An unmanned aerial vehicle patrol system and a smart-enabled city emergency device

Abstract

The application discloses an unmanned aerial vehicle patrol system and a smart-enabled city emergency equipment, belongs to the technical field of unmanned aerial vehicle patrol systems, and comprises an edge cooperative emergency command platform, a multi-mode redundant self-organizing communication module, an anti-interference multi-source fusion sensing and positioning module and a cross-domain heterogeneous cluster cooperation module; the multi-mode redundant self-organizing communication module is used for fusing three communication modes of a Beidou short message, a LoRa ad hoc network and an unmanned aerial vehicle relay, constructing an air-ground integrated multi-hop self-organizing communication network without dependence on a ground public network, and completely getting rid of dependence on ground communication infrastructure, so that the technical pain point of unmanned aerial vehicle disconnection under extreme conditions is effectively solved. The Beidou short message is used as a bottom-line communication means to realize long-distance emergency instruction transmission; the LoRa ad hoc network realizes decentralized multi-hop communication and supports automatic reconstruction after network node failure; and the unmanned aerial vehicle relay node is dynamically deployed to fill in a signal coverage blind area and realize no communication dead angle in an operation area.

Description

Technical Field

[0001] This application belongs to the technical field of drone patrol systems, specifically, it relates to a drone patrol system and smart-enabled urban emergency equipment. Background Technology

[0002] With the acceleration of urbanization, the continuous expansion of urban scale, the density of high-rise buildings, and the networking of infrastructure, the frequency of sudden major disasters such as earthquakes, floods, and fires has increased significantly, and the difficulty of handling them has increased. Urban emergency patrol and rescue work faces severe challenges. With its advantages of maneuverability, rapid response, and no need for ground infrastructure support, drones have been widely used in urban emergency patrols, disaster reconnaissance, and location of trapped personnel, becoming one of the core equipment in the urban emergency system.

[0003] However, existing drone patrol systems still have certain shortcomings in extreme emergency scenarios, as follows:

[0004] Insufficient communication reliability and susceptibility to loss of connection in extreme scenarios: Existing drone patrol systems generally rely on public ground networks or conventional radios for data transmission and command exchange. However, major disasters (such as earthquakes and floods) often paralyze ground communication infrastructure, causing drones to lose contact with command centers and ground rescue terminals. Patrol data cannot be uploaded, and command instructions cannot be issued, directly leading to the interruption of emergency patrol missions and delaying rescue opportunities. Even if some drones use a single backup communication method, they still suffer from limited coverage, weak anti-interference capabilities, and the inability to achieve multi-node collaborative communication, making them difficult to adapt to extreme environments.

[0005] Poor adaptability to harsh environments and insufficient perception and positioning accuracy: Harsh environments such as dense smoke in urban fires, torrential rain and floods, and darkness at night severely interfere with the operation of traditional optical sensors, causing target recognition failure. Simultaneously, in scenarios where GPS signals are blocked, such as densely populated high-rise buildings and underground spaces, existing UAV positioning systems mostly rely on a single GPS positioning system, resulting in large positioning errors. This makes it impossible to accurately locate the floor, window, or other three-dimensional spatial information of trapped personnel, hindering the provision of precise location guidance for rescue operations and significantly reducing rescue efficiency. Furthermore, existing sensors are mostly of a single type, lacking environmental adaptive adjustment capabilities, and are prone to failure in complex environments, further impacting system reliability.

[0006] Low efficiency in cross-domain collaboration and weak multi-drone swarm collaboration capabilities: Urban emergency response involves multiple departments such as public security, fire protection, medical care, and emergency management. Existing drone patrol systems are mostly deployed independently by a single department. The inconsistent data formats and incompatible communication protocols among departments result in significant delays in emergency data exchange (reportedly exceeding 20 minutes). Disaster data collected by drones cannot be synchronized with all rescue units in real time, easily leading to command confusion and unreasonable allocation of rescue forces. Furthermore, when multiple drones operate in complex urban environments, the lack of effective route planning and conflict avoidance mechanisms easily leads to problems such as route intersections and collisions, making it difficult to form swarm combat capabilities and achieve coordinated advancement of wide-area coverage and detailed reconnaissance. Summary of the Invention

[0007] To address the aforementioned problems and technical deficiencies, this application adopts the following technical solution: a drone patrol system, comprising an edge-collaborative emergency command platform, a multi-mode redundant self-organizing communication module, an anti-interference multi-source fusion sensing and positioning module, and a cross-domain heterogeneous cluster collaborative module.

