the activity radiation range information set based on the UAV road surface information set to quantify the real-time aggregation trend; compare the real-time aggregation trend with the preset aggregation trend predicted based on the activity event model, and identify at least one of the following trend contradictions: identify the situation where pedestrian aggregation causes the instantaneous passage demand of key entrances and exits and other nodes to exceed their design capacity, and mark it as a facility throughput contradiction; identify the situation where pedestrians, motor vehicles and non-motor vehicles interfere with each other in the limited right-of-way space, and mark it as a path competition contradiction; identify the situation where the aggregation behavior of pedestrians or vehicles encroaches on or weakens the smoothness of the preset emergency support path, and mark it as an emergency passage blockage contradiction; and generate the target aggregation contradiction information set based on the facility throughput contradiction, the path competition contradiction and the emergency passage blockage contradiction.

[0102] Optionally, when the control reporting module 303 analyzes and obtains a multi-UAV cooperative guidance strategy based on the target aggregation conflict information set to guide road targets to their destinations, it specifically performs the following: based on the facility throughput conflict, it analyzes the geographical boundaries, congestion intensity, and capacity of available diversion paths at congested entrances and exits, and generates a dynamic diversion guidance strategy for multi-UAV collaborative diversion guidance; based on the path competition conflict, it analyzes the type, speed difference, and geometric characteristics of conflicting traffic flows and intersection points, and generates a spatiotemporal rights allocation guidance strategy for allocating spatiotemporal rights for paths; based on the emergency passage blockage conflict, it analyzes the length of the blocked passage segment, the nature of the blockage, and the location of the emergency destination, and generates an emergency passage priority guidance strategy for rapid clearance and priority passage guidance; and synthesizes the multi-UAV cooperative guidance strategy based on the dynamic diversion guidance strategy, the spatiotemporal rights allocation guidance strategy, and the emergency passage priority guidance strategy.

[0103] Optionally, when the control reporting module 303 analyzes the geographical boundaries, congestion intensity, and capacity of available diversion paths at congested entrances and exits based on the facility throughput contradiction, and generates a dynamic diversion guidance strategy for multi-UAV collaborative diversion guidance, it specifically performs the following: by analyzing the location data of congested nodes in the facility throughput contradiction, combined with a pre-stored electronic map of the venue, it determines the geographical boundaries of multiple entrances and exits where congestion occurs, and calculates the real-time congestion intensity level of each entrance and exit based on the density and flow velocity of the target aggregation in the UAV road surface information; based on the activity radiation range information set, it analyzes the functional role of each entrance and exit in the overall road network: directly connecting the core activity area with the external main... The entrances and exits of main roads are identified as primary diversion exits, and the passages connecting to secondary roads or parking lots are identified as secondary diversion exits. By coupling the real-time congestion intensity level of each entrance and exit with its functional role in the road network, a diversion priority is dynamically assigned to each congested entrance and exit: nodes with high congestion intensity and that are primary diversion exits are assigned the highest diversion priority. Based on the diversion priority, surrounding diversion paths with sufficient traffic capacity and consistent destinations are preferentially matched to high-priority congested nodes, and the dynamic diversion guidance strategy is generated accordingly: multiple drones are designated to fly to the corresponding entrances and exits, with the highest diversion priority node as the core, to perform collaborative diversion guidance tasks.

[0104] Optionally, when the control reporting module 303 generates a spatiotemporal weight allocation guidance strategy for allocating spatiotemporal weights for path allocation based on the path competition conflict, it specifically performs the following: by analyzing the UAV road surface information set, it identifies conflicts between pedestrians, motor vehicles, and non-motor vehicles at the junction of the main roads and entrances / exits around the venue: marking high-speed difference conflicts between motor vehicles and pedestrians / non-motor vehicles as first-level conflicts, and low-speed difference conflicts between pedestrians and non-motor vehicles as second-level conflicts; for the junctions where conflicts occur, it analyzes their geometric characteristics: focusing on identifying the junctions between the venue exits and the main roads. Whether a funnel-shaped diffusion area is formed, and whether road channelization facilities form a bottleneck-shaped contraction area during the activity; perform conflict correlation analysis on the geometric features and conflict types: determine that the funnel-shaped diffusion area will interrupt the main road traffic flow and directly trigger the high-risk first-level conflict; while the bottleneck-shaped contraction area will compress the coexistence space of different traffic flows, exacerbating the intertwining and congestion of the first-level conflict and the second-level conflict; based on the conflict correlation analysis results, generate the spatiotemporal rights allocation guidance strategy: assign drones to key conflict points to allocate independent passage time windows and light-guided passage zones for different traffic flows to achieve the separation of spatiotemporal rights.

