A land-air coordinated fire extinguishing system and method for unmanned aerial vehicles and robot dogs

Abstract

This invention belongs to the field of intelligent firefighting equipment and multi-robot collaborative control technology, and relates to a land-air coordinated firefighting system and method using drones and robotic dogs. It includes an aerial reconnaissance and relay subsystem, a ground assault and operation subsystem, and a land-air coordinated command and control platform. The aerial reconnaissance and relay subsystem consists of a reconnaissance drone equipped with a multimodal perception module. The ground assault and operation subsystem consists of a quadrupedal firefighting robotic dog equipped with a replaceable mission payload to perform firefighting or material transport tasks. The land-air coordinated command and control platform generates a coordinated firefighting plan and establishes an aerial communication relay link through the reconnaissance drone to remotely guide and dynamically schedule the robotic dog. This invention deeply integrates aerial global perception and communication relay capabilities with ground high-mobility obstacle-crossing and payload operation capabilities, realizing a closed-loop operation of unmanned reconnaissance, path planning, close-range firefighting, and dynamic hazard avoidance at high-risk fire sites.

Description

Technical Field

[0001] This invention belongs to the field of intelligent fire-fighting equipment and multi-robot collaborative control technology, and in particular relates to a land-air coordinated fire-fighting system of drones and robot dogs. Background Technology

[0002] With the acceleration of urbanization, the risk of fire accidents in complex terrains such as high-rise buildings, underground spaces, chemical plants, and energy storage power stations is increasing. Fire scenes pose dangers such as high temperatures and dense smoke, toxic gases, structural collapse, and secondary explosions. Firefighters entering the fire scene for close-range reconnaissance and firefighting carries a great risk of personal injury or death.

[0003] In recent years, firefighting robot technology has developed rapidly, mainly divided into ground-based mobile firefighting robots and aerial firefighting drones. Ground-based firefighting robots mainly include bionic quadrupedal robot dogs, which have terrain adaptability and a strong advantage in the firefighting industry. Chinese invention patent CN202510508369 discloses an intelligent numerically controlled fire early warning and automatic fire extinguishing robot dog. This robot dog moves through a bionic motion module and is equipped with a multimodal perception module to perceive the surrounding environment and determine the fire type. Then, the fire extinguishing module sprays the corresponding type of extinguishing agent according to the fire type. Its obstacle-crossing ability can replace traditional wheeled or tracked firefighting equipment in complex environments, expanding the operating area of ​​firefighting equipment to a certain extent. However, due to the limitations of the installation height and field of view of its own sensors, the robot dog's perception of the overall fire situation is poor. In dense smoke and under the obstruction of walls, it cannot obtain the overall distribution and spread trend of the fire source. Ground communication is easily blocked by building structures and the fire environment. Once the communication link is lost, the robot dog is at risk of losing control and cannot independently complete the reconnaissance and fire extinguishing tasks in complex fire scenes.

[0004] In the field of firefighting drones, aerial drones possess the characteristics of a wide field of view and flexible maneuverability, making them suitable for fire reconnaissance and fire extinguishing agent delivery. Chinese invention patent CN121606859A proposes an intelligent firefighting method, system, equipment, and medium for firefighting drones. It establishes an environmental state vector by combining depth images with infrared images, and outputs continuous action commands to control the drone's flight using a temporal decision model and a strategy generation model, enabling the drone to autonomously avoid obstacles and extinguish fires within confined spaces. Chinese invention patent CN202511235352 proposes an unmanned intelligent firefighting drone system and method for energy storage stations. The drone carries a fire extinguishing device, and a control terminal plans the flight path and triggers the fire extinguishing. However, the drone platform is limited by its endurance and payload, making it unable to carry large-capacity fire extinguishing agents for continuous and powerful firefighting, unable to carry heavy rescue equipment, and unable to enter buildings to attack fires from within, requiring external spraying for extinguishing.

[0005] In addition, Chinese invention patent CN202510517259 discloses a comprehensive method for rapidly extinguishing building fires, which uses fire rescue vehicles and drones to work together. The method involves "firefighters carrying the main pipelines and water distribution devices on their backs or using robotic dogs to carry them" and "the cargo robotic dogs will play an extremely important role, which can greatly reduce the burden on firefighters carrying fire extinguishing equipment and can also replace firefighters in performing high-risk fire extinguishing tasks." It also proposes the idea of ​​robotic dogs replacing firefighters in transporting materials. However, it does not provide specific solutions on how drones and robotic dogs can coordinate between land and air, or how to solve the technical problems of communication and navigation guidance for robotic dogs in complex buildings. The overall coordination mechanism is missing.

[0006] Firefighting robot dogs are limited by sensor installation height and field of view, making it impossible to perceive the overall fire situation. Communication links are easily blocked by buildings, preventing them from independently completing the entire process from global path planning to interior firefighting in complex buildings. Firefighting drones are limited by endurance and payload, making it impossible to carry large-volume extinguishing agents for continuous and powerful firefighting, and unable to enter buildings to complete precise interior attacks. Existing land-air coordinated firefighting methods only provide preliminary ideas for robot dogs to replace humans in material transportation, without solving key technical problems such as autonomous communication relay between drones and robot dogs, dynamic path guidance, collaborative fire monitoring, and hazard avoidance.

[0007] In summary, current firefighting equipment technologies, including ground-based robotic dogs and aerial drones, are mostly used for individual operations, lacking systematic land-air coordination solutions. How to deeply integrate the advantages of drones in global aerial perception and communication relay with the high obstacle-crossing capabilities and ground-based load-bearing capacity of robotic dogs to achieve a closed-loop operation from fire reconnaissance, path planning, close-range firefighting, to safe evacuation is a technical challenge that urgently needs to be overcome in this field. Summary of the Invention

[0008] The purpose of this invention is to provide a land-air coordinated firefighting system for drones and robot dogs, which has a heterogeneous robot deep integration and collaborative architecture with the capabilities of global aerial perception, aerial communication relay, high ground mobility obstacle crossing and ground load operation. This solves the technical problem in the prior art where firefighting robot dogs and firefighting drones operate independently and lack a system-level land-air coordination mechanism, resulting in the limitations of a single device.

