Fire passage intelligent patrol method and system based on vehicle-mounted unmanned aerial vehicle

Abstract

The application discloses a kind of fire-fighting passage intelligent patrol method and system based on vehicle-mounted unmanned plane.The method comprises: based on task instruction, vehicle platform route and unmanned plane patrol route are planned to travel;Vehicle platform arrives after releasing unmanned plane to execute cruise task, and multi-source perception data are collected and real-time back transmission;Intelligent analysis platform fusion analysis data, identify occupation behavior and calculate comprehensive risk index from four dimensions of occupation object type, estimated occupation time, historical violation record and passage blockage degree to determine event level;Finally, according to event level, corresponding disposal process is executed.The application realizes rapid response and wide range patrol through vehicle-mounted unmanned plane cooperation;Through multi-dimensional fusion analysis and hierarchical disposal mechanism, intelligent identification, accurate evaluation and automatic closed-loop management to fire-fighting passage occupation are realized, and the patrol efficiency and law enforcement standardization are significantly improved.

Description

Technical Field

[0001] This invention relates to the fields of drones and fire protection technology, specifically to a method and system for intelligent inspection of fire lanes based on vehicle-mounted drones. Background Technology

[0002] Traditional manual patrols for addressing issues such as illegally parked vehicles and piled-up debris obstructing fire lanes suffer from low efficiency, numerous blind spots, difficulty in obtaining evidence, and slow response.

[0003] In recent years, drones have been introduced into the field of fire inspection due to their flexibility and wide field of view. Currently, there are existing technologies utilizing drones for fire inspection, such as the utility model patent "A Real-time Fire Inspection System for Drones" (patent number CN 213008793U). This patent discloses a real-time fire inspection system for drones, including a ground inspection control center and several inspection drone terminals. Each inspection drone terminal includes a mobile remote control terminal and a drone aircraft. The drone aircraft integrates a flight control device, an image acquisition device, an image transmission device, and a storage device. The image transmission device includes a drone-end control module, an encoder, a COFDM digital modulation module, a 5G module, and a transmitting antenna. The drone-end control module is connected to the storage device, and the encoder, COFDM digital modulation module, 5G module, and transmitting antenna are connected sequentially. The encoder, COFDM digital modulation module, and 5G module are each connected to the drone-end control module. This system can utilize the image transmission function of drones for ground inspection, allowing the control center to promptly observe abnormal conditions within the monitored area, thereby ensuring fire safety.

[0004] However, this system only utilizes the image transmission function of drones and cannot automatically identify or judge information. This significantly increases the workload of control center staff, requiring substantial manpower and resulting in low identification efficiency. Furthermore, the solution employs fixed-location drone deployment or purely manual operation, which suffers from inflexible deployment, limited coverage radius, low automation, and inability to effectively coordinate with ground forces. Consequently, it fails to meet the needs for routine, efficient, and intelligent monitoring of fire lanes. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the aforementioned background technology and provide a method and system for intelligent inspection of fire lanes based on vehicle-mounted drones.

[0006] The technical solution of this invention is: a method for intelligent inspection of fire lanes based on vehicle-mounted drones, comprising: Based on mission instructions, plan the travel route of the mobile vehicle platform and the drone patrol route within the target area; After the mobile vehicle platform arrives at the target area, it releases a drone equipped with a multi-functional pod, which then performs a patrol mission according to the planned patrol route. The drone collects multi-source sensing data of the target fire lane and transmits it back to the intelligent analysis platform in real time; The intelligent analysis platform integrates and analyzes the multi-source sensing data to identify whether there is any occupation behavior. If there is occupation behavior, it calculates a comprehensive risk index based on four dimensions: type of object occupied, estimated duration of occupation, historical violation records, and degree of channel blockage, and then determines the event level. Develop corresponding handling procedures based on the event level.

[0007] According to the present invention, a method for intelligent patrol of fire lanes based on vehicle-mounted drones is provided. The method for the drone to perform patrol missions according to a planned patrol route includes: the mobile vehicle-mounted platform moves to the next deployment point during the drone's mission; when the drone's battery level is lower than a threshold or after completing a local area patrol, the drone returns to the vehicle-mounted platform at the next deployment point for charging or battery replacement, and then takes off again to continue performing subsequent segments of the mission.

[0008] According to the present invention, a method for intelligent inspection of fire lanes based on vehicle-mounted drones is provided. The types of occupants are classified into temporary soft obstacles, non-motorized vehicles / motorcycles, and small private cars or large vehicles / fixed facilities using a visual recognition model, and each is assigned a different first risk score. α .

[0009] According to the present invention, a method for intelligent inspection of fire lanes based on vehicle-mounted drones is provided. The method for determining the estimated duration of road occupancy includes: combining thermal imaging to determine whether the vehicle is turned off, whether the driver is in the vehicle, and the status of the lights to identify momentary stopping behavior; if the object occupying the road remains stationary for more than a preset time threshold, the duration score is increased; and the score is corrected by combining the historical parking duration distribution of the license plate or geographical location to obtain the estimated duration of road occupancy, and a corresponding second risk score β is obtained based on the estimated duration of road occupancy.