[0105] Optionally, when the control reporting module 303 generates an emergency passage priority guidance strategy for rapid clearance and priority passage guidance based on the emergency passage blockage conflict, it specifically performs the following: It calls a pre-stored venue-specific emergency passage map, compares the blocked sections identified in the emergency passage blockage conflict with the preset emergency support paths marked on the map, accurately defines the start and end points of the blocked sections, and obtains the length of the blocked sections with reference to the core activity area; it fuses and analyzes the group heat map and the low-resolution image optical flow data, and determines the nature of the blockage based on the heat density distribution pattern and contour movement consistency: areas exhibiting high heat density and chaotic movement vectors are identified as blockages. If a dense crowd is identified as causing congestion, areas with lower heat density and whose outlines move along the road direction are identified as areas of vehicle stagnation. By associating the core geographic coordinates of the activity with the preset emergency support route terminals within the activity's radiation range information set, the absolute location of the emergency destination is determined. Combined with the positioning data transmitted back by the UAV in real time, the spatial orientation relationship of this destination relative to the obstructed section is analyzed. Based on the spatial orientation relationship analysis results, the emergency passage priority guidance strategy is generated: for short-distance congestion caused by dense crowds, rapid one-way diversion is adopted, with UAVs providing directional warnings and lateral guidance; for long-distance congestion caused by stationary vehicles, graded clearing based on vehicle size and remote warning via voice broadcast are adopted, with multiple UAVs working together to handle the source of congestion and ensure the smooth flow of emergency passage entrances.

[0106] Optionally, when generating the UAV target control report, the control report module 303 is specifically used to: compile the multi-UAV cooperative guidance strategy into a flight instruction set and guidance task sequence that can be parsed by the UAV swarm; according to the flight instruction set and task sequence, schedule designated UAVs to fly to the corresponding spatial location and perform coordinated diversion, space-time rights allocation and emergency channel priority guidance operations; collect the status data of the UAV execution process and the change data of the road surface information set after guidance in real time, and generate the UAV target control report containing the strategy execution effectiveness assessment and residual contradiction identification.

[0107] The system in this embodiment can be used to execute the methods of any of the above embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

Claims

1. A method for controlling a drone for monitoring an urban environment, characterized in that, The method comprises the following steps: Obtain large-scale event information, analyze the activity subject and the radiation range centered on the subject according to the large-scale event information, and generate activity radiation range information set; Obtain unmanned aerial vehicle road surface information set, analyze the gathering trend and trend contradiction of the monitoring target according to the unmanned aerial vehicle road surface information set and the activity radiation range information set, and determine the target gathering contradiction information set; According to the target gathering contradiction information set, the multi-unmanned aerial vehicle cooperation guidance strategy is analyzed, the road surface target is guided to the destination, and the unmanned aerial vehicle target control report is generated.

2. The method of claim 1, wherein, According to the large-scale event information, the activity subject and the radiation range centered on the subject are analyzed, and the activity radiation range information set is generated, which comprises: The large-scale event information includes event subject type, event estimated scale level and event core geographic coordinates; Based on the core geographic coordinates, the pre-collected city road network topology data is coupled, and the radiation range analysis is performed: Taking the activity core geographic coordinates as the center, the path radiation is preferentially carried out along the city trunk road network, the isochronal circle and the equidistant circle based on the road network accessibility are analyzed, the covered trunk roads, key traffic hubs and public places where people may gather are generated, and the activity radiation range information set is generated.