[0009] To achieve the above objectives, the technical solution adopted by this invention is as follows: This invention provides a land-air coordinated firefighting system using drones and robotic dogs, including an aerial reconnaissance and relay subsystem, a ground assault and operation subsystem, and a land-air coordinated command and control platform. The aerial reconnaissance and relay subsystem includes at least one reconnaissance drone, which is equipped with a multimodal perception module, a first communication module, and a first edge computing unit. The multimodal perception module is used to collect visible light images, infrared thermal imaging images, and point cloud data of the fire area to identify the location of the fire source, the fire spread trend, and geographical environment information. The first edge computing unit is used to construct a three-dimensional environmental map of the fire area based on the data collected by the multimodal perception module and identify safe approach paths and obstacle information.

[0010] The ground assault and operation subsystem includes at least one quadruped firefighting robot dog, which is equipped with a bionic motion module, a second communication module, a second edge computing unit, and a replaceable mission payload interface; the mission payload interface is used to mount firefighting equipment or material transport boxes.

[0011] The land-air coordinated command and control platform is connected to the air reconnaissance and relay subsystem and the ground assault and operation subsystem, respectively. It is used to generate a coordinated fire extinguishing plan based on the three-dimensional environmental map of the fire site, and to parse the coordinated fire extinguishing plan into air reconnaissance commands for reconnaissance drones and ground assault commands for quadrupedal firefighting robot dogs.

[0012] Among them, the reconnaissance drone flies to the designated airspace according to the aerial reconnaissance instructions and serves as a mobile communication relay node to establish a communication link between the land-air collaborative command and control platform and the quadrupedal firefighting robot dog.

[0013] Based on ground assault commands and the 3D environmental map and safe approach route of the fire scene received via communication link, the four-legged firefighting robot dog autonomously navigates and overcomes obstacles to reach the target firefighting operation point to perform firefighting or material delivery tasks.

[0014] Specifically, when the communication link between the reconnaissance drone and the land-air coordinated command and control platform is blocked or the signal quality is lower than a preset threshold, the second edge computing unit of the quadruped firefighting robot dog activates the autonomous emergency navigation mode. The autonomous emergency navigation mode is based on the local perception sensor carried by the quadruped firefighting robot dog itself and the pre-loaded three-dimensional fire scene environment map, and autonomously decides to move to the preset safe assembly point or execute the strategy of waiting for rescue in place.

[0015] Specifically, the collaborative firefighting plan generated by the land-air collaborative command and control platform includes a land-air airspace division strategy: defining the altitude layer for reconnaissance drone operations and the ground area for quadrupedal firefighting robot dog operations; according to the requirements of the firefighting mission, the land-air collaborative command and control platform dynamically adjusts the flight trajectory of the reconnaissance drone to avoid the impact of heat waves, water mist or smoke generated by the quadrupedal firefighting robot dog operations on the flight stability and sensor accuracy of the reconnaissance drone.

[0016] Specifically, the fire extinguishing equipment mounted on the mission payload interface of the quadrupedal firefighting robot dog includes a high-flow-rate spray device; after arriving at the target fire extinguishing site, the quadrupedal firefighting robot dog adjusts the spray angle of the high-flow-rate spray device through its onboard robotic arm or fixed gimbal to carry out a comprehensive spray at the base of the fire source; at the same time, the reconnaissance drone monitors the fire extinguishing effect in the air and transmits the fire source temperature change data back in real time.

[0017] Specifically, when the fire area expands and multiple quadruped firefighting robot dogs are needed to work together, the reconnaissance drone divides the communication time slots or communication frequency bands according to the position of each quadruped firefighting robot dog to achieve air-to-ground time division multiple access or frequency division multiple access communication and avoid co-channel interference.

[0018] Specifically, the multimodal perception module also includes a gas detection sensor; the reconnaissance drone identifies the concentration distribution of toxic and harmful gases in the fire scene through the gas detection sensor and marks high-risk toxic gas areas on the three-dimensional environmental map of the fire scene; the land-air collaborative command and control platform adjusts the path of the quadrupedal firefighting robot dog according to this information to make it avoid toxic gas accumulation areas.

[0019] This invention provides a firefighting method for a land-air coordinated firefighting system using drones and robotic dogs, comprising the following steps:

[0020] 1. The reconnaissance drone flies over the fire site and collects real-time images and three-dimensional spatial data of the fire site through the multimodal perception module, identifying the core area of ​​the fire source, the edge area of ​​fire spread and surrounding terrain obstacles; the first edge computing unit constructs and updates the three-dimensional environmental map of the fire site in real time, marking dangerous areas and passable areas;

[0021] Second, based on the three-dimensional environmental map of the fire scene, the land-air coordinated command and control platform plans a safe approach path for the quadrupedal firefighting robot dog from its initial position to the target work point and generates mission instructions; at the same time, the land-air coordinated command and control platform dynamically adjusts the hovering position of the reconnaissance drone according to the predetermined position of the quadrupedal firefighting robot dog to ensure the stability and coverage of the communication link.

[0022] 3. The quadrupedal firefighting robot dog receives the safe approach path and autonomously overcomes obstacles and traverses complex terrain using its bionic motion module; during the journey, the reconnaissance drone monitors changes in nearby obstacles and heat sources around the quadrupedal firefighting robot dog in real time and sends obstacle avoidance fine-tuning instructions or danger warnings to the quadrupedal firefighting robot dog through the communication link.

[0023] IV. During firefighting operations, if the reconnaissance drone detects a sudden change in the fire situation or a new high-risk hotspot, it will update the three-dimensional environmental map of the fire scene and synchronize it to the land-air collaborative command and control platform; the land-air collaborative command and control platform will recalculate the safe area and send the updated operating position or evacuation command to the quadrupedal firefighting robot dog.

[0024] Specifically, step three also includes: using image recognition algorithms to compare real-time images collected by the reconnaissance drone with first-view images transmitted back by the quadrupedal firefighting robot dog, and using visual feature matching to assist in positioning, correcting the cumulative position error of the quadrupedal firefighting robot dog in environments with no or weak satellite signals.