[0010] According to the present invention, a method for intelligent inspection of fire lanes based on vehicle-mounted drones is provided. The historical violation records are statistically analyzed based on the number of times the same license plate number or the same geographical coordinate point is blocked within a set time period, and are divided into three levels: first violation, occasional violation, and high-frequency repeated violation, which correspond to the third risk score γ respectively.

[0011] According to the present invention, a method for intelligent inspection of fire lanes based on vehicle-mounted drones is provided. The degree of lane obstruction is obtained by acquiring three-dimensional point cloud data through lidar, calculating the ratio of the remaining effective width of the fire lane to the design width, obtaining the obstruction rate, and classifying it into three levels: slight obstruction, moderate obstruction, or severe obstruction, which correspond to the fourth risk score δ respectively.

[0012] According to the present invention, a method for intelligent inspection of fire lanes based on vehicle-mounted drones is provided, wherein the method for calculating the comprehensive risk index includes: Where: R – Comprehensive Risk Index; α —First risk score; β —Second risk score; c —Third risk score; d —Fourth risk score.

[0013] According to the present invention, a method for intelligent inspection of fire lanes based on vehicle-mounted drones is provided. The method for determining the event level includes: judging the event level from light to severe based on the comprehensive risk index as Level 1 event, Level 2 event and Level 3 event. However, if the degree of blockage is severe blockage or the historical violation record is high-frequency repeated violation, then regardless of the size of the comprehensive risk index, it is directly judged as a Level 3 event.

[0014] According to the present invention, a method for intelligent inspection of fire lanes based on vehicle-mounted drones is provided. The method for formulating corresponding handling procedures according to the event level includes: triggering a drone to remotely issue a warning and start a countdown monitoring; if the warning is not lifted, the event is automatically upgraded to a level two event. Level 2 incidents automatically generate electronic work orders and push them to the terminal of the local management personnel, requiring them to upload post-handling images for verification; Level 3 incidents are automatically packaged into complete electronic evidence and pushed to the administrative law enforcement platform through cross-departmental interfaces.

[0015] This invention also relates to an intelligent fire lane patrol and early warning system based on a vehicle-mounted drone that implements the above method, comprising: The mobile vehicle-mounted platform integrates an automated hangar, communication relay equipment, edge computing units, and a power supply system. The unmanned aerial vehicle (UAV) subsystem is equipped with a positioning module, an autonomous obstacle avoidance system, and a multi-functional pod that integrates a high-definition zoom camera, a thermal imaging camera, and a lidar. The collaborative control network, including public network communication links and private network communication links, is used to transmit different types of data respectively to ensure high bandwidth and low latency communication requirements. The intelligent analysis platform includes an analysis engine, a task scheduling engine, an early warning and workflow engine, and a visual data management interface, used to perform multi-dimensional risk assessments and automated closed-loop handling.

[0016] The advantages of this invention are: 1. Rapid response mobile collaborative patrol capability. Integrating unmanned aerial vehicle (UAV) systems into mobile vehicle platforms frees patrol forces from being confined to fixed locations. The system can move to any area that needs patrolling according to mission instructions, realizing a shift from passively waiting for alarms to actively conducting close-range reconnaissance. The drone does not need to return to its origin, but instead seeks out a vehicle-mounted platform to recharge or swap batteries at the next deployment point. This dynamic collaboration not only solves the inherent pain point of short drone battery life, but also enables uninterrupted scanning of long-distance and wide-area fire lanes, ensuring the efficiency and continuity of patrol missions, and truly achieving rapid arrival and continuous patrol. 2. Multi-dimensional integrated intelligent risk assessment model The system integrates multi-dimensional data such as visual recognition, thermal imaging analysis, historical data statistics, and lidar point clouds; this fusion analysis can accurately distinguish between temporary parking and malicious illegal parking, and between cardboard box stacks and vehicle congestion, providing comprehensive data support for decision-making; Quantifiable risk calculation models transform complex law enforcement discretion into transparent mathematical logic; by adjusting the weighting factors of each dimension, managers can flexibly configure control strategies. A veto mechanism was introduced, which ensured the comprehensiveness of daily assessments, safeguarded the bottom line of fire safety, and ensured that the system could automatically take the highest level of response in extreme situations. 3. A differentiated and flexible precision handling mechanism For low-risk incidents, the system first triggers a drone to remotely issue a warning and starts a countdown monitoring; this reflects the concept of flexible law enforcement, giving violators the opportunity to correct their mistakes, achieving the purpose of education, reducing confrontation, and increasing social acceptance. When an incident is not resolved in a timely manner or the risk is moderate, the system automatically generates an electronic work order and pushes it to the specific person in charge, requiring them to upload image data after rectification; this realizes online closed-loop management of discovery, handling, and verification, ensuring that the problem is truly resolved; For serious violations or repeated offenses that cross the red line, the system automatically packages complete electronic evidence and pushes it directly to the administrative law enforcement platform through the interface, enabling rapid cross-departmental collaboration and strong enforcement. 4. A fully closed-loop automated collaborative governance system The handling of Level 3 incidents is clarified, including pushing the data to the administrative law enforcement platform through cross-departmental interfaces. This feature connects the technical system with the administrative management system, enabling the automatic transfer of evidence of violations, greatly reducing manual intervention and communication costs, and improving the efficiency of cross-departmental collaborative law enforcement. From task planning, data collection, intelligent analysis, risk assessment, task assignment to result verification, the entire process is completed online by the intelligent analysis platform (including task scheduling engine and workflow engine), forming a complete business closed loop and ensuring the traceability and supervision of the governance process. 5. A complete and standardized automated electronic evidence chain The system collects not only single images or videos, but also thermal imaging data (used to prove the vehicle's off / on state) combined with visible light images, which strongly proves the nature of the occupancy time; LiDAR point cloud data provides objective quantitative evidence of channel blockage rate; these multi-source data corroborate each other, constructing a solid and tamper-proof chain of evidence. The entire process is automatically recorded and packaged by the system, avoiding omissions or disputes that may occur during manual evidence collection. When an incident escalates to a Level 3 incident, the electronic evidence package pushed out is clearly structured and complete in content (including images of the illegal parking process, thermal status, point cloud data, historical violation records, etc.), ensuring that the evidence transferred to law enforcement agencies has a high degree of standardization and legal validity, effectively reducing subsequent legal disputes. Attached Figure Description

