Unmanned aerial vehicle drowning rescue intelligent triggering and executing system

Abstract

The invention relates to the technical field of unmanned aerial vehicle autonomous rescue, and discloses an unmanned aerial vehicle drowning rescue intelligent triggering and executing system, which integrates water surface ripples and underwater acoustic signals collected by a distributed sensor network, quantitatively calculates and generates drowning characteristic indexes, and constructs a dynamic real-time energy field covering a monitored water area according to the drowning characteristic indexes. The core of the system is that the system generates a navigation instruction by calculating the gradient of an energy field, and the rescue unmanned aerial vehicle with the self-floating capacity and a grabbing structure is guided in real time to accurately fly to a target point and land on the water surface. After landing, the unmanned aerial vehicle can detect specific low-frequency oscillation generated by grabbing of a person falling into water by analyzing data of an inertial measurement unit of the unmanned aerial vehicle, so that the effectiveness of rescue is intelligently verified, and a subsequent task is autonomously triggered. According to the invention, full-closed-loop autonomous rescue from intelligent sensing and dynamic decision making to accurate execution and validity verification is realized, and the rescue efficiency and success rate are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of autonomous drone rescue technology, specifically to an intelligent triggering and execution system for drone water rescue. Background Technology

[0002] Water safety has always been a key focus of public safety, and drowning accidents are extremely dangerous due to their suddenness and short golden rescue time. Traditional rescue methods mainly rely on human lookout and patrols. Once a danger is detected, lifeguards or rescue boats then go to the scene. While this model is effective to some extent, its detection range is limited, and its response speed is constrained by human reaction time and physical distance. Especially at night, in inclement weather, or in open water, its rescue efficiency and success rate are greatly reduced.

[0003] In recent years, with the development of drone technology, its application in water search and rescue has become a new technological direction. Existing drone rescue applications typically function as an aerial "eye" or delivery tool. For example, a pilot remotely controls the drone for aerial reconnaissance, and after detecting a suspected target, manually controls the drone to fly over the target and drop lifebuoys or other floating objects. However, such solutions still have significant limitations in practical applications. They heavily rely on the pilot's personal skills and on-site judgment, with an unavoidable delay between target detection and rescue action. More importantly, the entire rescue process is an open-loop system; after dropping rescue equipment, the drone cannot autonomously determine whether the rescue was truly effective—that is, whether the person in the water has successfully grabbed the rescue equipment. The system lacks an intelligent, closed-loop verification mechanism, preventing it from autonomously deciding on the next step based on the actual rescue status, such as whether a second drop is needed or whether the mission can be confirmed and the drone can return autonomously. This deep reliance on human intervention and the lack of autonomous perception and confirmation of the rescue mission status limit the automation level and overall rescue efficiency of existing drone rescue systems. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent triggering and execution system for drone water rescue, which solves the problems of response delays and low efficiency caused by the high dependence on remote manual control in existing drone water rescue solutions, as well as the fact that the rescue process is an open-loop system and cannot autonomously verify whether the rescue is truly effective.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] This invention provides an intelligent triggering and execution system for drone water rescue, comprising:

[0007] The intelligent decision triggering module is used to receive surface feature ripple signals and underwater acoustic signals from a distributed sensor network, fuse the received signals to generate a drowning feature signal, generate a rescue start command when the drowning feature signal exceeds a preset trigger threshold, and continuously parse the drowning feature signal into a dynamic target location.

[0008] The autonomous rescue execution module is used to respond to the rescue initiation command and control the flight system of the UAV according to the dynamic target position, navigate to the target water area and perform rescue deployment actions;

[0009] The rescue effectiveness verification module is activated after the autonomous rescue execution module completes the rescue action. It determines whether the person in the water has made effective contact with the drone by monitoring the changes in the physical state of the drone itself. When contact is determined to have occurred, it generates a rescue success confirmation signal and sends it to the autonomous rescue execution module.

[0010] Preferably, the drowning characteristic signal is a quantization index, which is generated by weighted fusion of the energy of the water surface characteristic ripple signal, the energy of the underwater acoustic signal, and correlation parameters that assess the spatiotemporal consistency of the two signals.

[0011] The quantitative index, namely the drowning characteristic index DCI(i,t), is generated by the following formula:

[0012] DCI(i,t)=w r ·E r (i,t)+w a ·E a (i,t)+w c ·C st (i,t);

[0013] in:

[0014] DCI(i,t) is the drowning characteristic index of the i-th sensing node at time t;

[0015] E r (i,t) represents the energy of the water surface characteristic ripple signal;

[0016] E a (i,t) represents the energy of the underwater acoustic signal;

[0017] C st (i,t) is a correlation parameter for evaluating the consistency of the spatiotemporal sources of the two signals;

[0018] w r w a w c These are the corresponding weighting coefficients.