[0008] The multi-mode redundant self-organizing communication module, the anti-interference multi-source fusion sensing and positioning module, and the cross-domain heterogeneous cluster collaboration module are all bidirectionally connected to the edge collaborative emergency command platform.

[0009] The multi-mode redundant self-organizing communication module is used to construct an air-ground integrated multi-hop self-organizing communication network without reliance on the terrestrial public network, providing the system with redundant data transmission channels under extreme disaster scenarios;

[0010] The anti-interference multi-source fusion sensing and positioning module is used to collect scene and target data in complex and harsh environments to achieve accurate three-dimensional spatial positioning in GPS signal-blocked scenarios.

[0011] The cross-domain heterogeneous cluster collaboration module is used to realize real-time data exchange among multiple emergency departments and collaborative scheduling of multiple drone clusters.

[0012] The edge-collaborative emergency command platform is used to integrate and process the data uploaded by each module, issue collaborative control commands, and form a closed-loop emergency patrol system of perception, communication, collaboration, and decision-making.

[0013] Preferably, the multi-mode redundant self-organizing communication module includes a BeiDou short message communication unit, a LoRa self-organizing network communication unit, an UAV relay communication unit, and a dynamic spectrum adaptive unit;

[0014] The Beidou short message communication unit is used for long-distance backup command and data transmission in extreme network outage scenarios;

[0015] The LoRa self-organizing network communication unit is used for decentralized multi-hop communication between drones and between drones and ground emergency terminals, and supports automatic route reconstruction after network node failure.

[0016] The UAV relay communication unit is used to dynamically deploy airborne relay nodes to fill signal coverage blind spots in high-rise buildings and disaster core areas.

[0017] The dynamic spectrum adaptive unit is used to detect environmental electromagnetic interference in real time and automatically switch communication frequency bands and adjust transmission power.

[0018] Furthermore, the multi-mode redundant self-organizing communication module also integrates a lightweight blockchain consensus unit, which is used to perform distributed encrypted storage and integrity verification of transmitted emergency data to ensure reliable data exchange among multiple departments.

[0019] Preferably, the anti-interference multi-source fusion sensing and positioning module includes a millimeter-wave radar sensor, an infrared thermal imaging sensor, a lidar sensor, an ultrasonic sensor, and a multi-sensor data fusion processing unit.

[0020] The multi-sensor data fusion processing unit is used to perform spatiotemporal registration and fusion processing on multi-source data collected by four types of sensors, and output target recognition results under severe environments such as dense smoke, heavy rain, and nighttime.

[0021] Furthermore, the anti-interference multi-source fusion sensing and positioning module also includes a triple positioning calibration unit, which integrates a BeiDou RTK positioning subunit, a visual inertial odometry (VIO) positioning subunit, and a city 3D model matching subunit.

[0022] The Beidou RTK positioning subunit is used for centimeter-level absolute positioning in unobstructed environments;

[0023] The visual inertial odometry (VIO) positioning subunit is used for relative positioning in environments where GPS signals are obscured.

[0024] The city 3D model matching subunit is used to match the lidar scanning data with the pre-stored city 3D building model and output the 3D positioning result of the target's floor and spatial location.

[0025] Preferably, the anti-interference multi-source fusion sensing and positioning module is further provided with a harsh environment adaptive adjustment unit, which is used to dynamically adjust the working weight and operating parameters of each sensor according to real-time environmental parameters.

[0026] Preferably, the cross-domain heterogeneous cluster collaboration module includes a cross-domain data interoperability unit, a heterogeneous cluster scheduling unit, and a collaborative instruction intelligent allocation unit.

[0027] The cross-domain data interoperability unit is bidirectionally connected to the edge collaborative emergency command platform to achieve decentralized fusion of emergency data from multiple emergency departments and real-time synchronization across all nodes.

[0028] The cross-domain data interoperability unit adopts a federated learning framework to achieve feature sharing and collaborative processing of raw data from multiple departments without leaving their respective domains, with a cross-departmental data interoperability delay of no more than 3 seconds.

[0029] Furthermore, the heterogeneous cluster scheduling unit includes a digital twin scenario construction subunit and a distributed conflict avoidance subunit;

[0030] The digital twin scenario construction subunit is used to build a digital twin of the urban emergency scenario, which maps the drone's location, flight path, operating status and perception data in real time.