3. The method of claim 2, wherein, The method comprises the following steps: Collect low-resolution image data through the wide-angle visual sensor carried by the unmanned aerial vehicle, the resolution of which is not higher than the preset privacy protection resolution threshold, so that the individual facial features in the image cannot be identified, and only the overall light and shadow profile can be extracted for optical flow calculation; Collect thermal radiation data through the thermal imaging sensor and generate a group thermal map, the group thermal map identifies different object types with different colors, the different object types including pedestrians, motor vehicles and non-motor vehicles; The collected low-resolution image data and the group thermal map are subjected to real-time fusion processing to generate the unmanned aerial vehicle road surface information set containing only group motion features without personal identity information.

4. The method of claim 3, wherein, According to the unmanned aerial vehicle road surface information set and the activity radiation range information set, the gathering trend and trend contradiction of the monitoring target are analyzed, and the target gathering contradiction information set is determined, which comprises: Based on the unmanned aerial vehicle road surface information set, the overall motion vector of the monitoring target is analyzed in the corresponding sub-region of the activity radiation range information set to quantify the real-time gathering trend; Compare the real-time gathering trend with the preset gathering trend predicted based on the activity event model, and identify at least one of the following trend contradictions: Identify the case that the instantaneous traffic demand of the key entrance and exit nodes caused by the gathering of pedestrians exceeds the designed capacity, and mark it as a facility throughput contradiction; Identify the case that pedestrians, motor vehicles and non-motor vehicles interfere with each other in the limited road right space, and mark it as a path competition contradiction; Identify the case that the gathering behavior of pedestrians or vehicles invades or weakens the smoothness of the preset emergency guarantee path, and mark it as an emergency passage obstruction contradiction; According to the facility throughput contradiction, the path competition contradiction and the emergency passage obstruction contradiction, the target gathering contradiction information set is generated.

5. The method of claim 4, wherein, According to the target gathering contradiction information set, the multi-unmanned aerial vehicle cooperation guidance strategy is analyzed, the road surface target is guided to the destination, which comprises: based on the facility throughput contradiction, analyze the geographical boundary of the congestion entrance and exit, the congestion intensity and the traffic capacity of the surrounding available shunting path, generate a dynamic shunting guidance strategy for multi-unmanned aerial vehicle cooperative shunting guidance; based on the path competition contradiction, analyze the type of conflict traffic flow, speed difference and geometric characteristics of intersection, generate a time-space right allocation guidance strategy for allocating path time-space right; based on the emergency channel blockage contradiction, analyze the length of the channel blocked paragraph, the nature of the blockage and the direction of the emergency destination, generate an emergency channel priority guidance strategy for rapid clearance and priority passage guidance; based on the dynamic shunting guidance strategy, the time-space right allocation guidance strategy and the emergency channel priority guidance strategy, the multi-unmanned aerial vehicle cooperation guidance strategy is synthesized.

6. The method of claim 5, wherein, The dynamic shunting guidance strategy for multi-unmanned aerial vehicle cooperative shunting guidance based on the facility throughput contradiction includes: By analyzing the position data of the congestion node in the facility throughput contradiction, the geographical boundary of the multiple entrances and exits where congestion occurs is determined in combination with the pre-stored venue electronic map, and the real-time congestion intensity level of each entrance and exit is calculated according to the density and flow rate of the unmanned aerial vehicle road surface information set target aggregation; Based on the activity radiation range information set, analyze the functional role of each entrance and exit in the overall road network: identify the entrances and exits directly connected to the activity core area and the external trunk road as the main shunting outlet, and identify the paths connected to the secondary road or parking lot as the secondary shunting outlet; By coupling analysis of the real-time congestion intensity level of each entrance and exit and its functional role in the road network, a shunting priority is dynamically allocated for each congestion entrance and exit: for nodes with high congestion intensity and main shunting outlet, the highest shunting priority is given; Based on the shunting priority, the high-priority congestion node is preferentially matched with a surrounding shunting path with sufficient traffic capacity and consistent path endpoint, and the dynamic shunting guidance strategy is generated accordingly: specify multiple unmanned aerial vehicles to fly to the corresponding entrance and exit, take the node with the highest shunting priority as the core, and perform cooperative shunting guidance task.