[0019] Preferably, the dynamic target position is determined in the following way:

[0020] The intelligent decision triggering module uses the drowning characteristic signal from the distributed sensor network to construct a real-time energy field, calculates the gradient of the energy field, and defines the gradient direction as the dynamic target position to guide the flight of the drone.

[0021] Preferably, the method by which the intelligent decision-triggered module constructs the real-time energy field is as follows:

[0022] The known geographical locations of each sensor node in the distributed sensor network are set as spatial coordinates, the generated quantization index is set as the field strength value on the spatial coordinates, and then the real-time energy field is formed by spatial interpolation.

[0023] The real-time energy field Φ(x,y,t) is generated using inverse distance weighted interpolation.

[0024]

[0025] in:

[0026] Φ(x,y,t) is the energy field intensity at coordinate (x,y) at time t;

[0027] N is the total number of sensor nodes participating in the calculation;

[0028] DCI(i,t) is the drowning characteristic index of the i-th sensing node at time t;

[0029] d(x,y,pos i Let (x, y) be the position of point (x, y) and the position pos of the i-th sensor node. i The distance between them;

[0030] p is the power of the distance;

[0031] To sum the weighted values ​​of all N sensor nodes;

[0032] This is the weighted sum of the drowning characteristic indices of all sensing nodes;

[0033] This is the sum of the weights of all sensor nodes.

[0034] The flight direction of the drone is determined by the gradient of the energy field. The target speed command of the UAV is determined, and the target speed command V of the UAV is determined. cmd (t) is set to a vector with the same direction as the gradient:

[0035]

[0036] in:

[0037] V max The preset maximum cruising speed;

[0038] (x u ,y u Let t be the coordinates of the UAV itself at time t;

[0039] It is the norm (or modulus) of the gradient.

[0040] Preferably, the rescue effectiveness verification module monitors the changes in the physical state of the UAV by analyzing the readings of the inertial measurement unit on the UAV itself. The effective contact is defined as the low-frequency oscillation signal generated after the person who fell into the water and the UAV itself, which serves as the rescue platform, form a physical coupling.

[0041] Preferably, the drone is also equipped with a gimbal camera and a real-time image transmission device mounted on the head of the drone, which are used to provide real-time video images of the target water area to the remote monitoring terminal, and the rescue success confirmation signal generated by the rescue effectiveness verification module is used to send a confirmation notification to the remote monitoring terminal that the target status has become stable.

[0042] Preferably, the value of the drowning characteristic signal generated by the intelligent decision triggering module and the rescue success confirmation signal generated by the rescue effectiveness verification module are used together as core status data and superimposed on the real-time video screen provided to the remote monitoring terminal through the data channel of the real-time image transmission device.

[0043] Preferably, the drone is configured in a smart standby dock and interacts with the distributed sensor network through a decentralized peer-to-peer communication network.

[0044] Preferably, the autonomous rescue execution module is further configured to generate a return command after the rescue effectiveness verification module generates the rescue success confirmation signal, and control the flight system of the UAV according to the return command to perform autonomous navigation back to the smart standby dock.

[0045] Preferably, the drone has the ability to float on water, and the drone also includes a protective grille disposed around the drone propeller, and a quick-locking mechanism disposed on the body for a person falling into the water to grab.

[0046] Furthermore, the autonomous rescue execution module is also used to control the drone to take off again after landing on the water.

[0047] This invention provides an intelligent triggering and execution system for drone water rescue. It has the following beneficial effects:

[0048] 1. This invention generates a quantization index by fusing surface ripple signals and underwater acoustic signals, and triggers drone rescue based on the quantization index, thereby achieving rapid and automated identification and response to drowning incidents and significantly reducing rescue delays caused by misjudgment from a single source.

[0049] 2. This invention constructs a real-time energy field using drowning characteristic signals from a distributed sensor network, and continuously analyzes the dynamic target position based on the gradient of the energy field to guide the drone's flight. This solves the problem of positioning failure caused by target drift in traditional rescue operations, and ensures accurate and dynamic tracking of people who have fallen into the water.