[0017] Figure 1 Flowchart of the intelligent inspection method for fire lanes based on vehicle-mounted drones of the present invention. Detailed Implementation

[0018] Embodiments of the present invention are described in detail below, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0019] In the description of this invention, it should be understood that the terms "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0020] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0021] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0022] This invention relates to a method and system for intelligent patrol of fire lanes based on vehicle-mounted drones, creating a vehicle-drone relay patrol mode that greatly expands the system's spatiotemporal coverage; it constructs a composite risk assessment model that integrates objective physical data and subjective behavioral characteristics, taking into account both the scientific nature of the assessment and the flexibility of law enforcement; it opens up data links for cross-departmental business collaboration, realizing fully automated closed-loop management from on-site persuasion to administrative law enforcement, significantly improving the intelligence and refinement of social governance.

[0023] Specifically, such as Figure 1 As shown, the core steps of the intelligent inspection method for fire lanes based on vehicle-mounted drones of the present invention include: Mission planning phase: Receive mission instructions, plan the ground travel route of the mobile vehicle platform and the aerial patrol route of the UAV after it arrives at the target area based on the instructions; Close-range deployment phase: The mobile vehicle platform travels to the target fire lane area according to the planned route; Aerial patrol phase: After the vehicle-mounted platform arrives, it releases a drone equipped with a multi-functional pod; the drone flies autonomously along the pre-planned patrol route to perform the patrol mission; Data Acquisition and Transmission: During the cruise, the UAV uses its onboard high-definition camera, thermal imager, lidar and other equipment to collect multi-source perception data (such as images, videos, temperature data and 3D point cloud data) of the target fire lane in real time, and transmits it back to the intelligent analysis platform in real time through wireless communication links. Intelligent Analysis and Risk Assessment: After receiving the data, the intelligent analysis platform performs integrated analysis to identify whether there is any behavior that obstructs fire lanes. If obstruction is identified, it uses a preset multi-dimensional rule model to quantify and score the obstruction based on four dimensions: type of object, estimated duration of obstruction, historical violation records of the location or vehicle, and current degree of obstruction of the lane, and calculates a comprehensive risk index. Risk Level Assessment and Handling: Based on the calculated comprehensive risk index, the system automatically determines the risk level of the occupancy event; finally, according to the determined risk level, the system calls and executes the corresponding handling plan from the preset handling process library (e.g., sending a warning, notifying property management, transferring to law enforcement, etc.).

[0024] The mobile vehicle-mounted platform of this invention solves the problem of limited coverage of traditional fixed monitoring, and can quickly deploy sensing nodes (drones) to any location that needs to be inspected; as an aerial mobile sensor, the drone provides a wide field of view and overcomes the blind spots of ground patrols; the intelligent analysis platform gathers multi-dimensional data from the front end, and transforms physical phenomena (occupancy behavior) into quantitative indicators (risk index) in the digital world through algorithm models, and then maps them to specific business actions (handling procedures) according to preset rules, forming a complete closed loop of perception, analysis, decision-making and action.