[0025] This also includes: after the firefighting operation is completed, the reconnaissance drone guides the quadrupedal firefighting robot dog to withdraw along a preset safe route to the designated equipment recovery point; at the same time, the reconnaissance drone conducts a second thermal imaging scan of the fire site ruins, and after confirming that there is no risk of reignition, it sends a mission termination command to the land-air coordinated command and control platform.

[0026] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0027] 1. This invention deeply integrates aerial global perception and communication relay capabilities with ground high mobility obstacle crossing and payload operation capabilities to achieve unmanned reconnaissance, path planning, close-range fire extinguishing and dynamic risk avoidance of high-risk fire sites.

[0028] 2. The present invention has a heterogeneous robot deep integration and collaborative architecture with the capabilities of global aerial perception, aerial communication relay, high ground mobility obstacle crossing and ground load operation. It solves the technical problem in the prior art that fire-fighting robot dogs and fire-fighting drones operate independently and lack a system-level land-air collaboration mechanism, resulting in the limitations of a single device.

[0029] 3. This invention constructs a reconnaissance drone as an aerial reconnaissance and communication relay node and a quadrupedal firefighting robot dog as a ground assault and operation node. It completes the entire unmanned closed-loop operation from the construction of a three-dimensional map of the fire scene, safe path planning, autonomous obstacle crossing and approach by the robot dog, execution of firefighting operations, to dynamic fire monitoring and safe evacuation. This reduces the risk of personal injury to firefighters entering high-risk fire scenes and improves the success rate and efficiency of firefighting in complex fire scene environments. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 This is a schematic diagram of a land-air coordinated firefighting system combining a drone and a robot dog, as an example.

[0032] Figure 2 This is a schematic flowchart illustrating a fire extinguishing method using a land-air coordinated fire extinguishing system combining drones and robotic dogs, as an example of one embodiment.

[0033] In the above diagram, 1. Aerial reconnaissance and relay subsystem; 2. Ground assault and operation subsystem; 11. Reconnaissance UAV; 12. Multimodal perception module; 13. First communication module; 14. First edge computing unit; 21. Quadrupedal firefighting robot dog; 22. Bionic motion module; 23. Second communication module; 24. Second edge computing unit; 25. Mission payload interface; 26. Firefighting equipment. Detailed Implementation

[0034] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0035] Numerous specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways than those described herein, and therefore the invention is not limited to the specific embodiments disclosed in the following specification.

[0036] Example 1: This example provides a land-air coordinated firefighting system combining drones and robot dogs. It constructs a heterogeneous robot deep integration and collaborative architecture, organically integrating the advantages of the aerial platform's all-domain perception and communication relay with the advantages of the ground platform's high obstacle-crossing and load-bearing capabilities, to achieve unmanned closed-loop operation in high-risk fire scenes.

[0037] like Figure 1 As shown, the system comprises three main components: the air reconnaissance and relay subsystem 1, the ground assault and operations subsystem 2, and the air-ground coordinated command and control platform. The air-ground coordinated command and control platform, not shown separately in the diagram, can be deployed in a ground command vehicle, a remote command center, or a cloud server. The air-ground coordinated command and control platform establishes wireless communication connections with both the air reconnaissance and relay subsystem 1 and the ground assault and operations subsystem 2, serving as the central hub for overall situational awareness gathering, coordinated mission planning, and dynamic command scheduling.

[0038] The aerial reconnaissance and relay subsystem 1 includes at least one reconnaissance drone 11. Depending on the size of the fire and the complexity of the buildings, it can be flexibly configured as a single drone or in a multi-drone formation. The reconnaissance drone 11 is preferably a multi-rotor drone with vertical take-off and landing, hovering, and agile maneuverability, enabling it to quickly reach the airspace above the fire and establish a stable aerial operational base in the designated airspace.

[0039] The reconnaissance drone 11 is mainly equipped with a multimodal perception module 12, a first communication module 13, and a first edge computing unit 14. The multimodal perception module 12 is used to comprehensively collect multi-dimensional environmental information of the fire area. Specifically, the multimodal perception module 12 integrates at least the following sensors: a high-definition visible light camera, an infrared thermal imager, and a lidar or binocular stereo vision camera. The high-definition visible light camera is used to acquire visible light images of the fire area and identify the location of open flames, building structural features, surrounding terrain, and road information. The infrared thermal imager is used to penetrate smoke to acquire thermal distribution images of the fire area and accurately locate the core area of ​​the fire source, the edge area of ​​fire spread, and potential reignition hazards through temperature differences. The lidar or binocular stereo vision camera is used to collect three-dimensional point cloud data of the fire area and obtain information on building outlines, obstacle distances, and terrain elevation, providing spatial geometric data for constructing a high-precision three-dimensional environmental map of the fire area.

[0040] In a further optimization scheme, the multimodal sensing module 12 can also integrate a gas detection sensor to identify the concentration distribution of toxic and harmful gases such as carbon monoxide, hydrogen sulfide, and volatile organic compounds in the fire environment, providing data support for delineating high-risk toxic gas areas.

[0041] The first edge computing unit 14 is a localized intelligent processing core deployed on the reconnaissance drone 11. This unit consists of a high-performance embedded AI processor, such as the NVIDIA Jetson series or an equivalent computing platform, capable of performing on-orbit real-time processing of raw data collected by the multimodal perception module 12 without relying on ground communication links. First, it executes multi-sensor data fusion and synchronous localization and mapping (SLAM) algorithms to build and continuously update a three-dimensional fire scene environment map with geographic coordinate information in real time. Second, it runs a deep learning-based fire source identification and fire spread trend prediction model to automatically identify and mark the core area of ​​the fire source, the spread edge area, and the direction of smoke diffusion. Third, combining obstacle information and terrain constraints in the three-dimensional map, it uses path search algorithms, such as the improved A* algorithm or the fast random exploration tree algorithm, to initially identify safe approach paths and feasible areas that can be used by the ground-based quadrupedal firefighting robot dog 21. Fourth, it transmits the processed structured data back through the first communication module 13, reducing communication bandwidth requirements and enabling efficient transmission of critical information even in fire scene environments with limited communication conditions.