[0050] 3. This invention determines the success of the rescue by analyzing the low-frequency oscillation signal generated after the drone and the person in the water form a physical coupling, and triggers the drone to return to its home autonomously. It establishes an objective verification mechanism based on physical state confirmation and realizes an autonomous closed loop from hazard identification to mission completion. Attached Figure Description

[0051] Figure 1 A schematic diagram of the overall architecture of a water surface autonomous rescue system provided in an embodiment of the present invention;

[0052] Figure 2 This is a schematic diagram of the intelligent decision-making and navigation instruction generation process according to an embodiment of the present invention;

[0053] Figure 3 This is a flowchart illustrating the full lifecycle execution of an autonomous rescue mission according to an embodiment of the present invention. Detailed Implementation

[0054] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] See attached document Figure 1 , Figure 1 This is a schematic diagram of the overall architecture of an autonomous rescue system according to an embodiment of the present invention. The present invention provides a water surface autonomous rescue system, which includes: an intelligent decision triggering module 10, an autonomous rescue execution module 20, a rescue effectiveness verification module 30, and a remote monitoring terminal 40. Through the coordinated operation of the above modules, the system achieves autonomous rescue of people who have fallen into the water.

[0056] The intelligent decision-making triggering module 10 is responsible for receiving data from an external distributed sensor network and generating navigation instructions for the rescue mission. This module acquires signals from surface feature sensors and underwater acoustic sensors deployed in the water area. After preprocessing and spatiotemporal consistency assessment, it fuses and quantizes the multi-source signals. This module utilizes the drowning characteristic index DCI(i,t) of all effective sensor nodes to construct a real-time energy field Φ(x,y,t) covering the monitored water area. The energy field is calculated using an inverse distance-weighted interpolation method, expressed as:

[0057]

[0058] in:

[0059] Φ(x,y,t) is the energy field intensity at coordinate (x,y) at time t;

[0060] N is the total number of sensor nodes participating in the calculation;

[0061] DCI(i,t) is the drowning characteristic index of the i-th sensing node at time t;

[0062] pos i The known geographical coordinates of the i-th sensor node;

[0063] d(x,y,pos i Let (x, y) be the position of point (x, y) and the position pos of the i-th sensor node. i The Euclidean distance between them;

[0064] p is the power of the distance;

[0065] To sum the weighted values ​​of all N sensor nodes;

[0066] This is the weighted sum of the drowning characteristic indices of all sensing nodes;

[0067] This is the sum of the weights of all sensor nodes.

[0068] After obtaining the real-time energy field Φ(x,y,t), the intelligent decision triggering module 10 calculates the energy field at the current position (x,y,t) of the UAV. u ,y u gradient of ) This gradient vector points in the direction of the fastest increase in the energy field value, guiding the drone's flight. The intelligent decision triggering module 10 generates the target velocity command V based on this gradient. cmd (t), this instruction is a vector, and its expression is:

[0069]

[0070] in:

[0071] V cmd (t) is the target velocity command vector of the UAV at time t;

[0072] V max This is the maximum cruising speed of the drone;

[0073] (x u ,y u Let t be the geographical coordinates of the UAV itself at time t;

[0074] The energy field gradient calculated at the current location of the drone;

[0075] ||·|| is the Euclidean norm of the vector.

[0076] The instruction was sent to the autonomous rescue execution module 20 in real time.

[0077] The autonomous rescue execution module 20 is the physical execution unit for the rescue mission, integrated within the UAV platform. This module receives the target speed command V sent by the intelligent decision triggering module 10. cmd (t), and accordingly control the drone to complete flight maneuvers such as autonomous takeoff, dynamic route planning, and precise water landing. This module also receives manual intervention commands from the remote monitoring terminal 40 and prioritizes executing these commands, such as forced return or mission termination. This module is also responsible for adjusting the drone's position on the water to ensure that it is within the preset grasping range of the person who falls into the water.

[0078] The rescue effectiveness verification module 30 is responsible for determining the effectiveness of the rescue after the drone lands on the water. This module is integrated inside the drone platform and analyzes the data collected by the drone's own inertial measurement unit (IMU) to detect the presence of specific low-frequency oscillation signals caused by the person grabbing the water.

[0079] The rescue effectiveness verification module 30 performs frequency domain analysis on the IMU data to extract frequency components. When the rescue effectiveness verification module 30 detects a frequency component whose duration exceeds a preset threshold Δt... confirm At that time, the rescue effectiveness verification module 30 generates a rescue success confirmation signal and sends it to the autonomous rescue execution module 20 to trigger the subsequent autonomous return process.