[0025] This invention enables the forward deployment of law enforcement forces and the reengineering of law enforcement processes. By using vehicle-mounted drones to quickly approach targets, response time is significantly shortened. Through multi-dimensional fusion analysis, risk assessment becomes more scientific and objective, laying the foundation for precise subsequent handling. The automation of the overall process reduces human intervention and improves the efficiency and standardization of patrol and law enforcement.

[0026] In some embodiments of the present invention, the patrol method of the UAV has been optimized. Specifically, this embodiment further defines the specific way in which the UAV performs patrol missions. After the UAV takes off to perform the patrol mission, the mobile vehicle platform does not remain stationary, but moves to the next preset deployment point according to the mission plan. When the UAV detects that its own battery level is lower than the safe return threshold, or has completed the patrol mission of the current local area, it will automatically fly to the new deployment point where the mobile vehicle platform is located. The automatic hangar on the vehicle platform provides automatic charging services for the UAV, or the robotic arm can quickly replace the battery. After the energy is replenished, the UAV takes off again to continue to perform the remaining patrol missions, and so on, until the patrol mission of the entire target area is completed.

[0027] This embodiment utilizes the principles of leapfrog tactics and relay resupply, treating the mobile vehicle-mounted platform as a mobile energy mothership and forward base. Through dynamic coordination between the vehicle and the drone, the drone's combat radius is no longer limited by its own battery life but extends to the mobility range of the vehicle-mounted platform. The drone's mission is segmented reconnaissance, while the vehicle-mounted platform's mission is mobile resupply; together, they achieve continuous and uninterrupted coverage of a vast area.

[0028] This embodiment completely solves the battery life bottleneck that has plagued the drone industry, enabling a single patrol mission to cover a longer fire lane or to achieve long-term, high-frequency monitoring of the same lane. At the same time, this on-the-fly, charging-while-flying mode greatly improves the system's efficiency and continuity, ensuring uninterrupted patrols and deterrence of key areas during critical periods (such as nighttime and holidays).

[0029] In other embodiments of the present invention, this embodiment elaborates on the specific implementation of the occupant type dimension. A deep learning-based visual recognition model is deployed in the intelligent analysis platform. The visible light video stream transmitted back by the drone is fed into this model, which performs target detection and classification on the occupants in the image. The classification results include, but are not limited to: temporary soft obstacles (such as clothes drying, temporarily stacked cardboard boxes, construction barriers), non-motorized vehicles / motorcycles, small private cars, and large vehicles / fixed facilities (such as vans, concrete blocks, containers). After the model outputs the classification results, the system assigns a corresponding first risk score to different types according to a preset mapping table. α For example, temporary soft barriers. α =0.2, non-motorized vehicles / motorcycles α =0.4, small private car α =0.8, large vehicles / fixed facilities α =1.0 (a higher score indicates a greater risk), see Table 1 for details: This embodiment utilizes computer vision and object classification technology. Through training on a large number of images of fire lane obstruction, the model learns the visual features of the degree of obstruction to passage by different objects. The principle is that different types of objects obstruct the passage of fire and rescue vehicles to completely different degrees. Therefore, it is necessary to distinguish them at the initial stage of risk assessment. Lightweight and movable objects and heavy fixed obstacles represent fundamentally different risk levels.

[0030] This embodiment achieves refined qualitative analysis of occupancy behavior, avoiding the drawbacks of a one-size-fits-all approach to all occupancy behaviors, enabling the system to understand what is blocking the road. This provides physically logical input for subsequent comprehensive scoring and also provides a basis for flexible law enforcement—for example, for temporarily piled-up cardboard boxes, a warning might be issued first, while for deliberately parked large vehicles blocking the passage, a higher-level response might be triggered directly.

[0031] In a further embodiment of the present invention, this embodiment optimizes the above-described method for determining the estimated road occupancy duration. This embodiment details how to determine the estimated road occupancy duration and generate a second risk score. β Once the visual model identifies an object (especially a vehicle), the system initiates a comprehensive judgment process: Instantaneous Behavior Recognition: Combining thermal imaging camera data from the drone, the system determines the thermal characteristics of the vehicle's engine area to infer whether the vehicle is off; it also analyzes whether there is anyone in the driver's cab and the vehicle's lighting status (such as hazard lights); if the system is determined to be off, unoccupied, and without lights, it may have been parked for a long time; if the system is determined to be on, with someone in the vehicle and hazard lights on, it is identified as an instantaneous stop. βThe value is set to a low level; Continuous timing: After the drone detects occupancy, it automatically switches to fixed-point monitoring mode to record the duration of the object's stillness. When the stillness time exceeds a preset threshold (such as 3 minutes), the system automatically increases the duration score. Historical data correction: The system calls the database to query the vehicle's license plate number or the GPS coordinates of the location, analyzing the distribution pattern of the vehicle's parking duration over a past period. For example, if a vehicle frequently parks at this location for more than 2 hours, even if it only parks for 5 minutes currently, the system will increase the weight in the estimated duration calculation, resulting in a higher estimate. β The values ​​are detailed in Table 2: Instantaneous behavior is considered temporary passenger pick-up / drop-off or delivery, with a second risk score of 0.1. If the stationary time exceeds a preset threshold, it is no longer necessary to continue observing and timing the stationary object. The estimated road occupancy time is 5-30 minutes, with a second risk score of 0.5. At the same time, the database is accessed to query the vehicle's license plate number or the GPS coordinates of the location to determine the distribution pattern of the object's parking time over a period of time. If the object has been monitored for long-term occupancy over a period of time (e.g., 30 days), the estimated road occupancy time of the object is greater than 30 minutes, with a second risk score of 1.0.