[0031] The distributed conflict avoidance subunit is used to dynamically plan conflict-free routes based on the UAV's mission type and flight parameters.

[0032] The heterogeneous cluster scheduling unit adopts a master-slave collaborative mode, supporting at least 50 heterogeneous UAVs to work collaboratively at the same time, with a flight path conflict rate of no more than 0.1%.

[0033] Furthermore, the multi-mode redundant self-organizing communication module provides an end-to-end low-latency transmission channel for the anti-interference multi-source fusion sensing and positioning module and the cross-domain heterogeneous cluster collaborative module, with an end-to-end transmission delay of no more than 50ms; the system has a communication success rate of no less than 99% under extreme disaster conditions such as power outage, network outage, and circuit outage, a target recognition accuracy of no less than 98% in harsh environments, and a three-dimensional positioning error of no more than 5cm.

[0034] A smart urban emergency response device, including the aforementioned drone patrol system.

[0035] Compared to existing technologies, the beneficial effects of this application are as follows:

[0036] (1) This application integrates three communication methods—BeiDou short message, LoRa self-organizing network, and UAV relay—through a multi-mode redundant self-organizing communication module to construct an air-ground integrated multi-hop self-organizing communication network that does not rely on the ground public network. This completely eliminates the dependence on ground communication infrastructure and effectively solves the technical pain point of UAV disconnection under extreme conditions. Among them, BeiDou short message serves as a backup communication method to realize long-distance emergency command transmission; LoRa self-organizing network realizes decentralized multi-hop communication and supports automatic reconstruction after network node failure; and UAV relay nodes are dynamically deployed to fill signal coverage blind spots and realize no communication dead zones in the operation area.

[0037] Meanwhile, the dynamic spectrum adaptive unit can detect electromagnetic interference in real time, automatically switch communication frequency bands and adjust transmission power to improve anti-interference capabilities; the lightweight blockchain consensus unit realizes encrypted storage and verification of emergency data, ensuring reliable data exchange among multiple departments.

[0038] (2) This application adopts an anti-interference multi-source fusion sensing and positioning module, which integrates four anti-interference sensors: millimeter-wave radar, infrared thermal imaging, lidar, and ultrasonic. It achieves multi-source data fusion through an improved federated Kalman filter algorithm, effectively overcoming the interference of harsh environments such as dense smoke, heavy rain, and night to optical sensors. The target recognition accuracy is not less than 98%, solving the problem of sensing failure in harsh environments of existing systems.

[0039] In addition, the triple positioning calibration unit integrates Beidou RTK, visual inertial odometry, and urban 3D model matching technology to achieve accurate positioning in all scenarios: centimeter-level absolute positioning in unobstructed environments, relative positioning within ±5cm in GPS signal obstruction scenarios, and accurate identification of the floor and window location of trapped personnel by matching LiDAR point cloud with pre-stored urban 3D model. This effectively solves the pain points of traditional positioning, such as large errors and inability to achieve accurate 3D spatial positioning, providing accurate location guidance for rescue operations, greatly improving rescue efficiency, and reducing casualties.

[0040] (3) This application constructs a cross-domain data communication unit based on the federated learning framework through a cross-domain heterogeneous cluster collaboration module, realizing the decentralized fusion and real-time synchronization of emergency data from multiple departments. The original data does not leave the domain, and only the feature parameters are shared. This not only ensures data privacy and security, but also effectively reduces the delay of cross-departmental data communication, solving the problems of inefficient data communication and chaotic command in multiple departments.

[0041] The heterogeneous cluster scheduling unit employs digital twin visualization and an improved speed obstacle method combined with an artificial potential field method to achieve efficient scheduling of heterogeneous UAV clusters. It supports at least 50 heterogeneous UAVs to work collaboratively simultaneously with a flight path conflict rate of no more than 0.1%. The master-slave collaborative mode and cluster management mechanism realize optimal task allocation and real-time control of cluster status, thereby significantly improving the operational efficiency of UAV clusters and forming a cluster combat capability for wide-area patrol, detailed reconnaissance, and material delivery. Emergency response efficiency is also effectively improved. Attached Figure Description

[0042] In the attached diagram:

[0043] Figure 1 This is a system diagram from an embodiment of this application. Detailed Implementation

[0044] 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 some embodiments of this application, but not all embodiments. Generally, the components of the embodiments of this application described and shown in the accompanying drawings can be arranged and designed in various different configurations.