[0080] The remote monitoring terminal 40 provides a human-machine interface for the operator. This terminal receives real-time video streams, location information, and system status data transmitted from the drone. The operator can monitor the drone through this terminal and, when necessary, send commands to the autonomous rescue execution module 20 to intervene in the drone's mission execution.

[0081] The physical components of this invention include an autonomous execution platform, a distributed sensor network, an intelligent standby dock, and a remote monitoring terminal 40.

[0082] The autonomous execution platform is the physical carrier of the rescue mission; in this embodiment, it is a multi-rotor unmanned aerial vehicle (UAV). The main body of the UAV platform adopts a sealed and waterproof structural design and is made of high-strength composite materials. The UAV's body has the ability to float on water and provides 190 Newtons of buoyancy.

[0083] The drone platform features grip-friendly structures on its sides or bottom, such as rigid handles, flexible ropes, or mesh structures. Protective grilles are installed beneath each rotor of its power system to isolate the high-speed rotating blades and prevent them from catching the hair or clothing of those falling into the water, thus avoiding secondary injuries. The drone platform is equipped with the inertial measurement unit (IMU) and GPS module necessary for flight missions, as well as optical cameras, thermal imaging cameras, or LiDAR for close-range target identification.

[0084] The distributed sensor network consists of multiple sensor nodes deployed in the monitored water area. This network includes surface feature sensors and underwater acoustic sensors. Surface feature sensors monitor abnormal changes in water surface ripples, while underwater acoustic sensors, such as hydrophones, collect underwater sound signals. These sensor nodes interact via a decentralized peer-to-peer communication network to send the collected data to the intelligent decision triggering module 10. This network constitutes a self-organizing wireless mesh network, where each sensor node acts as both a data sender and a routing relay for other nodes. When a sensor node collects a signal, its data packet is transmitted in a multi-hop manner through the network using a dynamic routing protocol until it is received by the communication unit located in the intelligent standby dock or directly by the UAV platform. In this network structure, when some nodes go offline due to failure, data packets can still be routed and forwarded through other normally functioning nodes in the network to reach the target node.

[0085] A smart standby dock is the ground infrastructure for a drone platform. Its structure includes a platform for parking the drones, a charging interface for autonomous charging of the drone batteries, and a communication unit that acts as a data relay. The drones are parked here while in standby mode and after completing their missions.

[0086] The remote monitoring terminal 40 is a human-machine interface device equipped with a display screen, a data processing unit, and input devices. This terminal receives and displays video and status data transmitted from the drone, and allows operators to send manual intervention commands via the input devices.

[0087] See attached document Figure 2 , Figure 2 This is a schematic diagram of the intelligent decision-making and navigation command generation process according to an embodiment of the present invention. The intelligent decision-making triggering module 10 collects raw environmental signals through a distributed sensor network. Each sensor node is equipped with a millimeter-wave Doppler radar for collecting water surface characteristic ripple signals, a vector hydrophone for collecting underwater acoustic signals, and a signal processing unit for preprocessing the collected raw signals.

[0088] Millimeter-wave Doppler radar is used to continuously transmit electromagnetic waves of a preset frequency towards the water surface and receive the echo signals. When the water surface generates ripples within a specific frequency range due to the struggle of a target, the echo signal will exhibit a Doppler frequency shift. The signal processing unit receives the echo signal and processes it through a short-time Fourier transform, converting the time-domain echo signal into a time-spectrum diagram to obtain the preprocessed water surface characteristic ripple signal R. p (i,t,f). Where i is the index of the sensor node, t is the time, and f is the frequency.

[0089] Vector hydrophones are used to synchronously acquire underwater acoustic signals and determine the location of sound sources. The signal processing unit receives the underwater acoustic signals, filters out noise signals outside a preset frequency range using a bandpass filter, and then performs a short-time Fourier transform to obtain the pre-processed underwater acoustic signal A. p (i,t,f).

[0090] Through the above preprocessing steps, the intelligent decision triggering module 10 converts the raw sensor data from different physical modes into time-frequency domain signals that can be used for energy analysis and data fusion in the same dimension, providing a data foundation for the subsequent generation of drowning characteristic index.

[0091] The intelligent decision triggering module 10 fuses the received preprocessed water surface characteristic ripple signal with the underwater acoustic signal to generate a quantization index as a drowning characteristic signal.