[0032] This embodiment integrates thermal imaging analysis, behavioral logic judgment, and big data prediction. Thermal imaging provides objective physical conditions (whether the engine is off or not), behavioral logic (person inside the vehicle, hazard lights on) provides a basis for judging subjective intentions, and historical big data reveals behavioral patterns. The combination of these three allows the system to more accurately predict how long the object will remain in a certain location, rather than simply how long it has been there.

[0033] This embodiment represents a key technology for achieving flexible and precise law enforcement. It effectively distinguishes between short-term stops that are merely for immediate departure and long-term, intentional illegal parking, avoiding misjudgments and excessive penalties for vehicles normally picking up or dropping off passengers or temporarily loading or unloading. Furthermore, by incorporating historical data for correction, it enables earlier warnings and more sophisticated responses to repeat offenders, demonstrating the precision and credibility of law enforcement.

[0034] In a preferred embodiment of the present invention, this embodiment optimizes the above-mentioned historical violation record statistics and scoring method. Specifically, this embodiment describes the specific implementation method of the historical violation record dimension. The intelligent analysis platform maintains a violation record database. Whenever the system confirms an occupation event, it records the event based on the identified license plate number (if the object is a vehicle) and geographical coordinates (if there is no license plate, such as a fixed obstacle). When evaluating the current event, the system queries the database for a recent period (e.g., the past 30 days) to count the number of times the same license plate number or the same coordinate point is flagged for road occupancy. Based on the number of counts, the system classifies the violators into three levels: First violation: No historical records found, assigned a third risk score. c For lower values ​​(e.g.) c =0.1); Occasional violations: Those with a history of 2-3 violations or less are given... c The median (e.g., γ=0.5); High-frequency repeated violations: If the history exceeds 3 or more, an penalty will be imposed. c For high values ​​(e.g., γ=1.0), see Table 3 for details: This embodiment utilizes data statistics and behavioral profiling technology to create a compliance file or behavioral profile for each entity (person or location) by mining historical big data. The underlying logic is the legal principle of "repeat offenses are punished more severely" and the risk management concept: an entity that frequently violates regulations is far more likely to violate regulations again and poses a much greater potential threat than an occasional violator.

[0035] This embodiment endows the system with learning and memory capabilities, moving beyond simple reactions to isolated events to considering current events within a historical context. This allows for more precise allocation of law enforcement resources to areas and vehicles with high-frequency violations, enabling targeted control of key individuals, effectively deterring potential violators, and enhancing the long-term effectiveness of governance.

[0036] In some embodiments of the present invention, this embodiment optimizes the above-described method for quantifying and evaluating channel congestion. Specifically, this embodiment describes how to accurately quantify the degree of channel congestion and generate a fourth risk score. dThe drone's onboard LiDAR scanner scans the fire lane during its patrol, acquiring high-precision 3D point cloud data. The point cloud processing algorithm in the intelligent analysis platform processes this data, first identifying the ground, side walls, or curbs of the lane, then calculating the width of the remaining space after it is occupied. The system compares the calculated remaining effective width with the design width of the lane stored in the GIS system to determine the congestion rate (congestion rate = 1 - remaining width / design width). Based on the congestion rate, the system classifies the degree of congestion into three levels: Minor blockage: Blockage rate <30%, the remaining width is still sufficient for most small fire trucks to pass through, corresponding to δ=0.1; Moderate congestion: The congestion rate is between 30% and 70%, and only small vehicles can barely pass through, corresponding to δ=0.6; Severe congestion: Congestion rate > 70%, remaining width is insufficient for any fire truck to pass through, corresponding to δ = 1.0, see Table 4 for details: This embodiment is based on lidar ranging and 3D reconstruction technology. LiDAR emits a laser beam and measures its return time to accurately obtain the 3D spatial coordinates of an object's surface, unaffected by lighting conditions. By comparing design drawings and real-time point clouds, the system can calculate the actual passage capacity of the channel with centimeter-level accuracy, transforming the vague concept of congestion into a precise mathematical expression of how much congestion there is.

[0037] This embodiment provides the most objective and robust physical evidence for risk assessment. The congestion rate is an irrefutable quantitative indicator that directly relates to the success or failure of fire and rescue operations. This quantitative result is not only used to calculate the comprehensive risk index, but more importantly, it provides indisputable law enforcement evidence for severe congestion incidents, giving solid technical support to the rigid enforcement of veto power.