7. The method of claim 6, wherein, The time-space right allocation guidance strategy for allocating path time-space right based on the path competition contradiction includes: By analyzing the unmanned aerial vehicle road surface information set, identify the conflicts between pedestrians, motor vehicles and non-motor vehicles at the junction of the venue peripheral trunk road and the entrance: mark the high-speed difference conflict between motor vehicles and pedestrians / non-motor vehicles as a first-level conflict, and mark the low-speed difference conflict between pedestrians and non-motor vehicles as a second-level conflict; For the intersection where the conflict occurs, analyze its geometric characteristics: focus on identifying whether the horn-shaped diffusion area is formed at the junction of the venue exit and the trunk road, and whether the road channeling facilities form a bottleneck-shaped contraction area during the activity; Perform conflict association analysis on the geometric characteristics and conflict types: determine that the horn-shaped diffusion area will interrupt the trunk road traffic and directly cause high-risk first-level conflict; The bottleneck-shaped contraction region compresses the coexistence space of different traffic flows, intensifying the interweaving and congestion of the primary conflict and the secondary conflict; Based on the conflict correlation analysis result, the space-time right allocation guide strategy is generated: the UAV is assigned to allocate independent time window and light guide passage for different traffic flows at the key conflict point, so as to realize the separation of space-time right.

8. The method of claim 7, wherein, Based on the emergency passage blockage contradiction, the length of the blocked passage, the nature of the blockage and the direction of the emergency destination are analyzed to generate an emergency passage priority guide strategy for rapid clearance and priority passage guidance, including: The blocked passage is compared with the preset emergency guarantee path in the pre-stored venue special emergency passage map to accurately define the start and end points of the blocked passage, and the length of the blocked passage is obtained with the activity core area as the reference; The nature of the blockage is determined by fusing the crowd heat map and the low-resolution image optical flow data according to the consistency of heat density distribution form and contour movement: the area with high heat density and chaotic movement vector is determined as dense crowd blockage, and the area with low heat density and contour moving along the road direction is determined as vehicle stop blockage; The absolute direction of the emergency destination is determined by correlating the activity core geographic coordinates in the activity radiation range information set and the preset emergency guarantee path terminal, and the spatial pointing relationship of the destination relative to the blocked passage is analyzed combined with the real-time positioning data returned by the UAV; Based on the spatial pointing relationship analysis result, the emergency passage priority guide strategy is generated: for short-distance blockage of dense crowds, rapid one-way dredging is adopted, and directional warning and lateral guidance are performed by the UAV; For long-distance blockage of stopped vehicles, hierarchical clearance and voice broadcast remote warning are adopted according to the vehicle volume, and the source processing and emergency passage entrance unblocking guarantee are completed by multi-machine cooperation.

9. The method of claim 8, wherein, The UAV target control report is generated, including: Based on the multi-UAV cooperation guide strategy, it is compiled into a flight instruction set and a guide task sequence that can be parsed by a UAV cluster; According to the flight instruction set and the task sequence, the specified UAV is dispatched to fly to the corresponding spatial position to perform cooperative shunting, space-time right allocation and emergency passage priority guide operation; Real-time collection of state data and guided road information set change data during UAV execution process generates the UAV target control report containing strategy execution effectiveness evaluation and remaining contradiction identification.

10. A drone control system for monitoring urban environments, characterized in that, Applied to the method of any one of claims 1-9, comprising: A range confirmation module is configured to obtain large-scale event information, analyze the activity subject and the radiation range centered on the subject according to the large-scale event information, and generate an activity radiation range information set; A contradiction analysis module is configured to obtain a UAV road information set, analyze the aggregation trend and trend contradiction of the monitoring target according to the UAV road information set and the activity radiation range information set, and determine a target aggregation contradiction information set. A control report module is configured to analyze the target aggregation contradiction information set to obtain a multi-UAV cooperation guidance strategy, guide the road target to the destination, and generate a UAV target control report.

Images