[0092] The intelligent decision-making triggering module 10 first extracts energy features from the preprocessed signal. This is specifically for the water surface characteristic ripple signal R. p (i,t,f), the module calculates the first frequency range corresponding to the preset frequency of water ripples generated by a human struggling. By integrating the signal energy within the range, the ripple energy spectrum E is obtained. r (i,t). Its calculation formula is:

[0093]

[0094] in:

[0095] i is the index of the sensor node, t is the time, and f is the frequency;

[0096] R p (i,t,f) is the frequency domain representation of the preprocessed water surface characteristic ripple signal at time t, obtained by Fourier transform;

[0097] |R p (i,t,f)| 2 The energy intensity or power of the water surface characteristic ripple signal at a single frequency point f;

[0098] This is the lower limit frequency of the preset first frequency range;

[0099] The upper limit frequency of the preset first frequency range;

[0100] For the function values ​​within the parentheses, starting from the lower frequency limit Up to frequency limit Summation is performed continuously over the entire interval.

[0101] For underwater acoustic signal A p (i,t,f), the module calculates the second frequency range corresponding to the preset frequency of abnormal drowning sounds. By integrating the signal energy within the range, the acoustic energy spectrum E is obtained. a (i,t). Its calculation formula is:

[0102]

[0103] in:

[0104] i is the index of the sensor node, t is the time, and f is the frequency;

[0105] A p (i,t,f) represents the frequency domain representation of the preprocessed underwater acoustic signal at time t;

[0106] |A p (i,t,f)| 2 The energy intensity of the acoustic signal at a single frequency point f;

[0107] The lower limit frequency of the preset second frequency range;

[0108] The upper limit frequency of the preset second frequency range;

[0109] For the function values ​​within the parentheses, starting from the lower frequency limit Up to frequency limit Summation is performed continuously over the entire interval.

[0110] The intelligent decision triggering module 10 also evaluates the spatiotemporal consistency of the two signals detected at the same sensing node, generating a correlation parameter C that takes a value between 0 and 1. st (i,t). When two signals occur synchronously in time and have the same spatial origin, the correlation parameter C... st The value of (i,t) approaches 1.

[0111] Subsequently, the intelligent decision triggering module 10 will extract the ripple energy spectrum E r (i,t), acoustic energy spectrum E a (i,t), and the correlation parameter C st The values ​​(i,t) are weighted and fused to construct a quantitative index, namely the Drowning Characteristic Index (DCI(i,t)). Its calculation formula is as follows:

[0112] DCI(i,t)=w r ·E r (i,t)+w a ·E a (i,t)+w c ·C st (i,t);

[0113] Among them, w r w a w c These are preset weighting coefficients used to adjust the contribution of ripple energy, acoustic energy, and spatiotemporal correlation to the final index.

[0114] When the drowning characteristic index DCI(i,t) of any sensing node remains above a preset trigger threshold τ for a preset duration Δt. dci At that time, the intelligent decision triggering module 10 generates a rescue initiation command.

[0115] See attached document Figure 2 , Figure 2 This is a schematic diagram of the intelligent decision-making and navigation command generation process according to an embodiment of the present invention. After generating the rescue initiation command, the intelligent decision triggering module 10 continuously parses the drowning characteristic signals into dynamic target locations, which are used to provide navigation basis for the autonomous rescue execution module 20.

[0116] The intelligent decision triggering module 10 utilizes the drowning characteristic index DCI(i,t) reported by all sensor nodes in the distributed sensor network to construct a two-dimensional real-time energy field Φ(x,y,t) covering the target water area. The construction method is as follows: the known geographical locations of each sensor node in the distributed sensor network are set as spatial coordinates, the generated quantization index is set as the field strength value on the spatial coordinates, and then a real-time energy field is formed through spatial interpolation.

[0117] In one specific embodiment, the spatial interpolation operation employs inverse distance weighted interpolation. In this case, the energy field intensity Φ(x,y,t) at any coordinate point (x,y) in the energy field at time t is calculated using the following formula:

[0118]

[0119] in:

[0120] Φ(x,y,t) is the energy field intensity at coordinate (x,y) at time t;

[0121] N is the total number of sensor nodes participating in the calculation;

[0122] DCI(i,t) is the drowning characteristic index of the i-th sensing node at time t;

[0123] pos i The known geographical coordinates of the i-th sensor node;

[0124] d(x,y,pos i Let (x, y) be the position of point (x, y) and the position pos of the i-th sensor node. i The Euclidean distance between them;

[0125] p is the power of distance, a positive real number whose value determines the rate at which the weight of neighboring nodes decays with increasing distance.

[0126] To sum the weighted values ​​of all N sensor nodes;

[0127] This is the weighted sum of the drowning characteristic indices of all sensing nodes;

[0128] This is the sum of the weights of all sensor nodes.