[0045] Example 1

[0046] like Figure 1 As shown, a drone patrol system comprises four core components: an edge-collaborative emergency command platform, a multi-mode redundant self-organizing communication module, an anti-interference multi-source fusion sensing and positioning module, and a cross-domain heterogeneous cluster collaboration module.

[0047] The multi-mode redundant self-organizing communication module, the anti-interference multi-source fusion sensing and positioning module, and the cross-domain heterogeneous cluster collaboration module all establish bidirectional communication connections with the edge collaborative emergency command platform.

[0048] The system hardware includes: one ground main command server, one cluster master UAV, up to 100 heterogeneous slave UAVs, and several ground emergency terminals (handheld terminals and command vehicle terminals) for public security, fire protection, and medical departments.

[0049] The system software architecture is divided into four layers: edge layer, network layer, platform layer, and application layer, forming a closed-loop emergency patrol system that integrates perception, communication, collaboration, and decision-making.

[0050] The multi-mode redundant self-organizing communication module is used to construct an air-to-ground integrated multi-hop self-organizing communication network without reliance on the terrestrial public network, providing the system with redundant data transmission channels under extreme disaster scenarios. Its core unit hardware and functional implementation include:

[0051] Beidou Short Message Communication Unit: Employs the Hexin Xingtong UM220-IVN Beidou-3 dual-frequency module, supporting Beidou-3 short message communication. The maximum transmission size per packet is 140 bytes, and the maximum transmission distance without relay is 1200km, serving as a backup communication channel in extreme network outage scenarios. In this embodiment, this unit prioritizes the transmission of core data such as emergency command instructions and the coordinates of trapped personnel. The heartbeat packet transmission cycle is 10 seconds, and the emergency command packet has the highest priority, allowing for priority transmission to ensure that the command link remains uninterrupted even in the event of network outages.

[0052] The LoRa self-organizing network communication unit uses the Semtech SX1262 RF chip, supporting tri-band switching of 433MHz / 868MHz / 915MHz, with an adjustable transmit power range of 10mW-100mW, a maximum single-hop line-of-sight transmission distance of 5km, and supports up to 10 hops for multi-hop transmission. The routing protocol adopts an improved OLSRv2 adaptive link-state routing protocol. When a drone node in the network fails, the route reconstruction time is ≤200ms, enabling automatic network rerouting without the need for ground base station support.

[0053] Unmanned Aerial Vehicle (UAV) Relay Communication Unit: A hexacopter relay UAV is used, equipped with a dual-channel LoRa communication module and a BeiDou short message module. Deployment altitude is 100m-300m, and the deployment location is dynamically adjusted by the edge-coordinated emergency command platform based on the location of signal blind spots. In this embodiment, for signal-obstructed areas such as high-rise building canyons, underground space entrances, and disaster core areas, 2-3 relay UAVs are deployed to form a relay link, filling signal coverage blind spots and achieving no communication dead zones in a 5km×5km operational area.

[0054] Dynamic spectrum adaptive unit: Employing the AD9361 broadband RF transceiver, it scans the entire 400MHz-1GHz frequency band in real time, identifies interference signals through an energy detection algorithm, and sets the interference power threshold to -80dBm. When interference in the current operating frequency band exceeds the threshold, it automatically switches to an idle frequency band with a switching time of ≤50ms. Simultaneously, it adjusts the transmit power in real time according to the link signal-to-noise ratio, reducing power consumption and minimizing interference in multi-device communication while ensuring communication quality.

[0055] Lightweight Blockchain Consensus Unit: Addressing the limited computing power of drones, an improved lightweight PBFT (Practical Byzantine Fault Tolerance) consensus algorithm is employed. The cluster supports a maximum of 50 consensus nodes, with a single-round consensus latency of ≤100ms. In this embodiment, all transmitted emergency data is packaged into block data. The block structure includes: data hash value, timestamp, sending node ID, receiving node ID, and digital signature. The SM2 national cryptographic algorithm is used for encrypted storage and transmission. During multi-department data exchange, each department acts as a consensus node. Data sharing requires node signature verification to ensure data immutability and non-repudiation, resolving the issue of reliable cross-departmental data exchange.

[0056] This module has an end-to-end data transmission latency of ≤50ms and a communication success rate of ≥99% under extreme conditions such as power outages, network outages, and circuit outages.

[0057] The anti-interference multi-source fusion sensing and positioning module is used to collect scene and target data in complex and harsh environments, and to achieve accurate three-dimensional spatial positioning in scenarios where GPS signals are blocked.