[0042] The first communication module 13 possesses multi-mode communication capabilities, including at least a 4G / 5G module or dedicated image and data transmission radio for long-distance communication with the air-to-ground coordinated command and control platform, and a Wi-Fi 6 or proprietary protocol broadband radio for establishing a short-range high-speed relay link with the ground-based quadruped firefighting robot dog 21. This allows the reconnaissance drone 11 to transform into a mobile aerial communication relay node when hovering in a specific airspace, effectively bypassing obstructions such as buildings. It establishes a stable and reliable communication bridge between the air-to-ground coordinated command and control platform and the quadruped firefighting robot dog 21 operating deep inside buildings or in low-lying areas, solving the inherent problem of ground communication being easily blocked.

[0043] The ground assault and operation subsystem 2 includes at least one quadruped firefighting robot dog 21. The quadruped firefighting robot dog 21 has a biomimetic leg-like locomotion structure that gives it a terrain adaptability far exceeding that of traditional wheeled or tracked robots. It can climb stairs, cross ditches, and walk stably on the surface of complex ruins piled with rubble and debris, and truly has the ability to penetrate deep into the fire scene inside the building.

[0044] The quadrupedal firefighting robot dog 21 is equipped with a bionic motion module 22, a second communication module 23, a second edge computing unit 24, and a replaceable task payload interface 25.

[0045] The bionic motion module 22 uses high-torque-density motors or hydraulically driven joints, with each robotic leg having at least three degrees of freedom, supporting free switching between various gait modes such as walking, jogging, climbing, and jumping. Its integrated inertial measurement unit and foot force sensor constitute a posture monitoring and balance control system, which can sense the body posture and foot contact status in real time, and autonomously adjust gait parameters when traversing rugged terrain to ensure motion stability.

[0046] The mission payload interface 25 enables the quadrupedal fire-fighting robot dog 21 to achieve diverse functions. It adopts a standardized quick-release mechanical connection and electrical communication protocol, supporting plug-and-play quick replacement of different types of mission payloads. In this embodiment, the preferred mission payload is a fire extinguishing device 26. The fire extinguishing device 26 can specifically be a high-flow-rate atomizing spray device, integrating a high-pressure fine atomizing nozzle and a fire extinguishing agent storage tank. It can atomize water or water-based foam fire extinguishing agent into tiny droplets, forming a high-flow-rate, highly penetrating water mist jet that instantly covers the entire space of the fire source room. It achieves efficient cooling and fire extinguishing by rapidly vaporizing and absorbing heat and isolating oxygen.

[0047] In another application scenario, the mission payload interface 25 can be equipped with a material transport box to deliver emergency supplies such as breathing masks and first aid kits to people trapped in the fire, or to send heavy demolition tools to firefighters who are breaking into the fire.

[0048] The second edge computing unit 24 is the localized decision-making core of the quadrupedal firefighting robot dog 21. In normal collaborative mode, it is responsible for receiving motion commands, safe approach paths, and updated data of the three-dimensional environment map of the fire scene sent through the communication relay link of the reconnaissance drone 11, and converting them into low-level motion control commands for the bionic motion module 22 to achieve autonomous navigation and obstacle crossing.

[0049] The second edge computing unit 24 has an internally embedded autonomous emergency navigation program. When the second communication module 23 detects that the uplink communication link signal quality with the reconnaissance drone 11 or the land-air coordinated command and control platform is continuously lower than a preset threshold, such as a continuous packet loss rate exceeding 50% or a clear "communication interruption" alarm flag being received, the autonomous emergency navigation mode is immediately activated. In this mode, the second edge computing unit 24 stops waiting for remote commands and instead relies entirely on the local perception sensors (not shown in the figure) carried by the quadrupedal firefighting robot dog 21 itself—including a front-facing binocular stereo vision camera, an ultrasonic perimeter obstacle avoidance radar, and a three-dimensional fire scene environment map pre-loaded offline into local storage—to independently plan an evacuation route toward a preset "safe assembly point" or execute a "wait for rescue in place" strategy, maintaining a silent communication state in place and periodically attempting to re-establish the connection until communication is restored or a new rescue command is received, eliminating the risk of the quadrupedal firefighting robot dog 21 losing control and connection due to unexpected interruption of the communication link in the fire scene environment.

[0050] The land-air coordinated command and control platform is the command center, running coordinated mission planning software and featuring a visual situation display interface. It can simultaneously access the 3D environmental map of the fire scene, fire situation assessment, and position status information of each platform transmitted back by the aerial reconnaissance and relay subsystem 1, as well as the self-check status and mission execution progress transmitted back by the ground assault and operation subsystem 2.

[0051] After the operator points to the location and preliminary extent of the fire on the interface, the land-air collaborative command and control platform algorithm automatically combines the three-dimensional environmental map of the fire scene, the distribution of obstacles, the passable area and the current position of each robot platform to generate a complete fire extinguishing plan with one click. This plan includes the hovering relay position of the reconnaissance drone 11, the task allocation of multiple quadrupedal firefighting robot dogs 21 and their respective safe approach paths, and the collaborative time window planning, which is then confirmed and executed by the commander.

[0052] The land-air collaborative command and control platform automatically defines the altitude layer for the reconnaissance UAV 11 and the ground area for the quadrupedal firefighting robot dog 21. During missions, when the quadrupedal firefighting robot dog 21 performs high-volume water mist spraying operations, it generates upward heat waves, water mist, and smoke, which may cause airflow disturbances, sensor obstruction, or even thermal damage to the reconnaissance UAV 11 located above it. Therefore, the land-air collaborative command and control platform can dynamically adjust the flight trajectory of the reconnaissance UAV 11 based on its current operational status and location, allowing it to temporarily fly around to an unaffected safe airspace and return to its original relay position during ground operation breaks, thus achieving safe coordination between the land and air platforms in three-dimensional space.

[0053] When the fire area expands and multiple quadrupedal firefighting robot dogs 21 need to work together simultaneously, the communication links of each quadrupedal firefighting robot dog 21 may experience co-channel interference in adjacent frequency bands. The platform or a reconnaissance drone 11 authorized by the platform can automatically allocate different communication time slots (time division multiple access) or assign different center frequencies (frequency division multiple access) to each quadrupedal firefighting robot dog 21 based on their current spatial distribution location. The multi-machine command and data transmission is non-interfering, stable and orderly.