[0129] The dynamic target position is determined by the gradient direction of the real-time energy field. The intelligent decision triggering module 10 calculates the energy field at the UAV's current position (x... u ,y u gradient of ) The gradient vector points in the direction of the fastest increase in energy field intensity.

[0130] The gradient was used to generate the target velocity command V for the UAV. cmd (t), this instruction is a speed V with the same direction as the gradient and a speed magnitude equal to the preset maximum cruising speed. max The vector. Its calculation formula is:

[0131]

[0132] in:

[0133] V cmd (t) represents the target velocity command vector generated and sent to the UAV at time t;

[0134] V max The preset maximum cruising speed for the drone;

[0135] (x u ,y u Let t be the geographical coordinates of the UAV itself at time t;

[0136] This indicates that at time t, the energy field Φ is at the current position (x) of the drone. u ,y u The gradient vector of ).

[0137] ||·|| is the Euclidean norm of the vector.

[0138] The intelligent decision triggering module 10 will continuously generate target speed commands V cmd (t) is sent to the autonomous rescue execution module 20 to guide the UAV to correct its course in real time and achieve dynamic tracking of the peak area of ​​the energy field.

[0139] The autonomous rescue execution module 20 uses a drone as the core entity of its autonomous execution platform. The drone's airframe is constructed from a single piece of high-strength composite material, measuring 935×935×121mm (+ / -10.5mm), and its structural design enables it to float on water. The airframe can provide 190 Newtons of buoyancy on the water surface to support the weight of one or more people in the water. The drone also includes a quick-lock mechanism on its airframe for people in the water to grip, and a protective grille surrounding the propulsion system's propeller to prevent the propeller from catching the hair or clothing of people in the water during surface operations.

[0140] The drone has an IP68 protection rating and can be submerged in 1.5 meters of water for 1 hour. The drone system can withstand an external ambient temperature range of -10°C to 55°C, and its design allows it to take off and land normally in environments with winds up to force 6 and sea state 2.

[0141] This drone platform integrates a gripping structure specifically designed for people falling into the water. Ring-shaped or rod-shaped rigid handles, injection-molded from high-strength engineering plastic, are bolted or clipped to the sides or perimeter of the fuselage. Alternatively, a network of ropes woven from high-strength nylon or polypropylene fibers is tensioned and connected between multiple anchor points on the fuselage frame. The design dimensions and surface texture of these structures ensure a stable grip even with low grip strength or while wearing gloves.

[0142] The drone is also equipped with a single-axis gimbal camera with a range of +90° to -90° and a resolution of 720P. It also features a real-time image transmission device to transmit live video footage captured by the camera to a remote monitoring terminal. Four maintenance windows are located at the rear of the drone for battery replacement and system maintenance.

[0143] See attached document Figure 3 , Figure 3 This is a flowchart illustrating the full lifecycle execution of an autonomous rescue mission according to an embodiment of the present invention. The autonomous rescue execution module 20 remains in a standby state while the UAV is in the intelligent standby dock. When the autonomous rescue execution module 20 receives a rescue initiation command generated by the intelligent decision triggering module 10, the rescue mission execution process is initiated.

[0144] After the process is initiated, the autonomous rescue execution module 20 first sends an autonomous takeoff command to the drone's flight system. The flight system responds to the command, controlling the drone to take off from the intelligent standby dock and climb to the preset cruising altitude.

[0145] After the drone completes takeoff, the autonomous rescue execution module 20 enters the dynamic navigation phase. During this phase, the autonomous rescue execution module 20 continuously receives the target speed command V, which is analyzed and sent in real time by the intelligent decision triggering module 10. cmd (t). The autonomous rescue execution module 20 converts the speed command into a low-level control signal for the UAV flight system to adjust the UAV's flight heading and speed in real time, so that it flies towards the dynamic target position along the energy field gradient direction.

[0146] When the drone flies within a preset range of the dynamic target location, the autonomous rescue execution module 20 controls the drone's flight system to execute the landing procedure. Under the control of the autonomous rescue execution module 20, the drone lands on the surface of the target water area. The drone's landing on the water and providing buoyancy to the person in the water constitutes the rescue deployment action of this invention.

[0147] The autonomous rescue execution module 20 is also responsible for controlling the UAV's water surface take-off and landing operations and autonomous return process. The UAV's flight system has the ability to take off again after landing on the water. After a rescue deployment, if the distance between the UAV and the person in the water exceeds the preset range, the autonomous rescue execution module 20 can send a water surface take-off command to the flight system and navigate and land again based on the dynamic target position until the UAV accurately lands next to the person in the water.