[0058] Multi-source sensing hardware and fusion algorithm implementation:

[0059] Sensor hardware configuration: All drones are equipped with a coaxially mounted multi-source sensing kit, with an IP67 protection rating and an operating temperature range of -20℃ to +60℃. Specific configurations are shown in the table below.

[0060]

[0061] Spatiotemporal registration and multi-sensor fusion algorithms:

[0062] First, spatiotemporal registration was performed on the multi-source sensor data: time synchronization was triggered by the PPS second pulse of the UAV flight control, with a time synchronization error ≤1ms; spatial registration was performed using the hand-eye calibration method, solving for the transformation matrix from each sensor coordinate system to the UAV body coordinate system. The transformation formula is as follows:

[0063]

[0064] in, The coordinates of a three-dimensional point in the sensor coordinate system. The coordinates of a three-dimensional point in the body coordinate system. This is the rotation matrix from the sensor to the machine body. It is a translation vector, which is solved in advance through calibration.

[0065] The data fusion employs an improved federated Kalman filter algorithm, setting up one master filter and four corresponding local filters for each single sensor. Local state estimation is performed first, followed by global optimal fusion. The core formula is as follows:

[0066] Local filter state update:

[0067]

[0068]

[0069]

[0070] in, For the first Local filters of individual sensors State estimate at time 10:00 These are sensor observations. For the observation matrix, For Kalman gain, Let be the state covariance matrix. To observe the noise covariance matrix.

[0071] Global fusion of the main filter:

[0072] ;

[0073] ;

[0074] in This is the globally optimal state estimate. This is the global covariance matrix.

[0075] Harsh Environment Adaptive Adjustment Unit: Equipped with a smoke concentration sensor, a rainfall sensor, and a wind speed sensor, it collects environmental parameters in real time and dynamically adjusts the working weight of each sensor based on the environmental parameters. The weight adjustment formula is as follows:

[0076]

[0077] in As the initial weights of the sensors, This is the environmental impact factor. This represents the confidence level of the sensor in the current environment; for example, the confidence level of a visible light camera in a dense smoke environment (smoke concentration ≥ 500 ppm). =0.1, weight automatically decreases; millimeter-wave radar, infrared thermal imaging confidence level =0.9, the weight is automatically increased to ensure that the target recognition accuracy is ≥98% in harsh environments.

[0078] The triple positioning calibration unit integrates a Beidou RTK positioning subunit, a visual inertial odometry (VIO) positioning subunit, and a city 3D model matching subunit to achieve accurate 3D spatial positioning in the absence of GPS, with a positioning error of ≤5cm.

[0079] Beidou RTK Positioning Subunit: Using Huace Navigation K803 Beidou-3 RTK module, it achieves an absolute positioning accuracy of ±1cm in an unobstructed open environment, serving as the main positioning source in scenarios without signal obstruction.

[0080] Visual Inertial Odometry (VIO) Positioning Subunit: Employing a combination of binocular camera and six-axis IMU hardware, and based on an improved VINS-Mono algorithm, it achieves a relative positioning accuracy of ±5cm in densely populated high-rise buildings and underground spaces where GPS signals are blocked, with a position drift rate of ≤0.5% / km, serving as a backup positioning source in GPS failure scenarios.

[0081] The city 3D model matching sub-unit uses a pre-stored LOD3 level city 3D building model, containing precise information such as the number of floors, floor height, window positions, and structural dimensions for each building. The ICP iterative nearest point algorithm is employed to register the real-time point cloud data from LiDAR scans with the pre-stored 3D model, determining the absolute 3D positions of the UAV and target within the building model. The core objective function is as follows:

[0082] ;

[0083] in Point cloud data scanned in real time by LiDAR, To pre-store the point cloud data of the 3D model, For rotation matrix, The vector is the translation vector; the optimal solution is obtained through Singular Value Decomposition (SVD). After registration, the floor, window number, and spatial coordinates of the trapped person can be accurately output, solving the problem that traditional positioning methods cannot achieve accurate three-dimensional spatial positioning.

[0084] The cross-domain heterogeneous cluster collaboration module is used to realize real-time data exchange among multiple emergency departments and collaborative scheduling of multiple drone clusters. The key requirement is to achieve the collaborative operation of at least 50 heterogeneous drones in the master-slave mode with a flight path conflict rate of ≤0.1%.

[0085] The cross-domain data interoperability unit is based on edge computing and federated learning framework to achieve decentralized fusion of emergency data from multiple departments and real-time synchronization across all nodes.