[0054] The above-mentioned land-air coordinated firefighting system proposed in this invention moves the reconnaissance and communication relay functions to the airborne UAV platform, releases the maneuverability and load-bearing capabilities to the ground quadrupedal firefighting robot dog 21 platform, and realizes the information loop and task linkage among the three through a unified command and control platform, thus constructing an unmanned firefighting system with significant synergistic advantages and strong robustness.

[0055] Example 2: This example provides a fire extinguishing method for a land-air coordinated fire extinguishing system using drones and robotic dogs, based on the land-air coordinated fire extinguishing system using drones and robotic dogs proposed in Example 1 above. The following is a detailed explanation... Figure 2 The paper takes a typical collaborative firefighting scenario involving a complex structural fire as an example to explain each step in detail.

[0056] Step S1: The drone takes off and the multimodal perception module collects fire data.

[0057] After the fire broke out, the reconnaissance drone 11, as an advance platform, took off autonomously from a pre-deployed point, such as a fire truck-mounted take-off and landing platform or a ground launch point in a safe area. Based on the approximate coordinates of the fire point issued by the land-air collaborative command and control platform, it automatically planned an air route and flew to the designated airspace above the fire site.

[0058] Upon arrival in the mission airspace, the multimodal perception module 12, carried by the reconnaissance UAV 11, was fully activated, simultaneously collecting multi-dimensional information about the fire scene: a high-definition visible light camera captured visible light images at a high frame rate, clearly recording the floors, window positions, facade structure, and surrounding road terrain of the burning building; an infrared thermal imager penetrated the initial smoke obscuration to obtain images of the fire's thermal distribution, with a temperature measurement range covering -20℃ to 1000℃, clearly identifying the core area of ​​the fire source and the edge area of ​​heat spread; and a lidar, with a 360° horizontal field of view, rotated and scanned at high speed, acquiring hundreds of thousands of three-dimensional spatial measurement points per second, generating high-precision point cloud data of the burning building and its surrounding environment in real time, providing accurate spatial geometric information for subsequent 3D modeling.

[0059] Meanwhile, in a further optimized solution, the multimodal sensing module 12 also integrates a gas detection sensor to simultaneously detect the concentration of toxic and harmful gases such as carbon monoxide, hydrogen sulfide, and volatile organic compounds in the fire environment, providing data support for identifying high-risk toxic gas areas.

[0060] Step S2: The edge computing unit of the drone constructs and updates a three-dimensional environmental map of the fire scene in real time, and identifies the fire source and obstacles.

[0061] The first edge computing unit 14 of the reconnaissance UAV 11 performs on-orbit real-time fusion processing on the multi-channel heterogeneous sensor data collected by the multimodal perception module 12, without the need to transmit the original data stream back to the ground.

[0062] Specifically, the first edge computing unit 14 uses the SLAM algorithm to stitch together the point cloud data collected by the lidar into a dense three-dimensional geometric model. Simultaneously, it projects the temperature distribution features extracted from the infrared thermal imaging image onto the corresponding surface of the three-dimensional geometric model using a spatial mapping algorithm, generating a three-dimensional environmental map of the fire scene with real-time temperature annotations. On this map, the system automatically marks the status of each area with different colors based on preset temperature thresholds: red indicates the core area of ​​the fire source (temperature exceeding 300℃), orange indicates a high-risk area for fire spread (temperature between 80℃ and 300℃), green indicates a currently safe and passable area, and gray indicates structural obstacles such as collapsed walls.

[0063] Meanwhile, the first edge computing unit 14 runs a deep learning-based fire source recognition model and obstacle detection model, automatically identifying and marking key information such as the location of open flames, the direction of smoke diffusion, accessible openings such as doors and windows, the topology of stairs and corridors, and collapsed debris on the map, laying a data foundation for subsequent safe path planning.

[0064] Step S3: The land-air coordinated command and control platform determines whether ground intervention by the robot dog is required.

[0065] The first edge computing unit 14 transmits the completed 3D fire scene environment map and fire assessment data back to the land-air collaborative command and control platform via the first communication module 13. The land-air collaborative command and control platform provides a panoramic view of the fire scene on a visualized 3D GIS interface.

[0066] Command personnel or the automatic decision-making module of the land-air coordinated command and control platform make judgments based on the following criteria: whether there are indoor fire points accessible via ground paths; whether the fire has exceeded the fire extinguishing capability of the drone's external spraying, requiring close-range attack; whether there are trapped personnel requiring material delivery; and whether the building structure is stable enough for the robot dog to safely enter. If any of the above conditions are met, it is determined that "ground intervention by the robot dog is required," and the process proceeds to step S4; if none of the above conditions are met, for example, if the fire is small and can be extinguished directly by drone spraying, or if the building is showing signs of overall collapse and cannot be entered, it is determined that "no," and the process returns to step S1, with the reconnaissance drone 11 continuing its aerial reconnaissance and monitoring mission. This judgment loop ensures that the robot dog is only dispatched to high-risk areas when necessary and conditions permit.

[0067] Step S4: Plan the robot dog's safe approach path and generate a land-air collaborative mission plan.

[0068] Once it is determined that ground intervention by the robot dog is required, the mission planning algorithm engine of the land-air coordinated command and control platform is immediately activated.

[0069] After the commander marks the "initial deployment point" (the robot dog's starting point) and the "target firefighting operation point" (the specific location where firefighting or material delivery needs to be carried out) on the 3D map interface, the land-air coordinated command and control platform comprehensively considers the distribution of obstacles marked on the 3D map, warnings of high-temperature danger zones, boundaries of toxic gas areas, the location of stairs and narrow passages, as well as the robot dog's own kinematic constraints such as maximum climbing angle, minimum turning radius, and obstacle crossing height. Using an optimized path search algorithm, such as the improved A* algorithm, it automatically plans a 3D "safe approach path" from the initial deployment point to the target operation point within a few seconds to more than ten seconds. This path will actively avoid all danger zones marked in red and yellow and gray obstacles, and the robot dog will travel in a relatively safe environment throughout the process.