[0148] See attached document Figure 3 , Figure 3 This is a schematic diagram of the autonomous rescue mission execution process according to an embodiment of the present invention. The autonomous rescue execution module 20 is activated after receiving a rescue success confirmation signal generated by the rescue validity verification module 30, or a forced return command from the remote monitoring terminal 40. The return command is sent to the UAV's flight system.

[0149] Upon receiving the return-to-home command, the flight system executes a pre-set autonomous return-to-home procedure. If the drone is currently on the water, the flight system first controls the drone to take off from the water and ascend to a pre-set safe return-to-home altitude. Subsequently, based on the geographical coordinates of the intelligent standby dock stored internally, the flight system plans and executes the return-to-home route, ultimately controlling the drone to autonomously land in the intelligent standby dock.

[0150] In addition, the autonomous rescue execution module 20 also monitors the communication link status with the remote monitoring terminal 40. When the communication link is interrupted and the duration exceeds a preset threshold, the autonomous rescue execution module 20 also generates a return command and triggers the aforementioned autonomous return procedure.

[0151] The rescue effectiveness verification module 30 is integrated inside the drone. The drone is equipped with an inertial measurement unit (IMU), which is used to continuously collect motion data such as acceleration and angular velocity of the drone in three-dimensional space.

[0152] The rescue effectiveness verification module 30 continuously receives and analyzes the motion data collected by the IMU. After the drone lands on the water, the rescue effectiveness verification module 30 continuously monitors the IMU data for safety status characteristics resulting from the physical coupling between the drone and the person in the water. These safety status characteristics are defined as a specific low-frequency oscillation signal.

[0153] To identify characteristic signals, the rescue effectiveness verification module 30 performs frequency domain analysis on the IMU data to extract frequency components. The rescue effectiveness verification module 30 detects whether any frequency component contains a duration exceeding a preset threshold Δt. confirm Furthermore, its energy is mainly concentrated in a preset low-frequency oscillation range [f] min ,f max Signals within ]

[0154] Low-frequency oscillation frequency range [f min ,f max This is used to characterize the unique vibration pattern produced when a human grabs a floating object on the water surface, a vibration pattern distinct from natural oscillations caused by water flow or wind. Once the characteristic signal is identified, the rescue effectiveness verification module 30 generates a rescue success confirmation signal. This signal is sent to the autonomous rescue execution module 20 to trigger the drone's autonomous return procedure.

[0155] See attached document Figure 1 , Figure 1 This is a schematic diagram of the overall architecture of an autonomous rescue system according to an embodiment of the present invention. The remote monitoring terminal 40 provides a human-machine interface for the operator at the rear to supervise the entire autonomous rescue mission and intervene when necessary.

[0156] The remote monitoring terminal 40 is a device that integrates a display unit, a data processing unit, a wireless communication module, and physical input devices (e.g., a joystick or button). It interacts bidirectionally with the drone platform via one or more wireless data links, such as 4G / 5G cellular networks or dedicated point-to-point radio frequency links.

[0157] The data stream transmitted from the UAV platform to the remote monitoring terminal 40 consists of two concurrent parts: a real-time video stream and a telemetry data stream. The real-time video stream comes from an optical or thermal imaging camera mounted on the UAV's single-axis stabilized gimbal, allowing the operator to visually observe the on-site environment. The telemetry data stream contains the UAV platform's precise geographic coordinates, altitude, flight speed, remaining battery percentage, and current operational status reported by various system modules, such as en route to the target, having landed on water, or verifying the effectiveness of the rescue operation.

[0158] The user interface of the remote monitoring terminal 40 displays the received video stream as the main background and overlays the parsed telemetry data in the form of text or graphic symbols onto a designated area of ​​the screen in real time. For example, the drone's location is marked on an electronic map, while battery level and status information are displayed in the status bar.

[0159] The data stream transmitted from the remote monitoring terminal 40 to the autonomous rescue execution module 20 of the UAV platform mainly consists of manual intervention commands. Operators can send high-level task commands with the highest execution priority. These commands include: a forced return command to interrupt the current task at any stage and immediately initiate the return-to-the-smart dock procedure; a task termination command to immediately stop all UAV actions; and a manual confirmation command, allowing the operator to manually send a signal indicating rescue success or failure after visual confirmation via video. This signal overrides the automatic judgment result of the rescue effectiveness verification module 30.