[0086] Hardware deployment: Each department, including public security, fire protection, medical care, and emergency management, deploys one edge computing node. The nodes use Jetson Xavier NX edge computing boxes and communicate bidirectionally with the edge collaborative emergency command platform.

[0087] Technical Implementation: The FedAvg federated averaging algorithm is adopted to achieve feature sharing and collaborative processing of raw data from multiple departments without leaving the domain. Disaster data collected by drones undergoes feature extraction at local edge nodes, and after encryption, only model feature parameters are shared, without transmitting raw sensitive data. After receiving the shared parameters, each node updates its local model, achieving full-node data synchronization. In this embodiment, the cross-departmental data communication delay is ≤3 seconds, far lower than the existing technology's delay of over 20 minutes, while simultaneously ensuring the data privacy and security of each department.

[0088] Access Control: A hierarchical access control mechanism is adopted, with the Emergency Management Bureau having the highest dispatch authority, and the public security, fire, and medical departments having the authority to view data and issue instructions for their respective business. All data operations are traceable and auditable.

[0089] The heterogeneous cluster scheduling unit includes a digital twin scenario construction subunit and a distributed conflict avoidance subunit. It adopts a master-slave collaborative mode and supports up to 100 heterogeneous drones to work together simultaneously with a flight path conflict rate of ≤0.1%.

[0090] Hardware and role definitions for master-slave collaborative mode:

[0091] Master node: 1 composite wing main UAV (backup node is ground edge collaborative emergency command platform), equipped with Jetson AGX Orin high-performance edge computing unit, full-band communication module, endurance ≥ 6h, responsible for global task allocation, cluster status monitoring, flight path conflict coordination, and emergency command issuance;

[0092] Slave nodes: Up to 100 heterogeneous slave drones, divided into fixed-wing drones: responsible for wide-area disaster patrol, with an endurance of ≥12h and a cruising speed of 80km / h; multi-rotor drones: responsible for detailed surveys of key areas and precise location of trapped personnel, with an endurance of ≥40min; and material delivery drones: responsible for emergency material delivery, with a payload of ≥5kg and an endurance of ≥30min.

[0093] Cluster management mechanism: For large-scale clusters of 50 or more drones, a cluster management mode is adopted. Every 10 drones are divided into 1 cluster, and a cluster head drone is set. Intra-cluster flight path conflicts are coordinated by the cluster head, and inter-cluster conflicts are coordinated by the master drone, which greatly reduces computing overhead and improves real-time scheduling.

[0094] Task allocation algorithm implementation:

[0095] An improved contract network protocol is used to achieve optimal task allocation for heterogeneous UAVs. The specific process is as follows:

[0096] The main drone issues an emergency mission to the cluster, including mission priority, operating area, and performance requirements;

[0097] The drone sends bidding information to the main drone based on its own model, remaining battery life, current location, and operational capabilities;

[0098] Based on the principle of optimal overall cost, the main drone selects the best bidder to allocate tasks, with a single round of task allocation time of ≤100ms, and supports more than 50 drones to bid simultaneously.

[0099] Digital Twin Scene Construction Subunit:

[0100] A 1:1 digital twin of an urban emergency scenario is built using Unity3D. The city's 3D building model, road network, water system, and key protection target information are imported. The system receives the location, speed, flight path, perception data, and operational status of drones in real time with a data refresh rate of 30Hz. This enables real-time visualization and mapping of the cluster status. Commanders can complete the entire process of task issuance, flight path planning, and status monitoring within the twin.

[0101] The distributed conflict avoidance algorithm (ensuring a conflict rate ≤ 0.1%) employs an improved velocity obstacle (VO) method combined with an artificial potential field method to achieve dynamic obstacle avoidance and flight path conflict avoidance. The core principles and formulas are as follows:

[0102] Speed ​​obstacle cone definition: For two drones A and B, the safe fuselage radii are respectively and relative speed Relative position vector Speed ​​barrier cone For all of this, it could lead to the two drones in the future. The set of relative velocities of collisions occurring within the space is given by the following formula:

[0103]

[0104] In this embodiment, the collision prediction time Set to 5 seconds, safety radius = =5m, with sufficient safety redundancy reserved.

[0105] Distributed coordination mechanism: Every 100ms, each drone broadcasts its position, speed, and planned flight path information via a LoRa ad hoc network; when it detects that its relative speed with other drones has entered the speed obstacle cone, the drone autonomously adjusts its heading and speed to avoid the conflict area; if multiple drones cross each other, the master drone issues a global coordination command to uniformly adjust the flight path and avoid deadlock.