[0070] Meanwhile, the land-air coordinated command and control platform analyzed the communication and line-of-sight conditions along the entire path and calculated one or more optimal hovering positions that would ensure a continuous wireless communication link between the reconnaissance drone 11 and the quadrupedal firefighting robot dog 21, which was about to enter the building. This hovering position was designated as the "drone relay docking point".

[0071] Step S5: Plan the drone relay hovering position.

[0072] The land-air coordinated command and control platform packages all of the above into a complete "land-air coordinated mission plan" and automatically parses it into two types of specific instructions: "air reconnaissance instructions" sent to reconnaissance drone 11, including relay hovering position coordinates, designated monitoring area and altitude layer, etc., and "ground assault instructions" sent to quadrupedal firefighting robot dog 21, including a complete safe approach path trajectory point sequence, offline data package of fire scene three-dimensional environment map, precise coordinates of target operation point and mission payload operation parameters, etc.

[0073] Step S6: The UAV establishes an airborne communication relay link and sends map and route instructions to the robot dog.

[0074] After receiving the "aerial reconnaissance command," the reconnaissance drone 11 autonomously flies to the designated relay station and hovers stably. Its first communication module 13 activates relay mode, establishing and maintaining an end-to-end bidirectional communication link between the land-air collaborative command and control platform and the quadrupedal firefighting robot dog 21. This link allows the robot dog to maintain communication indirectly with the land-air collaborative command and control platform through the drone relay node hovering in the air, even when it is deep inside a building and blocked by multiple layers of reinforced concrete slabs, effectively solving the problem of ground communication being easily blocked by buildings.

[0075] After the link is established, the reconnaissance drone 11 forwards the "ground assault command" issued by the land-air coordinated command and control platform—including the offline data packet of the three-dimensional environmental map of the fire site and the safe approach path trajectory—to the quadrupedal firefighting robot dog 21. After the robot dog's second communication module 23 completes the data reception, it sends a confirmation signal back to the land-air coordinated command and control platform, indicating that the command has been received and is ready to be executed.

[0076] Step S7: The robot dog autonomously navigates over obstacles, while the drone provides real-time overhead monitoring and obstacle avoidance guidance.

[0077] After confirming the mission command, the quadrupedal firefighting robot dog 21 immediately initiates the autonomous approach procedure. Its bionic motion module 22 adaptively switches gait according to the terrain: on flat surfaces, it maintains a speed of about 1.2 m / s with a "diagonal trot" gait for efficient movement; on stairs or steep slopes, it automatically switches to a "ladder climbing" gait, climbing cautiously at a lower speed, while dynamically adjusting the pushing torque and landing point of each leg based on real-time feedback from foot force sensors and inertial measurement units to ensure stable passage.

[0078] Throughout the journey, the reconnaissance drone 11 maintains an overhead view above the robot dog, continuously performing its "aerial escort" function. The drone's high-altitude view is unobstructed by ground obstacles, allowing for a clear observation of the overall environment within a radius of tens of meters around the robot dog. The first edge computing unit 14 runs a lightweight real-time target detection model, specifically designed to identify high-risk obstacles such as small protrusions, loose covers, and fallen objects close in front of the robot dog—details that the robot dog's own first-person perspective sensors may struggle to detect due to their narrow field of view. Once a threat is detected, the reconnaissance drone 11 immediately sends "obstacle avoidance fine-tuning commands" to the robot dog via a low-latency relay link, such as "shift 0.3 meters to the left" or "jump over the obstacle ahead," and the robot dog responds with an evasive maneuver within milliseconds.

[0079] Furthermore, a visual feature matching-assisted positioning mechanism can be introduced in this step to address the problem of complete loss of satellite positioning signals inside buildings. The high-altitude view image from the reconnaissance drone 11 and the first-view image transmitted back by the quadrupedal firefighting robot dog 21 are compared in real time using image recognition algorithms—for example, the same door or window with unique features that can be identified in both camera images—and by calculating the known absolute coordinates of this common feature in the 3D map, the precise position of the robot dog is calculated in reverse, effectively correcting the inertial navigation position drift error accumulated over long-term operation.

[0080] Step S8: The robot dog arrives at the work site to perform its task and dynamically monitors whether the fire situation changes abruptly.

[0081] After the quadrupedal fire-fighting robot dog 21 successfully arrives at the target fire-fighting site according to the planned path, it performs the preset fire-fighting or material delivery tasks. When equipped with fire-fighting equipment 26, the robot dog uses a high-flow-rate, comprehensive spray to extinguish the fire at its base.

[0082] While performing its mission, the reconnaissance drone 11 continuously monitors the operational area and its surrounding environment from the air. An infrared thermal imager records the temperature change curve of the fire source area in real time, and a gas sensor continuously detects the concentration of toxic and harmful gases in the environment. The first edge computing unit 14 continuously runs a fire anomaly detection algorithm to determine if any of the following "fire abrupt change" events have occurred: a sudden and rapid temperature rebound in an area that had previously cooled down; the sudden appearance of new high-temperature hotspots exceeding a threshold in an adjacent, unexpected area; a sharp increase in smoke or toxic gas concentration; or signs of structural instability in buildings.

[0083] If the determination is "yes", meaning the fire has undergone a significant change, the process proceeds to step S9; if the determination is "no", meaning the fire has stabilized and continues to recede under firefighting operations, the process proceeds directly to step S10.

[0084] Step S9: The drone updates the 3D environmental map of the fire site and marks new danger zones, synchronizing it to the land-air joint command and control platform to replan the route or issue evacuation orders.

[0085] When the reconnaissance drone 11 detects a sudden change in the fire situation, the first edge computing unit 14 immediately updates the newly emerging high-risk hotspots or dangerous areas to the current three-dimensional environmental map of the fire scene, and marks them as "new danger zones" in a conspicuous color on the map. The updated map data is synchronized to the land-air collaborative command and control platform in real time, and an emergency alarm is triggered at the same time.