[0038] In a further embodiment of the present invention, the above-mentioned comprehensive judgment and graded handling method is optimized. Specifically, the above-mentioned risk calculation, grade judgment and handling process are integrated into a complete decision and execution unit.

[0039] The method for calculating the comprehensive risk index includes: Where: R – Comprehensive Risk Index; α —First risk score; β —Second risk score; c —Third risk score; d—Fourth risk score.

[0040] Event severity assessment: Based on the comprehensive risk index R The magnitude of the value is used to initially determine the event level; for example, R<1 is defined as a level 1 event, 1≤R<3 as a level 2 event, and R≥3 as a level 3 event; however, the system also executes a veto logic: if the blocking level is severe blocking (i.e., d =1.0) or historical violation records are high-frequency repeated violations (i.e. c If the comprehensive risk index R value is 1.0, then regardless of the magnitude of the comprehensive risk index R value, it will be directly judged as a level 3 event.

[0041] Tiered handling process: Level 1 Event: The system triggers the drone's onboard loudspeaker to remotely issue a voice warning (e.g., this is a fire lane, do not obstruct it, please leave immediately), and simultaneously starts a countdown monitoring (e.g., 5 minutes); if the obstruction is cleared within the countdown, the event is closed; if the obstruction is not cleared within the time limit, the system automatically escalates the event to Level 2. Level 2 Event: The system automatically generates an electronic work order containing the event time, location, photos, videos, and risk score, and pushes it to the local management personnel responsible for the area (such as property managers or community grid workers) via APP or SMS; the responsible personnel must go to handle the situation within the specified time and upload on-site images after the handling for the platform's AI verification; once the verification is passed, the event is closed. Level 3 Event: The system automatically packages all evidence from the discovery of the event to the present moment (including visible light video, thermal imaging screenshots, LiDAR point cloud data, historical violation records, warning records, work order records, etc.) into a structured electronic evidence package; through a standardized API interface, the evidence package is directly pushed to the law enforcement platform of administrative law enforcement departments such as public security or fire and rescue, so that the law enforcement departments can impose penalties in accordance with the law.

[0042] This embodiment constructs a complete decision tree and automated workflow engine, which transforms complex business rules (such as weight calculation, veto, and tiered handling) into machine-executable code logic; by determining the event level, events of different natures and risks are diverted to different processing channels, thereby achieving optimized allocation of handling resources and standardization and automation of handling processes.

[0043] This embodiment realizes an ideal governance model of gentle persuasion for minor incidents, closed-loop management for general incidents, and forceful enforcement for serious incidents. The persuasion function enhances the approachability of law enforcement, the work order push strengthens local management responsibility, and the cross-departmental push breaks down the last mile of administrative law enforcement. The entire process is efficient, standardized, and traceable, significantly improving the level of intelligence in social governance.

[0044] The intelligent fire lane inspection method based on vehicle-mounted drones of the present invention, when actually implemented, follows the following process in practical application: Step 1: Task Initiation and Planning The city's fire management department issues an order to inspect fire lanes in a key area; the task scheduling engine of the intelligent analysis platform receives the order and, automatically or with operator assistance, plans the optimal route for the mobile vehicle platform from its current location to the target area; at the same time, based on the distribution and length of fire lanes in the target area, it plans the segmented patrol routes of the drones and sets up drone energy replenishment points (i.e., mobile deployment points of the vehicle platform) along the way.

[0045] Step Two: Approach and Relay Cruise The mobile vehicle-mounted platform carries drones and travels along a planned route to the target area. Upon arrival at the first deployment point, the platform releases the drones. The drones take off and fly autonomously along the first segment of the patrol route, with their multi-functional pods commencing operation. When the drone's battery level drops below a threshold or after completing the first segment of the route, it automatically flies to the second deployment point already reached by the vehicle-mounted platform for automatic charging or battery swapping, before taking off again to execute the next segment of the route. This process is repeated to achieve uninterrupted, full-coverage patrol of the entire fire lane.

[0046] Step 3: Multi-source data acquisition and fusion During the cruise, the drone simultaneously collects multi-dimensional data: High-definition camera: Captures visible light video and images of the passageway to identify the type of occupancy; Thermal imaging camera: Acquires thermal radiation images of vehicles and other occupants to determine their operating status (off / on); LiDAR: Scans the channel and generates 3D point cloud data for precise measurement of the remaining width; This data is transmitted back to the intelligent analysis platform in real time with low latency via a dedicated network link.