[0106] Static obstacle avoidance: Combining the artificial potential field method, the target point exerts an attractive force on the drone, while static obstacles (buildings, mountains) exert a repulsive force on the drone. The combined force guides the drone to fly along a collision-free path. The formulas for attraction and repulsion are as follows:

[0107]

[0108]

[0109] in This is the gravitational gain coefficient. The repulsive force gain coefficient, The coordinates of the target point, The current coordinates of the drone. The shortest distance from the drone to the obstacle. The radius of influence of the obstacle.

[0110] The intelligent allocation unit for collaborative instructions achieves optimal matching of emergency resources based on task priorities. Task priorities are divided from high to low as follows: rescue of trapped personnel (priority 1), early warning of secondary disasters (priority 2), wide-area disaster patrol (priority 3), and emergency material delivery (priority 4). The edge collaborative emergency command platform automatically assigns task instructions to drones and emergency departments with corresponding capabilities according to task priorities. For example, when a signal of trapped personnel is detected, it automatically pushes the precise location to the fire department, assigns the nearest multi-rotor drone to conduct secondary confirmation, and pushes the rescue request to the medical department, realizing collaborative linkage among multiple departments and multiple devices.

[0111] The edge-collaborative emergency command platform serves as the core hub of the system. The hardware carrier is a high-performance server mounted on a command vehicle, equipped with a portable command terminal. The software adopts a B / S architecture and includes a data fusion module, a cluster scheduling module, a digital twin visualization module, an emergency plan management module, and an access control module. It supports up to 20 users logging in simultaneously, enabling unified control and emergency command decision-making for the entire system.

[0112] In practice, the specific process is as follows:

[0113] Emergency Activation and Communication Networking: After the emergency response is activated, the edge-coordinated emergency command platform controls one main UAV and three relay UAVs to take off and build an integrated air-ground communication network of Beidou short message + LoRa self-organizing network. In an environment without a public network, it achieves full-link communication coverage of the command platform, UAV cluster, and ground rescue terminal.

[0114] Wide-area disaster patrol: The main UAV assigns wide-area patrol tasks to the fixed-wing UAVs through the contract network protocol. The fixed-wing UAVs conduct full-coverage patrols of the disaster area along the planned route, collect disaster data through the multi-source fusion sensing module, and transmit it back to the command platform in real time to generate a disaster heat map.

[0115] Precise location of trapped personnel: When a signal of a trapped person is detected during patrol, the command platform automatically assigns the nearest multi-rotor drone to the target area. Through a triple positioning calibration unit, combined with the matching of lidar point cloud and urban 3D model, the floor and window location of the trapped person are accurately located. The data is synchronized to the fire and medical rescue departments in real time, with a cross-departmental synchronization delay of ≤3s.

[0116] Cluster collaborative operation: 50 heterogeneous drones simultaneously carry out patrol, positioning and material delivery operations in the disaster area. The main drone coordinates the flight paths of each drone in real time through cluster management and distributed conflict avoidance algorithm to avoid flight path intersection and collision. The flight path conflict rate is ≤0.1%.

[0117] Closed-loop command and decision-making: The edge-coordinated emergency command platform integrates all collected disaster data and rescue force data, combines them with emergency plans to generate the optimal rescue plan, and issues coordinated instructions to various rescue units and drone clusters to complete the entire closed-loop emergency response from perception, communication, coordination to decision-making.

[0118] The above embodiments only illustrate preferred embodiments of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of this application's patent. It should be noted that those skilled in the art can make various modifications, improvements, and substitutions without departing from the concept of this application, and these all fall within the protection scope of this application.

Claims

1. A drone patrol system, characterized in that, This includes an edge-coordinated emergency command platform, a multi-mode redundant self-organizing communication module, an anti-interference multi-source fusion sensing and positioning module, and a cross-domain heterogeneous cluster collaboration module; The multi-mode redundant self-organizing communication module, the anti-interference multi-source fusion sensing and positioning module, and the cross-domain heterogeneous cluster collaboration module are all bidirectionally connected to the edge collaborative emergency command platform. The multi-mode redundant self-organizing communication module is used to construct an air-ground integrated multi-hop self-organizing communication network without reliance on the terrestrial public network, providing the system with redundant data transmission channels under extreme disaster scenarios; The anti-interference multi-source fusion sensing and positioning module is used to collect scene and target data in complex and harsh environments to achieve accurate three-dimensional spatial positioning in GPS signal-blocked scenarios. The cross-domain heterogeneous cluster collaboration module is used to realize real-time data exchange among multiple emergency departments and collaborative scheduling of multiple drone clusters. The edge-collaborative emergency command platform is used to integrate and process the data uploaded by each module, issue collaborative control commands, and form a closed-loop emergency patrol system of perception, communication, collaboration, and decision-making.