[0086] The dynamic replanning engine of the land-air coordinated command and control platform was immediately activated, comprehensively assessing the expansion speed and direction of the new danger zone, as well as the safety of the current location of the robotic dog. The land-air coordinated command and control platform quickly made a decision: if the current work point was still within the safe zone, it might only need to fine-tune the robotic dog's position or modify the subsequent mission path; if it determined that the current work point was about to be covered by the danger zone, it automatically generated an "emergency evacuation command" and a real-time calculated temporary evacuation path, which was sent to the quadrupedal firefighting robotic dog 21 with the highest priority, instructing it to move rapidly along the wall in a safe direction to the backup safe assembly point. At the same time, it sent instructions to the reconnaissance drone 11 to adjust its monitoring focus and flight path to the airspace above the new high-risk area.

[0087] After the path or plan has been replanned and adjusted, the process returns to step S4, and the complete collaborative task planning process is restarted with the new fire situation and robot position as the starting point, forming a dynamic closed loop of "perception-decision-adjustment-execution".

[0088] Step S10: Task completed, check to confirm there is no risk of reignition.

[0089] Once all predetermined firefighting targets have been achieved, and the reconnaissance drone 11 has completed at least two rounds of full secondary thermal imaging scans of the fire area, confirming that no abnormal hotspots with temperatures exceeding the preset safety threshold (such as 80°C) have been detected in the core firefighting zone and the edge of the spread zone for a continuous period of time, and that there are no new signs of smoke or toxic gas leaks, the land-air coordinated command and control platform determines that "the firefighting mission is successful and there is no risk of reignition."

[0090] Subsequently, the reconnaissance drone 11 planned the safe evacuation route for each of the four-legged firefighting robot dogs 21 to return to the preset "equipment recovery point" and issued the routes to the corresponding robot dogs. The reconnaissance drone 11 remained flying behind and above the robot dogs, monitoring the safe evacuation of all robot dogs from the danger zone and the completion of their assembly and recovery. This marked the end of a complete unmanned firefighting collaborative combat mission.

[0091] The process has entered the end state.

[0092] Comparative Example

[0093] Compared to the two solutions of "independent firefighting by drones" and "deployment of robot dogs by drones", this invention adopts a land-air collaborative architecture of "drone aerial relay guidance and robot dog autonomous ground obstacle crossing and approach", and discusses the intuitive advantages of the solution of this invention.

[0094] 1. Due to the limited energy density of lithium batteries, multi-rotor drones typically have a flight time of less than 20 minutes after being fully loaded with fire extinguishing agent, which is insufficient to support continuous firefighting operations. When the drone hovers, the strong downwash generated by the rotors can disperse flames, interfere with the trajectory of the fire extinguishing agent, and even supplement the fire with oxygen. The high-temperature hot airflow above the fire source can reach hundreds of degrees Celsius, causing severe turbulence to the drone, which can lead to flight control failure or thermal damage to electronic components in severe cases.

[0095] Drones cannot penetrate narrow passages such as doors, windows, and stairwells to enter the interior of buildings to extinguish fires; they can only perform external spraying and cannot carry out internal firefighting.

[0096] Therefore, the appropriate role of drones is "aerial reconnaissance and communication relay platform", not the main force in independent firefighting operations.

[0097] 2. A medium-sized firefighting robot dog with combat capabilities weighs approximately 60-100 kg. Deploying the robot dog to a building rooftop or floor via drone places extremely high demands on the drone's payload capacity. Furthermore, the robot dog is in a passive pendulum state during transport, making it prone to significant oscillations under wind and rotor airflow, threatening the drone's flight safety.

[0098] The rooftops of buildings on fire are often obstructed by air conditioning units, ventilation ducts, and other obstacles. Deployment points on windows or balconies are narrow and often obscured by smoke. Precisely deploying a 100-kilogram swing load to an area of ​​only a few square meters under the influence of hot air currents exceeds the reliable and controllable range of current civilian drones. During approach to the building, the hoisting cable may become entangled with external cables, billboards, security grilles, or other attachments, causing the drone to lose control or the payload to fall.

[0099] The hoisting and deployment is a one-way, irreversible operation; once deployed, the robot dog cannot return autonomously in the event of mission failure or a sudden change in the fire situation. When entering via ground approach, the robot dog always retains the ability to autonomously evacuate along the original route.

[0100] Compared to the two solutions mentioned above, the robot dog of this invention moves autonomously throughout its journey, eliminating the risk of being suspended in mid-air and not relying on deployment accuracy. In the event of a communication interruption, the robot dog can autonomously navigate to a safe rendezvous point, and the equipment is recoverable. During the robot dog's approach and movement, the drone simultaneously performs relay communication and panoramic monitoring, without the two competing for platform resources. It is not limited by building height, shape, or roof conditions, making it suitable for various complex fire situations.

[0101] In summary, the solution proposed in this invention is superior to the above alternatives in terms of engineering feasibility, mission reliability, and operational adaptability. It is a reliable technical route for achieving unmanned closed-loop operations in high-risk fire sites under current technological conditions.

[0102] All standard parts used in this invention can be purchased from the market, and irregular parts can be customized according to the description and drawings. The specific connection methods of each part adopt conventional methods such as bolts, rivets, and welding that are mature in the prior art. The machinery, parts and equipment adopt conventional models in the prior art. In addition, the circuit connection adopts conventional connection methods in the prior art, which will not be described in detail here. The contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0103] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments for application in other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A land-air coordinated firefighting system combining unmanned aerial vehicles (UAVs) and robotic dogs, comprising an aerial reconnaissance and relay subsystem, a ground assault and operation subsystem, and a land-air coordinated command and control platform, characterized in that, The aerial reconnaissance and relay subsystem includes at least one reconnaissance drone, which is equipped with a multimodal perception module, a first communication module and a first edge computing unit. The multimodal perception module is used to collect visible light images, infrared thermal imaging images and point cloud data of the fire area to identify the location of the fire source, the trend of fire spread and geographical environment information. The first edge computing unit is used to construct a three-dimensional environmental map of the fire scene based on the data collected by the multimodal perception module, and to identify safe approach paths and obstacle information; The ground assault and operation subsystem includes at least one quadruped firefighting robot dog, which is equipped with a bionic motion module, a second communication module, a second edge computing unit, and a replaceable mission payload interface; the mission payload interface is used to mount firefighting equipment or material transport boxes. The land-air coordinated command and control platform is connected to the air reconnaissance and relay subsystem and the ground assault and operation subsystem, respectively. It is used to generate a coordinated fire extinguishing plan based on the three-dimensional environmental map of the fire site, and to parse the coordinated fire extinguishing plan into air reconnaissance commands for reconnaissance drones and ground assault commands for quadrupedal firefighting robot dogs. Among them, the reconnaissance drone flies to the designated airspace according to the aerial reconnaissance instructions and serves as a mobile communication relay node to establish a communication link between the land-air collaborative command and control platform and the quadrupedal firefighting robot dog. Based on ground assault commands and the 3D environmental map and safe approach route of the fire scene received via communication link, the four-legged firefighting robot dog autonomously navigates and overcomes obstacles to reach the target firefighting operation point to perform firefighting or material delivery tasks.