[0160] In addition, operators can send control commands to the camera gimbal via the terminal to adjust the camera's pitch and yaw angles independently of the drone's flight attitude, thereby enabling detailed observation of specific areas. All these commands are encoded into data packets by the terminal and transmitted wirelessly to the autonomous rescue execution module 20 for parsing and execution.

Claims

1. An unmanned aerial vehicle water landing rescue intelligent triggering and executing system, characterized in that, The application relates to a rescue system for drowning people, which comprises the following modules: An intelligent decision trigger module is used for receiving water surface characteristic ripple signals and underwater acoustic signals from a distributed sensing network, fusing the received signals, generating a drowning characteristic signal, generating a rescue start instruction when the drowning characteristic signal exceeds a preset trigger threshold, and continuously analyzing the dynamic target position of the drowning characteristic signal; An autonomous rescue execution module is used for responding to the rescue start instruction, controlling the flight system of the unmanned aerial vehicle according to the dynamic target position, navigating to the target water area and executing a rescue action; A rescue effectiveness verification module is used for being activated after the autonomous rescue execution module completes the rescue action, judging whether the fallen person has effectively contacted the unmanned aerial vehicle by monitoring the physical state change of the unmanned aerial vehicle, and generating a rescue success confirmation signal and sending the signal to the autonomous rescue execution module when the contact is determined to occur. 2.The unmanned aerial vehicle water landing rescue intelligent triggering and executing system according to claim 1, wherein The drowning characteristic signal is a quantitative index, which is generated by weighted fusion of the energy of the water surface characteristic ripple signal, the energy of the underwater acoustic signal and a correlation parameter for evaluating the consistency of the space-time sources of the two signals. 3.The unmanned aerial vehicle water landing rescue intelligent triggering and executing system of claim 2, wherein, The dynamic target position is determined by the following method: The intelligent decision trigger module uses the drowning characteristic signal of the distributed sensing network to construct a real-time energy field, calculates the gradient of the energy field, and defines the gradient direction as the dynamic target position for guiding the flight of the unmanned aerial vehicle. 4.The unmanned aerial vehicle water landing rescue intelligent triggering and executing system of claim 1, wherein, The rescue effectiveness verification module realizes the monitoring of the physical state change of the unmanned aerial vehicle by analyzing the readings of an inertial measurement unit carried by the unmanned aerial vehicle, wherein the effective contact is defined as a low-frequency oscillation signal generated after the fallen person forms a physical coupling body with the unmanned aerial vehicle as a rescue platform. 5.The unmanned aerial vehicle water landing rescue intelligent triggering and executing system of claim 1, wherein, The unmanned aerial vehicle is also equipped with a gimbal camera and a real-time image transmission device installed on the head of the machine body, which are used for providing real-time video pictures of the target water area for a remote monitoring terminal, and the rescue success confirmation signal generated by the rescue effectiveness verification module is used for sending a confirmation notification that the target state has been stabilized to the remote monitoring terminal. 6.The unmanned aerial vehicle water landing rescue intelligent triggering and executing system of claim 5, wherein, The numerical value of the drowning characteristic signal generated by the intelligent decision trigger module and the rescue success confirmation signal generated by the rescue effectiveness verification module are used as core state data, which are superimposed on the real-time video pictures provided for the remote monitoring terminal through the data channel of the real-time image transmission device. 7.The unmanned aerial vehicle water landing rescue intelligent triggering and executing system of claim 3, wherein, The method for constructing the real-time energy field by the intelligent decision trigger module is as follows: The known geographical positions of the sensing nodes in the distributed sensing network are set as spatial coordinates, the generated quantitative index is set as the field strength value on the spatial coordinates, and the real-time energy field is formed through a spatial interpolation operation. 8.The unmanned aerial vehicle water landing rescue intelligent triggering and executing system of claim 1, wherein, The unmanned aerial vehicle is arranged in an intelligent standby dock and interacts with the distributed sensing network through a decentralized point-to-point communication network. 9.The unmanned aerial vehicle water landing rescue intelligent triggering and executing system of claim 1, wherein, The drone has the ability to float on water, and the drone also includes a protective grille around the drone propeller and a quick-locking mechanism on the body for people to grab when they fall into the water. Furthermore, the autonomous rescue execution module is also used to control the drone to take off again after landing on the water. 10.The unmanned aerial vehicle water landing rescue intelligent triggering and executing system of claim 8, wherein, The autonomous rescue execution module is also used to generate a return command after the rescue effectiveness verification module generates the rescue success confirmation signal, and control the flight system of the UAV according to the return command to perform autonomous navigation back to the smart standby dock.

Images