[0047] Step 4: Multidimensional Risk Assessment and Level Determination The intelligent analysis platform's analysis engine processes the received data in real time: Occupancy type analysis: The visual recognition model classifies occupants into temporary soft obstacles, non-motorized vehicles, small private cars, or large vehicles, and assigns them a first risk score. α ; Lane occupancy duration analysis: Combining thermal imaging and behavioral logic, it determines whether the parking is momentary or continuous; if the continuous occupancy exceeds a threshold, and after correcting by calling historical parking data for that license plate, the estimated lane occupancy duration is obtained, and a second risk score is assigned. β ; Historical record analysis: Query the database to count the number of violations committed by this license plate or coordinate point, determine whether it is a first-time offense, an occasional offense, or a high-frequency offense, and assign a third risk score. c ; Congestion level analysis: The ratio of the remaining effective width to the designed width is calculated using point cloud data to determine the congestion rate, classifying it as slight, moderate, or severe congestion, and assigning a fourth risk score. d ; Subsequently, the system calculates the comprehensive risk index. R=a+b+c+d Finally, determine the event severity level: Standard assessment: Based on the R value, it is initially classified into three levels: Level 1, Level 2, and Level 3. One-vote veto: If the level of congestion is severe or the historical record is high frequency, it will be directly judged as a level three event.

[0048] Step 5: Automated closed-loop processing Based on the determined event level, the alert and workflow engine initiates the corresponding handling procedures: Level 1 incident (low risk, such as temporary storage or brief shutdown): The drone automatically flew close to the scene and delivered remote voice persuasion via a loudspeaker. The system starts a countdown monitoring to observe whether the occupant has moved; If the event is resolved within the countdown, the system records the event and then closes the loop. If the issue is not resolved within the time limit, the event will automatically escalate to level two. Level 2 events (medium risk, such as prolonged occupation and low frequency): The system automatically generates electronic work orders that include event details and evidence; Work orders are pushed to the local management personnel (such as property managers) in the area via the APP. The responsible person must be present at the scene to handle the situation and upload photos / videos of the rectification. The platform's AI automatically compares the verification photos, and once the verification is successful, the process is closed. Level 3 event (high risk, such as severe congestion or repeated violations): The system immediately initiates the evidence preservation procedure, packaging all data from the entire event (high-definition video, thermal imaging, point cloud data, historical records, persuasion records, work order records) into a complete electronic evidence package with a timestamp and digital signature; Through the cross-departmental data exchange interface, the evidence package is automatically pushed to the case-handling system of the administrative law enforcement department (such as fire and rescue agencies or traffic police), and the law enforcement department shall impose administrative penalties in accordance with the law.

[0049] In addition, the present invention also relates to an intelligent fire lane inspection and early warning system based on a vehicle-mounted drone that implements all the aforementioned methods; the system consists of four core parts: Mobile vehicle platform: As the mobile carrier and ground support center of the system, it integrates an automated hangar for parking and taking off and landing drones, a communication relay device for long-distance communication with drones and intelligent analysis platforms, an edge computing unit for performing preliminary data processing locally to reduce transmission latency, and a power supply system to power the entire vehicle platform and drones. Unmanned Aerial Vehicle (UAV) Subsystem: As the system's aerial mobile sensing terminal, it is equipped with a high-precision positioning module, an autonomous obstacle avoidance system to ensure flight safety, and a multi-functional pod; the pod integrates a high-definition zoom camera, a thermal imaging camera, and a lidar, and can simultaneously collect multi-source sensing data; Collaborative control network: As an information superhighway connecting all nodes, it includes public network communication links using 4G / 5G networks for transmitting non-real-time business data and status information; it also includes private network communication links using microwave or proprietary frequency bands for transmitting high-definition video streams, lidar point cloud data, and key flight control commands from drones to ensure high bandwidth and low latency communication requirements. Intelligent Analysis Platform: As the brain of the system, it includes an analysis engine responsible for performing image recognition, point cloud analysis, and risk calculation; a task scheduling engine responsible for planning tasks and scheduling drones and vehicle platforms; an early warning and workflow engine responsible for executing graded handling processes, sending notifications, and connecting with external systems; and a visual data management interface that provides data display, system configuration, and historical query functions.

[0050] This system embodies the modern IoT architecture principle of cloud, edge, and terminal collaboration. The drone subsystem (terminal) is responsible for data collection; the mobile vehicle platform (edge) provides nearby physical support and preliminary data processing; the intelligent analysis platform (cloud) gathers all data and performs global intelligent analysis and decision scheduling; and the collaborative network ensures the smooth and reliable flow of information between all levels. The four are organically combined to form a complete closed loop of the system from the physical world to the digital world and then to business execution.

[0051] This system forms a complete product solution that can be implemented and deployed. The automated hangar and edge computing unit improve the system's automation level and response speed; the communication design that combines public and private networks balances cost and performance; the modular intelligent analysis platform design ensures the system's scalability and maintainability; the system as a whole achieves deep coupling between hardware and software, and can stably and efficiently perform intelligent inspection and early warning tasks for fire lanes.