2. The land-air coordinated firefighting system of drone and robot dog according to claim 1, characterized in that, When the communication link between the reconnaissance drone and the land-air coordinated command and control platform is blocked or the signal quality is lower than a preset threshold, the second edge computing unit of the quadruped firefighting robot dog activates the autonomous emergency navigation mode. The autonomous emergency navigation mode is based on the local perception sensor carried by the quadruped firefighting robot dog and the pre-loaded three-dimensional fire scene environment map, and autonomously decides to move to the preset safe assembly point or execute the strategy of waiting for rescue in place.

3. The land-air coordinated firefighting system of drone and robot dog according to claim 1, characterized in that, The collaborative firefighting plan generated by the land-air collaborative command and control platform includes a land-air airspace division strategy: defining the altitude layer for reconnaissance drone operations and the ground area for quadrupedal firefighting robot dog operations; according to the requirements of the firefighting mission, the land-air collaborative command and control platform dynamically adjusts the flight trajectory of the reconnaissance drone to avoid the impact of heat waves, water mist or smoke generated by the quadrupedal firefighting robot dog operations on the flight stability and sensor accuracy of the reconnaissance drone.

4. A land-air coordinated firefighting system combining a drone and a robot dog according to claim 1, characterized in that, The fire extinguishing equipment mounted on the mission payload interface of the quadrupedal firefighting robot dog includes a high-flow-rate spray device. After arriving at the target fire extinguishing site, the quadrupedal firefighting robot dog adjusts the spray angle of the high-flow-rate spray device through its onboard robotic arm or fixed gimbal to carry out a comprehensive spray at the base of the fire source. At the same time, the reconnaissance drone monitors the fire extinguishing effect from the air and transmits the fire source temperature change data back in real time.

5. A land-air coordinated firefighting system combining a drone and a robot dog according to claim 1, characterized in that, When the fire area expands and multiple quadruped firefighting robot dogs are needed to work together, the reconnaissance drone divides the communication time slots or communication frequency bands according to the position of each quadruped firefighting robot dog to achieve air-to-ground time division multiple access or frequency division multiple access communication and avoid co-channel interference.

6. A land-air coordinated firefighting system combining a drone and a robot dog according to claim 1, characterized in that, The multimodal perception module also includes a gas detection sensor; the reconnaissance drone identifies the concentration distribution of toxic and harmful gases in the fire scene through the gas detection sensor and marks high-risk toxic gas areas on the three-dimensional environmental map of the fire scene; the land-air collaborative command and control platform adjusts the path of the quadrupedal firefighting robot dog according to the information so that it avoids toxic gas accumulation areas.

7. A fire extinguishing method for a land-air coordinated fire extinguishing system combining drones and robotic dogs according to any one of claims 1-6, characterized in that, Includes the following steps, 1. The reconnaissance drone flies over the fire site and collects real-time images and three-dimensional spatial data of the fire site through the multimodal perception module, identifying the core area of ​​the fire source, the edge area of ​​fire spread and surrounding terrain obstacles; the first edge computing unit constructs and updates the three-dimensional environmental map of the fire site in real time, marking dangerous areas and passable areas; Second, based on the three-dimensional environmental map of the fire scene, the land-air coordinated command and control platform plans a safe approach path for the quadrupedal firefighting robot dog from its initial position to the target work point and generates mission instructions; at the same time, the land-air coordinated command and control platform dynamically adjusts the hovering position of the reconnaissance drone according to the predetermined position of the quadrupedal firefighting robot dog to ensure the stability and coverage of the communication link.

3. The quadrupedal firefighting robot dog receives the safe approach path and autonomously overcomes obstacles and traverses complex terrain using its bionic motion module; during the journey, the reconnaissance drone monitors changes in nearby obstacles and heat sources around the quadrupedal firefighting robot dog in real time and sends obstacle avoidance fine-tuning instructions or danger warnings to the quadrupedal firefighting robot dog through the communication link. IV. During firefighting operations, if the reconnaissance drone detects a sudden change in the fire situation or a new high-risk hotspot, it will update the three-dimensional environmental map of the fire scene and synchronize it to the land-air collaborative command and control platform; the land-air collaborative command and control platform will recalculate the safe area and send the updated operating position or evacuation command to the quadrupedal firefighting robot dog.

8. A fire extinguishing method for a land-air coordinated fire extinguishing system using drones and robot dogs according to claim 2, characterized in that, Step three also includes: using image recognition algorithms to compare real-time images collected by the reconnaissance drone with first-view images transmitted back by the quadrupedal firefighting robot dog, and using visual feature matching to assist in positioning, correcting the cumulative position error of the quadrupedal firefighting robot dog in environments with no or weak satellite signals.

9. A fire extinguishing method for a land-air coordinated fire extinguishing system using drones and robot dogs according to claim 2, characterized in that, Also includes: After the firefighting operation was completed, the reconnaissance drone guided the quadrupedal firefighting robot dog to withdraw to the designated equipment recovery point along a preset safe route. At the same time, the reconnaissance drone conducted a second thermal imaging scan of the fire site ruins. After confirming that there was no risk of reignition, it sent a mission termination command to the land-air coordinated command and control platform.

Images