2. A drone patrol system according to claim 1, characterized in that, The multi-mode redundant self-organizing communication module includes a Beidou short message communication unit, a LoRa self-organizing network communication unit, an UAV relay communication unit, and a dynamic spectrum adaptive unit. The Beidou short message communication unit is used for long-distance backup command and data transmission in extreme network outage scenarios; The LoRa self-organizing network communication unit is used for decentralized multi-hop communication between drones and between drones and ground emergency terminals, and supports automatic route reconstruction after network node failure. The UAV relay communication unit is used to dynamically deploy airborne relay nodes to fill signal coverage blind spots in high-rise buildings and disaster core areas. The dynamic spectrum adaptive unit is used to detect environmental electromagnetic interference in real time and automatically switch communication frequency bands and adjust transmission power.

3. A drone patrol system according to claim 2, characterized in that, The multi-mode redundant self-organizing communication module also integrates a lightweight blockchain consensus unit, which is used to perform distributed encrypted storage and integrity verification of transmitted emergency data to ensure reliable data exchange among multiple departments.

4. A drone patrol system according to claim 1, characterized in that, The anti-interference multi-source fusion sensing and positioning module includes a millimeter-wave radar sensor, an infrared thermal imaging sensor, a lidar sensor, an ultrasonic sensor, and a multi-sensor data fusion processing unit. The multi-sensor data fusion processing unit is used to perform spatiotemporal registration and fusion processing on multi-source data collected by four types of sensors, and output target recognition results under severe environments such as dense smoke, heavy rain, and nighttime.

5. A drone patrol system according to claim 4, characterized in that, The anti-interference multi-source fusion sensing and positioning module also includes a triple positioning calibration unit, which integrates a Beidou RTK positioning subunit, a visual inertial odometry (VIO) positioning subunit, and a city 3D model matching subunit. The Beidou RTK positioning subunit is used for centimeter-level absolute positioning in unobstructed environments; The visual inertial odometry (VIO) positioning subunit is used for relative positioning in environments where GPS signals are obscured. The city 3D model matching subunit is used to match the lidar scanning data with the pre-stored city 3D building model and output the 3D positioning result of the target's floor and spatial location.

6. A drone patrol system according to claim 4, characterized in that, The anti-interference multi-source fusion sensing and positioning module is also equipped with a harsh environment adaptive adjustment unit, which is used to dynamically adjust the working weight and operating parameters of each sensor according to real-time environmental parameters.

7. A drone patrol system according to claim 1, characterized in that, The cross-domain heterogeneous cluster collaboration module includes a cross-domain data interoperability unit, a heterogeneous cluster scheduling unit, and a collaborative instruction intelligent allocation unit. The cross-domain data interoperability unit is bidirectionally connected to the edge collaborative emergency command platform to achieve decentralized fusion of emergency data from multiple emergency departments and real-time synchronization across all nodes. The cross-domain data interoperability unit adopts a federated learning framework to achieve feature sharing and collaborative processing of raw data from multiple departments without leaving their respective domains.

8. A drone patrol system according to claim 7, characterized in that, The heterogeneous cluster scheduling unit includes a digital twin scenario construction subunit and a distributed conflict avoidance subunit; The digital twin scenario construction subunit is used to build a digital twin of the urban emergency scenario, which maps the drone's location, flight path, operating status and perception data in real time. The distributed conflict avoidance subunit is used to dynamically plan conflict-free routes based on the UAV's mission type and flight parameters. The heterogeneous cluster scheduling unit adopts a master-slave collaborative mode, supporting at least 50 heterogeneous drones to work collaboratively at the same time.

9. A drone patrol system according to claim 7, characterized in that, The multi-mode redundant self-organizing communication module provides an end-to-end low-latency transmission channel for the anti-interference multi-source fusion sensing and positioning module and the cross-domain heterogeneous cluster collaborative module.

10. A smart-enabled urban emergency equipment, characterized in that, Including the drone patrol system as described in any one of claims 1-9.

Images