[0052] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent inspection of fire lanes based on vehicle-mounted drones, characterized in that: include: Based on mission instructions, plan the travel route of the mobile vehicle platform and the drone patrol route within the target area; After the mobile vehicle platform arrives at the target area, it releases a drone equipped with a multi-functional pod, which then performs a patrol mission according to the planned patrol route. The drone collects multi-source sensing data of the target fire lane and transmits it back to the intelligent analysis platform in real time; The intelligent analysis platform fuses and analyzes the multi-source sensing data to identify whether there is any occupancy behavior; If there is an occupation, a comprehensive risk index is calculated based on four dimensions: type of object occupied, estimated duration of occupation, historical violation records, and degree of channel obstruction, and then the event level is determined. Develop corresponding handling procedures based on the event level.

2. The intelligent inspection method for fire lanes based on vehicle-mounted drones according to claim 1, characterized in that: The method for the UAV to perform patrol missions according to the planned patrol route includes: the mobile vehicle platform moves to the next deployment point during the UAV's mission; when the UAV's battery level is below a threshold or after completing a local area patrol, the UAV returns to the vehicle platform at the next deployment point to recharge or replace the battery, and then takes off again to continue performing the subsequent flight segment mission.

3. The intelligent inspection method for fire lanes based on vehicle-mounted drones according to claim 1, characterized in that: The types of occupants are classified using a visual recognition model into temporary soft obstacles, non-motorized vehicles / motorcycles, and small private cars or large vehicles / fixed facilities, and each is assigned a different first risk score. α .

4. The intelligent inspection method for fire lanes based on vehicle-mounted drones according to claim 1, characterized in that: The method for determining the estimated road occupancy duration includes: using thermal imaging to determine whether the vehicle is turned off, whether the driver is inside the vehicle, and the status of the lights to identify momentary stopping behavior; if the object occupying the road remains stationary for more than a preset time threshold, the duration score is increased; and the score is corrected by combining the historical parking duration distribution of the license plate or geographical location to obtain the estimated road occupancy duration, and a corresponding second risk score β is obtained based on the estimated road occupancy duration.

5. A method for intelligent inspection of fire lanes based on vehicle-mounted drones according to claim 1, characterized in that: The historical violation records are statistically analyzed based on the number of times the same license plate number or the same geographical coordinate point is blocked within a set period of time, and are divided into three levels: first violation, occasional violation, and high-frequency repeated violation, which correspond to the third risk score γ.

6. The intelligent patrol method for fire lanes based on vehicle-mounted drones according to claim 1, characterized in that: The degree of channel obstruction is determined by acquiring three-dimensional point cloud data using lidar, calculating the ratio of the remaining effective width of the fire lane to the design width, obtaining the obstruction rate, and classifying it into three levels: slight obstruction, moderate obstruction, or severe obstruction, which correspond to the fourth risk score δ.

7. A method for intelligent inspection of fire lanes based on vehicle-mounted drones according to claim 1, characterized in that: The method for calculating the comprehensive risk index includes: Where: R – Comprehensive Risk Index; α —First risk score; β —Second risk score; γ —Third risk score; δ —Fourth risk score.

8. A method for intelligent inspection of fire lanes based on vehicle-mounted drones according to claim 7, characterized in that: The method for determining the event level includes: judging the event level from light to severe based on the comprehensive risk index as Level 1 event, Level 2 event and Level 3 event. However, if the degree of blockage is severe blockage or the historical violation record is high-frequency repeated violation, then regardless of the size of the comprehensive risk index, it is directly judged as a Level 3 event.

9. A method for intelligent inspection of fire lanes based on vehicle-mounted drones according to claim 8, characterized in that: The method for formulating corresponding handling procedures based on the event level includes: triggering a drone to remotely issue a warning and start a countdown monitoring; if the situation is not resolved, it will automatically escalate to a level two event. Level 2 incidents automatically generate electronic work orders and push them to the terminal of the local management personnel, requiring them to upload post-handling images for verification; Level 3 incidents are automatically packaged into complete electronic evidence and pushed to the administrative law enforcement platform through cross-departmental interfaces.

10. A fire lane intelligent patrol and early warning system based on a vehicle-mounted unmanned aerial vehicle (UAV) for implementing the method as described in any one of claims 1 to 9, characterized in that, include: The mobile vehicle-mounted platform integrates an automated hangar, communication relay equipment, edge computing units, and a power supply system. The unmanned aerial vehicle (UAV) subsystem is equipped with a positioning module, an autonomous obstacle avoidance system, and a multi-functional pod that integrates a high-definition zoom camera, a thermal imaging camera, and a lidar. The collaborative control network, including public network communication links and private network communication links, is used to transmit different types of data respectively to ensure high bandwidth and low latency communication requirements. The intelligent analysis platform includes an analysis engine, a task scheduling engine, an early warning and workflow engine, and a visual data management interface, used to perform multi-dimensional risk assessments and automated closed-loop handling.

Images